When people chat in Microsoft Teams (MS Teams), a ‘compliance’ copy of the chat is saved to either personal or (Microsoft 365) Group mailboxes. This copy is subject to retention policies, and can be found and exported via Content Search.
But what happens if there is no Exchange Online mailbox? It seems the chats become inaccessible which could be an issue from a recordkeeping and compliance point of view.
This post explains what happens, and why it may not be a good idea (from a compliance and recordkeeping point of view) not to disable the Exchange Online mailbox option as part of licence provisioning.
Licences and Exchange Online mailboxes
When an end-user is allocated a licence for Microsoft 365, a decision (sometimes incorporated into a script) is made about which of the purchased licences – and apps in those licences – will be assigned to that person.
E1, E3 and E5 licences include ‘Exchange Online’ as an option under ‘Apps’. This option is checked by default (along with many of the other options), but it can be disabled (as shown below).
If the checkbox option is disabled as part of the licence assigning process (not after), the end-user won’t have an Exchange mailbox and so won’t see the Outlook option when they log on to office.com portal. (Note – If they have an on-premise mailbox, that will continue to exist, nothing changes).
Having an Exchange Online mailbox is important if end-users are using MS Teams, because the ‘compliance’ copy of 1:1 chat messages in MS Teams are stored in a hidden folder (/Conversation History/Team Chat) in the Exchange Online mailbox of every participant in the chat. If the mailbox doesn’t exist, those copies aren’t made and so aren’t accessible and may be deleted.
If end-users chat with other end-users who don’t have an Exchange mailbox as shown in the example below, the same thing happen – no compliance copy is kept. The chat remains inaccessible (unless the Global Admins take over the account).
The exchange above, between Roger Bond and Charles, includes some specific key words. As we will see below, these chats cannot be found via a Content Search.
(On a related note, if the ability to create private channels is enabled and they create a private channel and chat there, the chats are also not saved because a compliance copy of private channel chats are stored in the mailboxes of the individual participants.)
Searching for chats when no mailbox exists
As we can see above, the word ‘mosquito’ was contained in the chat messages between Roger and Charles.
Content Searches are carried out via the Compliance portal and are more or less the same as eDiscovery searches in that they are created as cases.
From the Content Search option, a new search is created by clicking on ‘+New Search’, as shown below. The word ‘mosquito’ has been added as a keyword.
We then need to determine where the search will look. In this case the search will look through all the options shown below, including all mailboxes and Teams messages.
When the search was run, the results area shows the words ‘No results found’.
Clicking on ‘Status details’ in the search results, the following information is displayed – ‘0 items’ found. The ‘5 unindexed items’ is unrelated to this search and simply indicates that there are 5 unindexed items.
Double-checking the results
To confirm the results were accurate, another search was conducted where the end-user originally did not have a mailbox, and then was assigned one.
If the end-user didn’t have a mailbox but the other recipient/s of the message did, the Content Search found one copy of the chat message in the mailbox of the other participants. Only one item was found.
When the Exchange Online option was enabled for the end-user who previously did not have a mailbox (so they were now assigned a mailbox), a copy of the chat was found in the mailbox of both participants, as shown in the details below (‘2 items’).
Summary and implications
If end users chat in the 1:1 area of MS Teams and don’t have an Exchange Online mailbox, no compliance copy of the chat will be saved, and so it will not be found via Content Search.
If any of the participants in the 1:1 chat have an Exchange Online mailbox, the chat will appear in the mailboxes of those participants.
If all participants in the 1:1 chat have an Exchange Online mailbox, the chat will be found in the mailbox of all participants.
Further to the above:
If end users can delete chats (via Teams policies) and don’t have a mailbox, no copy of the chat will exist.
If end-users with a mailbox can delete Teams chats, but a retention policy has been applied to the chats, the chats will be retained as per the retention policy (in a hidden folder).
And finally, if you allow private channels, end-users can create private channels in the Organisation Team. The chats in these private channels are usually stored in the personal mailboxes of participants (not the Group mailbox) – so these chats will also be inaccessible and cannot be found via Content Search.
The implications for the above are that, if you need to ensure that personal chat messages can be accessed (from Content Search), then the participants in the chat must have an Exchange Online mailbox.
Further, if you allow deletion of chats but need to be able to recover them for compliance purposes, a retention policy should be applied to Teams 1:1 chat.
In his April 2007 article titled ‘Useful Void: The Art of Forgetting in the Age of Ubiquitous Computing’ (Harvard University RWP07-022), Viktor Mayer-Schönberger noted that the default human behaviour for millenia was to forget. Only information that needed to be kept would be retained. He noted that the digital world had changed the default to remembering, and that the concept of forgetting needed to be re-introduced through the active deletion of digital content that does not need to be retained.
The harsh reality is that there is now so much digital information in the world, including digital content created and captured by individual organisations, that active deletion of content that does not need to be retained, seems an almost impossible task.
This post explores issues with the traditional model of records retention in the digital world, and why newer options such as the records retention capability of Microsoft 365 is a more effective way to manage the retention and disposal of records, and all other digital content.
The traditional retention model
The traditional model of managing the retention and disposal/disposition of records was based on the ability to apply a retention policy to a group or aggregation of information identified as records. For the most part, those paper records were the only copy that existed (with some allowance for working and carbon copies).
The model worked reasonably well for paper records, but started to falter when paper records became the printed versions of born-digital records, and where the original digital versions remained where they were created or captured – on network files shares, in email systems, and on backups. Although, technically, the official record was on a file, a digital version was likely to remain on network file shares or in an email mailbox after the paper version was destroyed at the end of the retention period, and remain overlooked.
How many of us have had to wade through the content of old network file shares to examine the content, determine its value, and perhaps see if it can even still be accessed? Or do the same with old backup tapes?
The volume of unmanaged digital content, not subject to any retention policy, only continued to increase. This situation continued to worsen when electronic document and records management (EDRM) systems were introduced from the late 1990s. End-users had to copy records to the EDRMS, thereby creating yet another digital copy, in addition to the born-digital originals stored in mailboxes or file shares.
Even if the record in the EDRMS were destroyed, there was a good chance the original ‘uncontrolled’ version of the digital record – along with an unknown volume of digital records that probably should have been consigned to the EDRMS but weren’t – remained in email mailboxs, on file shares, or on a backup tape somewhere.
eDiscovery was born.
The emergence of new forms of digital records, including instant messages, social media, and smart-phone based chat and other apps from the early 2000s only added to the volume of digital content, much of which was stored in third-party cloud-based and mobile-device accessible applications completely out of the reach and ability of the organisation trying to manage records.
Modern retention management
A modern approach to retention management should be based on the following principles:
Information, not just records, should only be kept for as long as it is required.
It is no longer possible to accurately and/or consistently identify and capture all records in a single recordkeeping system.
Duplication of digital content can be reduced by creating and capturing records in place, promoting ‘working out loud’, co-authoring and sharing (no more attachments and private copies).
None of the above points excludes the ability to manage certain types of records at a more granular level where this is required. But these records, or the location in which they are created or captured, should not be regarded as the only form of record.
Ideally, these records should be created (or captured) directly in the system where they are to be managed – not copied to it.
Change management is necessary
Some of these new ways of working are likely to come up against deeply ingrained behaviours, many of which go back several decades and have contributed to a reluctance to ‘forget’ and destroy old digital content, including:
hiding/hoarding content in personal drives (and personal cloud-based systems or on USB drives);
communicating by email, the content is which is inaccessible to anyone else;
attaching documents to emails;
printing and filing born-digital content; and
sometimes, scanning/digitising the printed copies of born-digital records and saving them back to a digital system.
What about destruction?
Records managers in organisations moving away from the authorised destruction of digital content identified as records, to the destruction of all digital content (including identified records) need to consider what is required to achieve this outcome, and the implications for existing process and practices (including those described above).
Some activities will remain unchanged. For example, the need to review certain types of records before they are destroyed (aka ‘disposition review’), to seek approval for that destruction, and to keep a record of what was destroyed.
Some activities are new and can replace other existing actions and activities. For example, the application of retention policies to mailboxes can remove the requirement to backup those mailboxes.
Some of activities or outcomes may be challenging. For example, the automatic destruction without review of digital content that is not the subject of more granular retention requirements, such as emails out of mailboxes, documents in personal working drives. This content will simply disappear after the retention period expires.
How Microsoft 365 can support modern retention management
Microsoft recognised some time ago that it was becoming increasingly difficult to manage the volumes and types of digital content that was being created every day by organisations.
Exisiting and newly released functionality in the Compliance portal of Microsoft 365 includes the ability to create and apply both label-based retention policies to specific types of records, including automatically based on machine learning capabilities, and broader ‘workload’ specific (e.g., mailboxes, SharePoint sites, OneDrive accounts, MS Teams chats) retention policies. This capability helps organisations to focus retention requirements on the records that need to be retained, while destroying digital content that is no longer relevant and can be forgotten.
Instead of directing end-users to identify records and copy them from one system to another (thereby creating two versions), Microsoft 365 allows end-users to create and capture records in place, providing a single source of truth that can be shared (rather than attached), be the subject of co-authoring, and protected from unauthorised changes (and even downloads).
Limitations with Microsoft 365
It is important to keep in mind that there are some limitations with the current (October 2020) retention capability in Microsoft 365.
Retention and disposal is based on individual digital objects, not aggregations. There are limited ways to group individual records by the original aggregations in which they may have been stored (e.g., document libraries in SharePoint).
Only the (minimal) details of records that were subject to a disposition review are recorded in the ‘disposed items’ listing, and this is only kept for a year (but can be exported). No record is kept of any other destroyed record, except in audit logs (for a limited period).
The metadata details of records subject to a disposition review that were destroyed is minimal – the document type and name, date destroyed, destroyed by whom.
When records are destroyed from SharePoint document libraries or lists, the library or list remains with no record kept of what was previously stored there. It is not possible to leave a ‘stub’ for a destroyed record.
The primary outcome from introducing modern ways to manage retention will be that all digital content, not just content that has been identified as records or copied to a recordkeeping system, will be subject to some form of retention and disposal management.
In other words, a change from exception-based retention (where all the other digital content is overlooked), to a more holistic method of retention with both granular controls on certain types of records where this is required, and broader retention capability allowing us to forget the content that is no longer relevant – the ‘redundant, trivial and outdated’ (ROT) content often scattered across network file shares.
At the 2020 Microsoft Ignite conference, Jeff Teper presented a diagram titled ‘Microsoft 365’. The diagram showed only four icons: Teams, Outlook, Office and Edge.
The implication of this diagram was that, for most end-users, Teams is now (or will become) their primary portal into Microsoft 365. As stated by Jeff Teper, SharePoint is a foundation platform, the out of sight content engine. Edge’s ability to serve up search results from Microsoft 365 further reduces the need to go to SharePoint.
So, what are the implications for managing records?
SharePoint as a recordkeeping system
For a long time, records have been created, captured and stored in recordkeeping systems.
In the paper world, the recordkeeping system consisted of paper records stored in files and boxes and detailed in registers. With the introduction of computers in the 1980s, registers were transferred to databases, making it a bit easier to find records. In the late 1990s, recordkeeping databases were linked with (separate) file stores and became electronic document and records management (EDRM) systems that continued to manage paper records (the so-called ‘hybrid’ systems).
For almost a decade (since SharePoint 2010 was introduced), SharePoint has contended with files shares and EDRM systems as an alternative recordkeeping system, providing almost all the same core functionality.
The ability to create a record in a single location, then share and co-author it from that location, has completely removed the requirement to copy a record to a separate recordkeeping system.
And then came Teams
Someone at Microsoft had incredible foresight to see the potential for a new user interface that would replace products like Lync and Skype for chat and conferencing, and would also provide access to files stored in SharePoint.
SharePoint has been a core part of the Microsoft productivity offerings for a very long time and people have built careers around developing functionality on the SharePoint platform to appeal to end-users, the intranet being the most common case in point, with customised team sites close behind.
The arrival of Microsoft 365 Groups and then Teams in 2017 was perhaps not widely noticed. One could argue that end by the beginning of 2020, it was still largely unnoticed.
And then came a pandemic and working from home. Teams – which may have been largely ignored or overlooked until then – was already ready to take its place next to Outlook, Office and Edge as a primary end-user interface.
New Teams were created, sometimes with abandon (and were sometimes just as quickly abandoned).
Both 1:1 (or 1:many) chats and channel chats took off. Files were created and shared via OneDrive for Business (‘Files’ in the 1:1 chat area), or via the back-end SharePoint sites (‘Files’ in the channel chat area).
There was (and maybe still is) a belief that files were being saved to Teams but not SharePoint. ‘We are storing everything in Teams’ was not an uncommon expression, sometimes followed by ‘but we’re not using SharePoint or OneDrive’.
The year 2020 saw a huge increase in the volume of records stored in SharePoint sites linked with Teams, as well as a completely new set of records – chats (‘compliance’ copies of which are stored in Exchange mailboxes).
The diagram below provides an overview of the relationship between Teams, Microsoft 365 Groups, Exchange mailboxes, SharePoint and OneDrive for Business.
What about SharePoint?
As the diagram above shows, SharePoint has not disappeared. Many organisations will continue to use, and ask end-users to access, SharePoint sites directly to store and manage records.
But accessing SharePoint from SharePoint may become less necessary over time. At Ignite 2020, the ability to pin a ‘home site’ (such as an intranet) to Teams was demonstrated. Even the intranet may end up in Teams.
As Jeff Teper said, SharePoint is a foundation platform, one that does not get in the way of collaboration and productivity but powers it.
Implications for records managers
Records managers, who were likely already on a steep learning curve regarding SharePoint, need to continue to improve their knowledge of the SharePoint platform. On a positive note, SharePoint Online is a much easier application to learn and manage, compared with its earlier on-premise predecessors.
In organisations that have been using SharePoint for a while and/or have allowed the free-creation of Teams in MS Teams, there will some requirement for retrospective analysis, review, and cleaning up.
In all organisations, there will be a requirement to establish some form of governance and oversight of records (files and chats) that have been created, including for the purpose of retention and disposal/disposition.
Where MS Teams has been implemented with little thought given to naming conventions, SharePoint site provisioning, or access controls, records managers should been given access to and review the list of all SharePoint sites that have been created, including from MS Teams. This will provide an initial idea of the volume of content and activity on each site, and what action needs to be taken on things like inactive Teams.
Ideally, records managers should be added to the Site Collection Administrators (SCA) group of every SharePoint site, including MS Teams-based sites. This action will give records managers access to the content on every site and to help advise on the management of records in those sites (including Team-based sites).
The best way to do this is to add records managers to a Security Group and then add that Group to the SCA group of every site. This access could be deferred for sites that contain very sensitive information, although typically records managers would have access to all records, including if they had an EDRMS. And, access is always recorded in audit logs or the local site ‘viewers’ (where enabled) and ‘last modified by’ information.
Access to the chat content of Teams (including 1:1 chats) will not normally be required; some understanding of the content could be inferred from the name of the Team or the SharePoint content. If necessary, Global Admins or a Compliance Admin can run a Content Search across Teams to find chat content, and/or export that content by an individual person or subject.
Records managers will also need to advise on the appropriate retention policy or policies that need to be created and then applied to:
The chat content in 1:1 chats.
The chat content in the various Teams.
SharePoint sites linked with Teams.
OneDrive for Business accounts. An additional consideration is how long the content of inactive ODfB acccounts should be retained via the ‘Storage’ policy (default is 30 days then permanent deletion).
SharePoint sites not linked with MS Teams. This includes whole sites as well as library-based retention policies.
Office 365 Groups (mailbox/SharePoint site). If linked with a Team, a second retention policy is required for the Team chat content retention (second dot point above). For example, one policy ‘GroupABC’ and a second policy ‘GroupABCTeamChat’.
As many of the above retention policies replace the need for backups, records managers need to discuss the options with their IT colleagues.
Forward looking implications
Ideally, there should be some form of governance around the creation of new Teams in MS Teams. These governance arrangements might include:
The necessary access for records managers. For example, Site Collection Administrator on every site, and/or a customised Compliance Admin role to create and access retention policies.
Controls around the creation of new Teams, including naming conventions. If not controlled, what processes will ensure that records are properly managed.
Retention implications. For example, can the new site and/or the channel chat content be covered by another retention policy – e.g., ‘All Teams with assessed low-level working content should be kept for 5 years’.
Simple best practice guidance for all new users, including on how to share and co-author.
Retention policies for all Microsoft 365 content, not just SharePoint.
Reviews of the content of OneDrive for Business accounts of departed end-users, especially for people in senior or decision making positions. It is relatively common practice for end-users to delete (and download) this content before they leave their jobs.
Monitoring and oversight of content, including access to reporting dashboards.
So, is Microsoft 365 just Teams, Outlook and Office (in Edge)?
For many, or not most information based end-users, MS Teams is likely to become the primary interface to Microsoft 365 collaboration team spaces including SharePoint and OneDrive. Just like Outlook, Teams will probably be left open all day.
In theory, the volume of low-value emails, and emails with attachments, should reduce over time.
The developing role of records managers
In this new world, the role of records managers will change from being the curators of records copied to and stored in a separate ‘records and document management’ system, to being records compliance analysts or perhaps, corporate knowledge and information managers and content analysts.
They will learn what the Graph can do, and help to guide AI tools including machine learning and machine teaching, Project Cortex and SharePoint Syntex. They will be responsible for monitoring content across the Microsoft 365 platform, creating and applying retention policies and managing the outcome of those policies, working more interactively with the Graph, and with a range of data.
In organisations that have a requirement to transfer records to archival institutions, the new knowledge and information managers will have a key role in ensuring that this data is suitable for transfer.
They might even have oversight of old paper records gathering dust until they can be destroyed.
Two types of retention policy can be created in Microsoft 365:
Label-based retention policies, where the label is used to define the retention and retention outcomes. Labels must be published in a retention policy, a process that includes determining where the labels will be applied and appear (‘explicit’) to end users.
Non-label-based retention policies, where the policy includes the retention details and the outcomes. As part of the policy creation, these policies are then applied to specific Microsoft 365 workloads where they are mostly invisible to end-users (except in Exchange mailboxes). In SharePoint and OneDrive for Business, these policies create a Preservation Hold library that is only visible to Site Collection Admins and above.
It is possible to apply both a label-based retention policy and a non-label retention policy to the same SharePoint site. In theory, this would allow for (a) everything on the site to be covered by an overarching retention policy and (b) specific libraries or lists to be covered by a label-based policy.
In practice, it gets a little complicated, as described in this post.
Creating the two labels
For the purpose of this post, I will apply the two types of policy to a SharePoint site (‘FinanceAP’) that contains specific types of financial information that needs to be kept for 7 years, but I want to allow other content on the site to be destroyed after 5 years.
Retention labels are created in the Information Governance section of the Compliance admin portal in Microsoft 365. I created a label titled ‘Financial records’ with a retention period of 7 years. I then published that label to a retention policy named ‘Financial Records – 7 years’ and applied it only to the FinanceAP site.
More than one label can be published in the same policy, making this a useful option if your SharePoint architecture ‘maps to your file plan or Business Classification Scheme (BCS) and your records retention classes are based on either. It also allows you to create and add the same retention class for types of records that occur in multiple functions where the classes have the same retention – for example, ‘Meetings – 7 years’ or ‘Policy – 10 years’.
Once the policy has been published to a site or sites, the option (in Library Settings) to ‘Apply label to items in this list or library’ can be used to choose which label will apply to the content in the library, as shown below.
If the column ‘Retention label’ is checked, the retention label name appears in that column.
Non-label retention policy
Non-label retention policies are also created in the Information Governance section of the Compliance admin portal which also (a little confusingly) lists all the label-based policies as well.
The process of creating these policies includes the retention (e.g, 5 years) and retention outcome (delete) definitions, as well as the location where the policy will be applied.
For the purpose of this post I created a retention label named ‘Financial Working Records – 5 years’ and applied it to the same site (only) as the label-based policy.
I should expect now to find a Preservation Hold library (via Site Contents as a SharePoint admin) when something is deleted.
At this point, I have two retention policies, (a) one label-based and applied to the site, and (b) one that applies to the whole site.
What happens now?
In the document library where the label-based policy has been selected, I can see that the retention label (Financial Records) that has been applied to items in this library.
This means that I cannot delete this document unless (as an end-user with edit rights or admins) the retention label is removed. However, as we will see below, another policy is working behind the scenes.
In a document library where no label-based policy has been applied, I can see that no label appears under the Retention label policy. From an end-user point of view, it appears that the record can be deleted – or is it?
As this site is the subject of an ‘implicit’ or invisible retention policy that has been applied to the entire site, any attempt to delete anything will be captured by the back-end Preservation Hold library seen below via Site Contents (visible to Admins only).
Interestingly, any attempt to delete a document from a library where a label-based retention policy has been applied, which is ‘denied’ in the actual library, is recorded in the Preservation Hold library, although the document remains in the original library.
If anyone with access to the Preservation Hold library tries to delete that item there, they will receive this message:
The only way to remove this item is to remove the policy.
(Note – the image above is a small ticket dated 1956 from my grandmother’s visit to Denmark. I used this because the word ‘Kontrolbillet’ seemed appropriate for this post.)
In response to several queries about this following my previous post about whether it is possible to manage records as data, it seemed apparent that the data-based nature of contemporary modern digital content formats, especially Office documents, is not well known.
This post provides details of the data structure content of a typical Word document, to help explain why such records could be seen (and managed) as self-contained data sets.
Just to be clear, the idea of managing records as data does not remove the need or business requirement to store and manage records in ‘local’ aggregations or context – a SharePoint document library or a mailbox for example (less so a network file share because of the limited metadata, but still possible). These aggregations will generally map to business activities, can have specific metadata requirements and can be used to control access to and retention of records as long as they need to be managed.
Managing records as data is a more holistic data analytics concept that allows organisations to better understand and analyse records amidst the volume of all other digital content. It should should help to ensure that all records on a given subject or context are managed appropriately through time, and that, wherever possible, only one copy exists.
A document in a SharePoint Online library
For this example, a document is stored in a SharePoint Online document library called ‘Client Agreements’. The library has a set of metadata columns that must be added to every record. The library uses document sets but it could equally use metadata or folders, the important point is that metadata is added to the library.
The metadata added to the library can be anything, including terms from a business classification scheme. The metadata can be mandatory or optional, and can be set as default options – for example, you may want every document in a library to automatically have the same function and activity terms.
In the screenshot below, we can see the document library with two document sets (a type of folder). The library has four added metadata options: Client Name, Client Reference, ClientRef, and Date of Birth (not visible in the screenshot but we’ll see it later).
The metadata properties
Here are the metadata columns for the library. As we will see below in the actual data, metadata columns with a space between words results in additional characters (‘_0020_’) replacing the space.
When I open the Harpin ‘folder’, I can see the metadata columns next to a document. In this case they were added to the document automatically as the documents inherit the same metadata properties as the document set. This is set via the Document Set settings – ‘Shared Columns’:
Alternatively, the metadata can be added to each new individual document when the document is added.
If the Harpin document is selected as shown below …
… the information panel on the far right shows the metadata properties for the document (and also the activity – when the document was modified and by whom, and who viewed it):
As this particular document is a Word template added to a content type in the library, an end user can to select it when they create a new document in the library as shown in the screenshot below. Alternatively, the ‘Client Folder’ option allows them to create a new document set folder with all the metadata that relates to the client; this data is then inherited by any document created in the library:
If the document is opened, you can click on File – Info and see the metadata properties already added TO the document in the library. These properties remain with the document even if it is downloaded and/or attached to an email. If Document IDs have been enabled, that metadata value is also added to the document properties, meaning we can see that it came from a SharePoint library (and which one):
Because it is used as a template, the Word document can make use of the metadata added to the record in the body of the document, in addition to the metadata forming part of the properties for the document.
The XML properties
Let’s now look at the XML of the document.
Download the document to an accessible location. Using the Command Prompt (CMD), rename the document to .zip (You cannot do this from File Explorer). From File Explorer, the original file will now have the extension .zip. In the list below, the other file with a similar name is a copy, but the size is identical.
Now, unzip the zip file (right click, Extract All).
Here is the top level output, which is standard for all Word documents.
Open the ‘customXml’ folder and you will see a set of XML files:
Open item1.xml, and you will see the custom properties which, as you can see, includes both the Document ID as well as the original path SharePoint site/library location. Just to be clear the Document ID ends in ‘119’, which is the actual document; the original document set folder’s ID ends in 118 (scroll up to check):
As can be seen, the document that was downloaded has the unique Document ID embedded in the metadata. Note that this ID will change if the document is uploaded to a different library.
In the ‘docProps’ folder we find three sets of XML files:
In the ‘coreXML’ file we see the Dublin Core (DC) metadata that you see in the document Properties above. You can add all the Dublin Core metadata to the library, they are built-in to every library, which means that every document can have all that metadata.
The actual content (the body) of the email is found in the ‘word’ folder of the XML files. Here is the content of that ‘word’ folder:
In the screenshot below you can see some of the ‘document.xml’ content including the metadata that has been added in the body of the document (separately from the properties of the document).
All this metadata is accessible and is used by the Microsoft Graph.
Excel files are interesting because, in a sense, they contain data within data. Here is some data in a spreadsheet:
This data is – strangely – stored in two different XML files. The text (including the column headings) is stored here: \xl\sharedStrings.xml:
The values are stored according to each worksheet. For example: \xl\worksheets\sheet1.xml (first two rows only)
A note about emails
Emails do not have the same XML-based structure as Office documents and generally cannot have additional metadata added (except as tags).
Emails in Outlook (sent or received) become ‘.msg’ files if saved to another location from Outlook.
The ‘.msg’ format is based on CFB_3, or compound file binary format, a format that was also used by earlier versions of Microsoft Office documents. It is ‘a general-purpose file format that provides a file-system-like structure within a file for the storage of arbitrary, application-specific streams of data’. (Source: Microsoft web page on Compound File Binary File Format).
Copies of Microsoft Teams chat messages are also stored in a hidden folder in Exchange mailboxes, as instant messages. They cannot be accessed directly but should be considered as a type of archive copy – the originals are stored in a separate database.
If emails are saved to a SharePoint document library, they can be described with additional metadata while stored in the library, but this metadata does not become part of the core metadata of the email or remain with it if it is downloaded, as it does with other Office documents.
In any case, whether they remain in Exchange/Outlook mailboxes, are copied and stored in SharePoint or other Microsoft-based locations, the metadata content in them is accessible via searches.
Active Directory completes the relationships
Every digital record has an author and is likely to have contributors (modified by). Every email is sent and received by someone. All of the internal names linked with digital content are recorded in an organisation’s Active Directory. Employees are also likely to be added to Security Groups (sometimes known as AD Groups) that provide a way to control access to IT resources.
The relationship between document-based content (documents, emails), and between people in AD Security Groups, provides the ability to establish relationships between content, people and business activities.
A final word
Importantly, managing records data does NOT remove or exclude the business need or requirement to aggregate (e.g., in document libraries, mailboxes), manage through time, and then destroy or transfer records according to business requirements. Instead, it enhances this capability by ensuring that all records about a given subject or context can be identified and that, as much as possible, only one copy of the record exists.
(Note – while drafting this post I became aware of an MA Dissertation on the subject of ‘Artificial Intelligence and Record-keeping’ being developed by Mohamed Ben Tahayekt at University College London. I have not had access to this material but I believe some of the concepts may be similar to those outlined in this post.)
Digital records have long been thought of (and described) as being ‘unstructured’.
The reality, however, is that almost all contemporary text-based digital record is made up of a defined, structured and mostly open or accessible package of data that is based on standards. For example:
Microsoft Word, PowerPoint and Excel documents are all based on an XML structure (indicated by the ‘x’ on the end of the file extension) described in ISO/IEC 29500 and ECMA 376.
Google Docs exist only in an online format (described in this Google site); to access them offline they must be converted to one of the following formats ISO/IEC 29500 format, ODT (ISO/IEC 26300), PDF or html.
Emails are now mostly based on the Internet Messaging Format (IMF), standardized by RFC 5322.
PDFs are based on the open standard ISO 32000.
All of these standards support interoperability between systems (and devices). (See my post about Metadata Payloads for more information on this subject).
An exception to the above are binary objects, including digital photos and images and where these are embedded in text-based documents. But even so, most binary objects are stored with a range of metadata to describe them.
Given that text-based digital records are already full of readable and accessible structured data (and binary objects come with a range of descriptive metadata), is it possible to manage digital records as self-contained data objects?
Records and context
Digital content (records and non-records) will be always be captured, saved to or stored somewhere:
In email mailboxes. Emails of course may include attachments that duplicate records stored elsewhere in the system, or are not stored anywhere else – e.g., received from outside the organisation.
In a drive/folder structure in a network file share location, including ‘personal’ drives.
In a library/folder in online file storage and collaboration platforms, including ‘personal’ online storage locations.
In corporate enterprise ‘social’ platforms such as the intranet.
In corporate messaging and chat applications.
Some of the above may have well-defined ‘filing’ or storage structures (including folders) that are used to store or ‘file’ records. Some of these may include the ability to classify and categorise records, and add additional metadata.
In an organisational setting, all of this digital content will be created, sent/received, or modified by someone listed in Active Directory (AD), a system that generally links employees through their organisational structure. Additionally, employees are likely to belong to several AD Groups that further define relationships between them.
These relationships are important as they help us to understand the context for records.
Isolating records from other content
But one of the challenges for any organisation is knowing what is a record and what isn’t. Perhaps that isn’t as important as it sounds, if all the digital content is considered a potential record.
Organisations create or receive and store a lot of digital content, and a lot of this content has traditionally been kept (on backup tapes) for a long time to support disaster recovery and investigation purposes.
Only a percentage of this content is likely to fit the standard definition of a record – ‘evidence of business activities’.
And some digital content may not obviously be a record until it is connected with or related to other content or activities. For example, a simple email that says ‘Yes’ or ‘OK’ may be the record of agreement to something that doesn’t form part of any other obvious records until it is identified as being a record.
Not uncommonly in traditional electronic recordkeeping systems, there could be no guarantee that everything copied there was a copy of every record that existed on a given subject. Additionally, a record stored in a recordkeeping system may be of relevance in other contexts.
The key to what a record might be is the word ‘evidence’; this is exactly what lawyers look for when they conduct eDiscovery activities.
Rather than assume all records can be accurately found and managed amidst the volume of all digital content, it may be more efficient and accurate to assume all digital content is a record and then apply rules and tools to manage that content, with the aim of identifying records and their potential context based on the data contained in the individual digital objects and their relationships with both other records and people.
In other words find records amongst the entire content, rather than seeking to isolate only those digital objects that are identified as records and copy them to another location – while leaving the originals and potentially other related records in place. Managing records this way avoids the problem of email threads or chats that continue after the copy has been made, or a new copy of a Word document appearing.
How can we achieve this outcome?
There are three potential ways to manage records as data.
The first is to understand, even in general terms, is that digital content is not unstructured, and to learn more about how they are structured. Some simple examples:
Every email (and instant messages) has a sender, recipient, date sent, date received. They usually (but not always) have a subject. The text-based body of the email provides an additional form of accessible data. A quick look at email headers reveals a huge amount about the email.
Every document (and web page) has an author, created dated, modified date and last modified by, and a name. They also have a large amount of other data, a lot of which is visible in the Properties section.
Photographs are stored as binary objects but have a range of EXIF metadata that includes the creation date, information about the camera settings, and may also include the name of the person who created it, as well as a GPS location.
The second is to understand that digital content may include added data or metadata. This added data may relate to or derive from the location where the record is stored, or may be added by end-users as part of their work. It may include a unique identifier and information about the aggregation where it is stored, as well as recordkeeping classification terms. Additionally, it may include both process metadata (modified by, and when) and security or access control metadata. Depending on where it is stored, this additional metadata may be embedded with the document properties (the metadata payload).
The third is to have access to (ideally) all digital content across the organisation, and the necessary tools (or access to people with them who can provide usable output) to search and retrieve, relate, and manage all digital content on any given subject or context through to disposal. A very simple example of this is to run a PowerBI report across the network file shares.
And lastly, while there will always be some form of ‘local’ aggregation for specific records where all the records are stored in the one place (mailbox, document library, folder), the only way to establish an aggregation of all digital records on a given subject or context using data only is through the use of advanced searches and/or eDiscovery tools and/or data reporting or visualisations and/or artificial intelligence to find, link and relate content.
Linking and relating content
The diagram below from Microsoft about its Graph technology, provides a simple example of how content can be linked and related through its data.
There are now many data analytics and data visualisation tools that help to understand digital content. These tools are just one part of the picture.
Data analytics tools (such as ‘Constellation‘, a joint project between the Australian Signals Directorate and the Australian CSIRO) are a starting point to understand digital content – including digital content from line of business systems.
These tools might be used to identify content or people related to a given subject, through chat messages, emails, documents etc, including content that is already linked through its own context – the mailbox or a SharePoint library. From that information it would be possible to build a picture – types and volume of content, people, and the relationships between them.
A global search should be able to retrieve, and if necessary export, the content, keeping in mind always that the nature of digital content means it may continue to be modified or new content may added at any time.
As searches improve, narrower set of content allows more granular analysis and visualisation, allowing the identification of sub-sets of records within broader sets. For example, of the potentially large group of ‘everything about COVID’, just the narrower set ‘Vaccines’.
All of this could be achieved through the data that makes up the digital content. And many data-driven organisations are likely to be doing just this, using a range of business intelligence tools to understand the information available to them, in both line of business systems and other content.
Can we manage records as data
Perhaps ‘manage’ is not the right word, or at least not in the sense of expecting digital records to be managed as exceptions to the rest of the digital content.
But there is a lot more we can do to make this outcome possible. We can leave the records where they are stored or captured, we can apply local structure to those records, or security controls. We can keep records of changes that are made. We can apply retention rules that prevent the destruction of any record, or potential record, before it can be legally destroyed.
Instead of ‘managing’ records as exceptions, we can leave the data where it was created or stored, and use a range of tools to help us understand and manage it.
This will allow us to manage records as data and finally achieve the ‘semantic office‘ I wrote about in 2010.
The international standard for records management, ISO 15489-1:2016 (‘Information and documentation – Records management – Part 1: Concepts and Principles’), defines records as ‘information created, received, and maintained as evidence and as an asset by an organization or person, in pursuit of legal obligations or in the transaction of business’.
Among other things, the standard notes that records systems may exist in a variety of forms, not necessary as or in a single or dedicated application. It also underlines the importance of appraisal; that is, the recurrent analysis of business context, business activity, processes and risk for the purpose of determining what records to make and keep and how to manage them over time – especially given the complexity of contemporary recordkeeping.
In terms of risks, the standard states that risk management is required to develop strategies for managing records and the management of records as a risk management strategy in itself.
Unlike traditional electronic document and records management (EDRM) systems that are used to store copies of records created and stored in other applications (‘exception management’), the Microsoft 365 environment is a single system in which records are a sub-set of the entire content (‘exception identification’).
This post discusses how records can be collated, grouped and aggregated in Microsoft 365 to meet requirements for management records. It emphases the point made in the international standard that the risk to records should be understood and minimised.
Records and context
Records are usually created or captured in some form of context – for example a business activity or project. This in turn provides the basis for collating, grouping or aggregating those records according to that context – commonly, a ‘subject’ or ‘topic’.
Records may be a subset of a broader subject (or series). They may be relevant or relate to more than one context or subject.
Digital records that may have no obvious context when they are first created or capture (for example a casual email about an ‘unusual virus outbreak’ in November 2019) may form part of a specific context only when their value is recognised (‘global pandemic’).
Grouping digital records
Grouping records in the digital world has up until now usually involved copying a digital record, created or captured in one system (such as email or a network file share), to a digital ‘file’ in another system such as an electronic document and records management (EDRM) system. The digital ‘file’ in those systems is a virtual representation; the records are actually stored in a file share, linked by metadata in the form of a file number.
The grouping of digital records as exceptions had (and continues to have) several flaws:
It assumed that all types of digital records could be stored in a digital ‘file’ from where they could be faithfully and reliably rendered (and not just stored as zipped versions of exported content from the originating system).
It relied on the willingness of end-users (often after training) and/or a technical third-party system, to copy a record to the system. This ‘exception management’ meant that some records were not copied to the EDRMS.
It was a ‘point in time’ capture. The original digital record remained in the system where it was created or captured, and might also be attached to emails and from there saved to multiple other locations.
There was no way of knowing if all the records in the file were all the records relating to the subject.
Where are the records created or captured in Microsoft 365
Most business records in Microsoft 365 will be created or captured in Outlook/Exchange mailboxes, SharePoint site libraries or MS Teams (which stores chat in Exchange mailboxes and documents in SharePoint or OneDrive). (For the purpose of this post, OneDrive is seen as a personal working space that should not be used to store business records.)
Regardless of whether they are created or captured in Exchange or SharePoint (including via Teams), all of the content – records and non records – created or captured in Microsoft 365 is stored in the Azure substrate. This effectively means that records in Microsoft 365 are a sub-set of all the other content stored in the Azure substrate.
Consequently, the management of records in Microsoft 365 involves exception identification. That is, identifying records and ensuring they are managed appropriately as much as possible where they are captured or created – and placing other controls over all the other content as necessary.
Everything created and stored in Microsoft 365 – including all the very rich metadata associated with every digital record – is subject to the Graph. The Graph identifies relationships and ‘signals’ not only between digital content but between people (agents) and business activities.
The Graph powers Delve and Discovery and the soon-to-be-released Project Cortex, presenting information (they have access to) to end-users that can sometimes be unsettling for people used to working in relative privacy. See below for further discussion about Project Cortex.
Additionally, as all the content in Microsoft 365 is stored in the Azure back-end, most of it can be searched and (where necessary) exported through the Content Search option in the Compliance portal, a capability that supports eDiscovery. This capability means that even when records are not ‘manually’ identified as records, there is a better chance they will be found.
How are records aggregated in Microsoft 365
There are three main ways that records are, or can be, aggregated in Microsoft 365: Exchange mailboxes, SharePoint site libraries, and Microsoft Groups that have a mailbox and a SharePoint site and can be linked to (or created from) a Team in MS Teams.
Exchange aggregates email records by:
Personal mailboxes, accessible only the ‘owner’ (end-user).
Shared mailboxes, accessible to those who have access.
Microsoft 365 Group mailboxes, accessible to the members of the Group (including anyone added to the Group).
Although a mailbox is a form of aggregation, there is no way to relate or link emails stored there with other related records stored in SharePoint unless they are copied to a SharePoint document library, as can be seen in the example below. This is recommended if an organisation wants to keep emails together with other records.
Emails copied to a SharePoint document library are a ‘point in time’ copy; there may be additional replies to the email, forming a thread that isn’t captured.
The alternatives to copying emails to SharePoint are:
Leave all emails in mailboxes and use Content Search to find and export them to SharePoint as a PST.
Creating a Microsoft 365 Group with an associated mailbox and SharePoint site, so that the records are retained in the context of the Group.
In any case, all mailboxes should be subject to a minimum retention period to ensure that any email that might be a record is preserved for that period. Certain mailboxes (for example, senior or key staff members) may be kept for longer periods and then exported for permanent storage.
SharePoint document libraries are logical aggregations for the storage of records, including emails copied from Exchange mailboxes.
Ideally, individual libraries that are used for the storage of records should map to a business activity and/or records retention class; this mapping should be reflected in the library name.
NOTE: Individual document libraries should not be used to store records relating to multiple subjects or mapping to more than one retention class or policy.
Document libraries may be assigned as much metadata as required, and content stored in them can be defined through the use of metadata and/or content types.
Microsoft 365 Groups (including Teams in MS Teams)
Microsoft 365 Groups provide a way to group and manage records, including MS Teams channel chats, in the context of the Group.
Every Group includes a mailbox (visible in Outlook) and a SharePoint site, and can be linked to new Team in MS Teams. Teams channel chats are stored in a hidden folder in the Group mailbox. Any documents and records are stored in the ‘Files’ tab of the channel, which surfaces the default ‘Documents’ library in the connected SharePoint site.
If the creation of Teams is allowed from the MS Teams application, every new Team creates a Microsoft Group (with the same name) and a SharePoint site (with the same name), however the mailbox (with the hidden folder for channel chats) is not visible from Outlook.
(The exception here are private channels; if these are allowed: (a) the chat content is stored in the Exchange mailbox of the each participant, and (b) a new SharePoint site is created for the ‘Files’.
The relationship between the content created by the Group is most obviously visible from the ‘Activity’ web part of the SharePoint site of the Group as can be seen in the screenshot below. This shows (right to left), an original incoming email from Outlook in the Group’s mailbox, the copy saved to the SharePoint document library, and the Word document reply. The specific context of the record (= the ‘file’) – ‘Correspondence 2020’ – is defined by the document library.
What about records in 1:1 Teams chat
As with OneDrive, Teams 1:1 chat should not be used to create or capture records, but may be used as a ‘working’ space.
However, ‘should’ and ‘reality’ can be different things. There are two ways to address this:
Explictly, through communication to end-users. Make it clear that Teams 1:1 chat and OneDrive are NOT to be used to create or capture records. Applying short-term retention policies to this content may assist with reducing (or increasing) this risk.
Implicitly, through monitoring and retention policies. Apply longer-term retention policies to the content and use Content Search/eDiscovery to look for content that may be records. Additionally, review the content of the OneDrive of departed staff and ensure that any records are kept.
Implications for managing records
The implications for collating, grouping and aggregating records in Microsoft 365 are as follows.
SharePoint document libraries will continue to be the primary aggregation for managing corporate records, including emails copied from Outlook.
Organisations should establish an architecture model for SharePoint sites that are used to manage records. The model may include a mix of the following: (a) sites mapped to business functions with libraries mapped to business activities and retention classes, (b) entire sites used to create and capture records relating to a single activity, where the entire site is mapped to a retention class, and (c) MS Groups (and Teams) with an associated SharePoint site, where the Group (mailbox/SharePoint site) is subject to a single retention class (and the Team channel chat also).
More effort, in terms of site/library set up, metadata, access controls, retention and end-of-retention process is likely to be required for the management of high-level, high-risk and permanent records.
Personal mailboxes in Exchange will continue to exist as a form of aggregation, and consideration should be given to having different retention policies for different ‘types’ of mailbox, to ensure that any email that could be a record is not deleted too quickly.
Addendum – Other options that collate, group and aggregate content in Microsoft 365
As noted earlier, all of the content created or captured in Microsoft 365 is stored in the backend Azure substrate. Consequently, it is possible to search across all or part of that content to find related information and, where required, export it to a different location.
The global Content Search is accessed from the Compliance portal and access requires elevated privileges – Global Admin or Compliance Admin.
Searches are created as cases and are based on keywords, conditions (such as ‘Sender’ for emails), and locations – all or specific. When a new content search is created or run, the Global Admins are alerted, providing a form of oversight in addition to audit logs.
While content searches find content is related to the search parameters, and legal holds can then be applied to that content, they do not create any form of aggregation in a recordkeeping sense.
The Graph, Delve, Discovery
Microsoft describe the Graph as being ‘the gateway to data and intelligence in Microsoft 365 [that can be used via the Microsoft Graph API] to access the tremendous amount of data in Microsoft 365, Windows 10, and Enterprise Mobility + Security’ and ‘… build apps that support scenarios spanning across productivity, collaboration, education, people and workplace intelligence, and much more. (Source ‘Overview of Microsoft Graph‘)
The Graph is commonly represented in diagrams similar to the one below.
Most end-users will encounter the Graph through either Delve or the Discover option in both the office.com portal and their OneDrive for Business accounts.
It is not uncommon for end-users to express surprise at the content (that they have access to) that is presented. Commonly this will show documents that a colleague is working on, or connections between people. Disabling Delve does not fix permissions; if a person has access to a document that appears in Delve, they will be able to search for it and find it that way.
Over time, the Graph can also provide other information based on the relationships or ‘signals’ it finds between all the different content in Microsoft 365.
While the Graph can present groups of records that have some relationship to the end-user, it does not aggregate those records or maintain a single consistent view. However, the Graph powers the new Project Cortex that does do something similar.
Project Cortex was announced by Microsoft in April 2019. To quote the announcement, Project Cortex:
Uses advanced AI to deliver insights and expertise in the apps you use every day, to harness collective knowledge and to empower people and teams to learn, upskill and innovate faster.
Uses AI to reason over content across teams and systems, recognizing content types, extracting important information, and automatically organizing content into shared topics like projects, products, processes and customers. Cortex then creates a knowledge network based on relationships among topics, content, and people.
From a recordkeeping aggregation point of view, a core functionality of Project Cortex is its ability to create ‘topic cards’ based on the rich metadata that makes up all the content in Microsoft 365. Again to quote the announcement:
Project Cortex securely collects content that is created and shared every day in Microsoft 365—including files, conversations, recorded meetings and video—and it categorizes the content based on its type, and tags it with extracted metadata.
AI then applies advanced topic mining logic—whether its content contained in Microsoft 365 or connected from external systems—to identify topics and relate content to those topics.
Topics can reflect any knowledge that’s important, including customers, products, projects, policies and procedures. Technically, AI is creating knowledge entities, a new object class, in the Microsoft Graph. The relationships between those topics—those knowledge entities—and the experiences that connect this knowledge with people creates your knowledge network.
Topic cards – or ‘knowledge entities’ – are a form of AI-generated aggregation.
However, topic cards will only present information that an end-user has access to and so the nirvana of presenting emails or Teams 1:1 chats in these cards as a form of aggregation for recordkeeping purposes is not likely to be realised through Project Cortex.
There are two ways to create retention policies in Microsoft 365 – (a) by creating a label and publishing it as retention policy, or (b) creating a retention policy without a label. Both are created from the Information Governance section of the Compliance portal in Microsoft 365.
This post describes how the policies are created and applied, and the outcome for each.
Retention policy based on a label
Retention labels are created from the Labels tab of the Information Governance section in the Compliance portal.
Ideally, the name of the label would be based on a disposal class in a records retention schedule or records disposal authority. For the purpose of this post, the new label was named ‘Test One – Retain five years then delete‘.
Adding the details of the retention provided a quick visual clue to the retention details.
The label was configured to have a retention period of 5 years after the date modified after which the content could be deleted, with no disposition review. No record would be kept of what was deleted.
Alternatively, it would be possible to select the Disposition Review option. This option sends an alert to the account indicated so the records could be reviewed before disposal. However, keep in mind that if the trigger is left as ‘Date created’ or ‘Date modified’, records will be ready for disposition review based on those dates. It might be better to choose ‘Date applied’ instead so that a group of records could be reviewed at the same time – for example, in a SharePoint library.
Retention labels have no life unless they are published. This is achieved from the ‘Publish labels’ option in the same tab.
One or more labels can be published in a retention policy. This could be useful if the records disposal authority contained multiple labels for a single function.
As part of the publishing process, a decision has to be made to apply the policy to somewhere in Microsoft 365. In this case, the decision was made to apply it to a single SharePoint site because that site contained records that mapped to the retention label.
Finally, the new policy was given a name. In this case the policy had only one label so the policy had the same name as the label.
If the organisation’s records disposal authority/retention schedule had multiple retention classes mapped to a business function, the policy could be named after the function and include all the relevant classes created as labels.
All of these labels would appear in the drop down list from the Library Settings on every site where the policy was applied.
But even when it was published in this way and applied to a specific SharePoint site, the label does not do anything. The policy has to be applied to a document library in that site – but even then it wouldn’t do anything until the label was selected from a drop down list.
The screenshot below shows how the specific label is selected on the library and whether it should be applied to all existing items.
The label can now be seen in the ‘Retention Label’ column (when made visible in the view settings).
Anyone who uses the library to add and edit content (site members) will notice an obvious change if they try to delete anything, including from their synced document libraries in File Explorer.
There is, however, a workaround that end-users can use to delete documents with a label: (a) check the box next to the document, (b) open the information panel and (c) remove the label on individual documents.
Once the label is removed it no longer appears against the record.
The document can now be deleted. It will move to the Recycle Bin, from where it could be restored for up to 93 days if a mistake was made.
The lessons to be learned from this are that:
It might be useful to make the library read only after the label is applied.
The label should only be applied when the library became inactive, to allow people to get on with their work while it is still active.
If anyone was concerned about content being deleted from an active library, they could set an alert on the library.
In the meantime, for all the other records that retained their label, when the retention period ended the records in the library would be transferred to the Recycle Bin where they would sit for 93 days before complete and permanent deletion.
If no-one restored a document from the Recycle Bin, the content would be deleted, permanently, forever, without any record. The records managers might need to make a copy of the metadata for the documents to be deleted before they are deleted.
Alternatively, if they had selected the option for a ‘disposition review’ when the label was created, this would alert the records managers (and others) of the need to review the contents of the library.
Once everything had been deleted, all that will be left will be an empty library with no record of what used to be there.
Retention policy without a label
Non-label retention policies are created from the ‘Retention’ tab of the Information Governance portal.
For the purpose of this post, this label will be named ‘Label Two – Retain 5 years then destroy’.
This policy has the same retention period, trigger, and action as Label One – 5 years after date modified – and then the content could be deleted without any record of this being kept. [Note – yes the screenshot below shows 7 years].
As the policy was being created it was applied directly to a SharePoint site.
The policy could start working immediately.
Nobody, apart from the person who created and applied it knew it had been applied. It didn’t appear anywhere on the SharePoint site where it had been applied.
But, as part of the application process, it created a Preservation Hold (PH) library visible only to Admins, including the Site Collection Admins but not the Site Owners.
(Note – if end-users are allowed to create SharePoint sites, or Teams in MS Teams that also creates a SharePoint site, they are made Site Collection Admins by default and can accordingly see the Preservation Hold Library).
Anything that was deleted in advance of the retention policy expiry would be moved to the Preservation Hold library.
The PH library had a specific purpose – to ensure that everything on the site was retained until the end of the retention period.
It was not possible to restore any items from the Preservation Hold library. It was also not possible to ‘clear’ that library. If an administrator tried, they would be advised that ‘something went wrong’.
The only way to delete from the Preservation Hold library would be to remove the policy.
However, if a record was deleted it would be placed in the Recycle Bin for up to 93 days, from where it could be restored. If a record was restored in this way (from the Recycle Bin), the copy in the Preservation Hold library would remain.
For most people this type of policy was fine. They could (appear to) delete records, unaware that nothing was actually deleted if it were subject to a retention policy.
The same thing happened if a similar type of retention policy was applied to OneDrive accounts. End-users could ‘delete’, but nothing was actually deleted as long as the retention policy was active.
At the end of the retention period, the content that was subject to this policy was transferred to the Recycle Bin where it remained for 93 days. It would then be deleted, permanently, forever, without any record.
All that was left would be an empty library with no record of its previous content. The only clue would be in the ID or the ID part of the Document ID (if enabled). If anything new was added to the library, the next number in sequence would indicate how many records had previously been saved to that library.
Lessons to be learned
There are several lessons to be learned.
The first is that you need a plan to create and apply retention labels or retention policies to content across Microsoft 365. The configuration options and implications of each option need to be understood before they are applied.
You need to understand the difference between (a) a retention label published as a policy and applied to a SharePoint site and then a library on that site, and (b) a retention policy applied to an entire SharePoint site.
Retention labels can be mapped to retention schedule/disposal authority classes and then applied to specific libraries on SharePoint sites where the policy is applied. Accordingly, it may be useful to create document libraries that map to those classes, and only store content in those libraries that is covered by a single retention class.
If single SharePoint document libraries are used to store content that maps to different retention classes, then it will become very complicated to accurately map retention labels to content.
In many cases, it may be possible to apply a single non-label retention policy to a single SharePoint site or sites. For example, all project sites might have the same 7 year retention applied to all the content on the site as there may be no real need to apply policies to individual libraries.
Whatever you do, have a plan to act, document your settings, and keep it up to date.
And understand what happens at the end of the retention period. There is no back up.
In the article, Tony describes where and how the chat component of MS Teams is stored and how this might affect eDiscovery.
He also makes the important point that, while it may be possible ‘… to backup Teams by copying the compliance records in an Exchange Online backup … you’ll never be able to restore those items into Teams.’ In other words, it is better to leave the data where it was created – in MS Teams. The post explains why this is the case.
This post draws on the article to describe the factors involving in managing the chat element of Teams as records. It notes that, while is is technically possible to export chat messages (in various ways), it may be much better from a recordkeeping point of view to leave them where they are and subject them to a retention policy.
Two key reasons for leaving chat messages in place are: (a) chat messages are dynamic and may not always be a static ‘thread’, and (b) the chat messages exported from Exchange may not contain the full content of the message.
What is a Teams chat?
A Teams chat consists of one or more electronic messages with at least two participants – a sender and a receiver.
There are two types of chat message in MS Chat:
One-to-one/one-to-many ‘chat’ (top icon above).
Channel-based Teams chat (second icon above). Teams chat is visible to all members of the Team. Within channel-based chats, a person may create a private channel which is visible only the person who created the private channel and any participants.
Messages created in both options could be regarded as records because they may contain evidence of business activity.
However, one-to-one chats have no logical subject or grouping. Only the chat messages in Team channel chat are connected through the context of the Team/channel.
Where and how are chat messages stored?
The following is a summary from Tony Redmond’s article.
Chat messages are stored directly in the backend Azure Cosmos DB (part of the so-called Microsoft 365 ‘substrate’). The version in the database is the complete version of the chat message.
The messages are then copied, less some content elements (for example: reactions, audio records, code snippets), to a hidden folder in either (a) end-user mailboxes for one-to-one chat and private channel chats, and (b) M365 Group mailboxes for channel chat.
Most export options, including the export option in Content Search and eDiscovery, draw their content from the mailbox version of the message. This has potential implications for the completeness of the chat message as a record.
Additionally, any export can only be a ‘point in time’ record unless there is absolute certainty that all chat on a given subject have ceased.
Implications for records managers
In addition to the concerns about a chat message (or exports of them) being complete, there are (at least) two other points relating to the management of chat messages as records in MS Teams:
Knowing if chat messages on any given subject exist.
Applying an appropriate retention policy.
Both of these points are discussed below.
The primary way to locate content on any given subject across Microsoft 365 is via the Content Search option in the Compliance portal. Access to the Content Search option is likely to be restricted. So, if records managers do not have access, they will need to ask the Global Administrators to conduct a search.
Searches can be configured to find content in any or all of the following locations:
Users, Groups, Teams
Office 365 group email
Skype for Business
Teams messages [the copy in the mailbox]
Office 365 group sites
Exchange public folders
Note that content search only works on the copies of the items in the Exchange mailboxes, not the backend Teams database. Accordingly, there is some potential for it to not find some content.
Both the mailbox content and the content discovered by the search can be exported. Teams chat messages can be exported as individual items or as a PST – but note that these message may exclude the elements as described in Tony’s article.
The problem with exporting the content either this way or via other export options (such as described in this post ‘How to export MS Teams chat to html (for backup)‘ (using the Microsoft Graph API) is that it creates a single ‘point in time’ copy; additional content could be added at any time and, if the chats were subject to a retention policy, they may already be deleted.
Managing chat messages ‘in place’ as records
As any export only creates a ‘point in time’ version, it makes more sense from a recordkeeping point of view to leave the chat messages where they are and apply one or more retention policies to ensure the records are preserved.
Ideally, organisations that may create or capture records on a given subject will have taken the time to establish a way for users to do this, including through the creation of a dedicated Microsoft 365 Group with an associated SharePoint site and Team in MS Teams.
For example, if there is a requirement to store all records relating to COVID-19, it would make sense (at the very least) to create a Microsoft 365 Group with that name; this will create: (a) a linked mailbox accessible by all members of the Group, (b) a SharePoint site with the same name, and (c) a Team in MS Teams. All of the content – emails, documents, chat, is linked via the same (subject) Group.
This model makes it easier to aggregate ‘like’ information and apply a single retention policy. It assumes there is (or will be) some degree of control over the creation of Teams (or very good communication to users) to prevent the creation of random Teams, Groups and SharePoint sites – AND to ensure that end-users chat about a given subject within a Team channel, not in one-to-one chat.
What retention period should be applied to chat messages?
The retention period applied to either one-to-one or Team channel messages will depend largely on the organisation’s business or regulatory requirements to keep records. There are two potential models.
The simplest model is to have a single retention policy for one-to-one chats, and a separate retention policy for all Teams channel chats.
As one-to-one chats are stored in the mailboxes of chat participants, it makes sense to retain the chat content for as long as the mailboxes. However, some organisations may seek to minimise the use of chat and have a much reduced retention period – even as little as a few days.
The creation and application of retention policies to Teams channel chat may require additional considerations. For example:
As every Team is based on a Microsoft Group that has its own SharePoint site, it is probably a good idea to establish Teams based on subjects that logically map to a retention class. For example, if ‘customer correspondence’ needs to be kept for a minimum 5 years, and there is a Group/SharePoint site/Team for that subject, then all the content should have the same retention policy – although the Group mailbox and SharePoint site may have a policy applied to the Group, with a separate (but same retention period) applied to the Team.
There may be a number of Teams that contain trivial content that does not need to be retained as records. These Teams could be subject to a specific implicit policy that deletes content after a given period – say 3 years.
In all cases, there is a requirement to plan for retention for records across all the Microsoft 365 workloads.
What happens to chat messages at the end of a retention period?
At the end of a Microsoft 365 retention policy period, both the mailbox version and the database version of the Teams chat message are deleted. To paraphrase Tony’s article, the Exchange Managed Folder Assistant removes expired records from mailboxes. Those deletions are synchronized back to Teams, which then removes the real messages from the backend database.
No record is kept of this deletion action except in the audit logs. Accordingly, if there is a requirement to keep a record of what was destroyed, this will need to be factored in to whatever retention policy is created.
When people chat in Microsoft Teams (MS Teams), a ‘compliance’ copy of the chat is saved to either personal or (Microsoft 365) Group mailboxes. This copy is subject to retention policies, and can be found and exported via Content Search. But what happens if there is no Exchange Online mailbox? It seems the chats become […]
In his April 2007 article titled ‘Useful Void: The Art of Forgetting in the Age of Ubiquitous Computing’ (Harvard University RWP07-022), Viktor Mayer-Schönberger noted that the default human behaviour for millenia was to forget. Only information that needed to be kept would be retained. He noted that the digital world had changed the default to […]
At the 2020 Microsoft Ignite conference, Jeff Teper presented a diagram titled ‘Microsoft 365’. The diagram showed only four icons: Teams, Outlook, Office and Edge. The implication of this diagram was that, for most end-users, Teams is now (or will become) their primary portal into Microsoft 365. As stated by Jeff Teper, SharePoint is a […]
Tony is a contributing author to the e-book ‘Office 365 for IT Pros‘, essential reading for anyone doing anything with Microsoft 365. Page 921 of the May 2020 edition contains the following paragraph, which expands on the quote above and contains probably the best guidance ever required in relation to this subject:
It is sensible to write down each of the retention labels that you plan to use before creating anything. It is much easier to delay the release of a label and the training of users to use the label properly than it is to launch a label into general circulation only to discover that you later need to withdraw it. Another thing to consider is how easy it is for users to decide between different retention labels when the time comes for them to apply a label. Too many labels, misleading names, or too much choice can lead to frustration and bad decisions.
How do you go about writing down each of the retention labels as part of a plan – especially for a Microsoft 365 environment that is already in full swing?
This post provides some suggestions to help you do this.
What is your records retention and disposal status?
A good starting point is to establish the current records retention and disposal status for your organisation. Do you have a records retention schedule, also known as a disposal authority or records authority?
If you have one of these documents, it would be useful to review it as a key part of the process is to ‘map’ the records retention classes to specific records across the various Microsoft 365 ‘workloads’ (e.g., Exchange, SharePoint, OneDrive, MS Teams etc), not just in one system (such as SharePoint).
You will need to know what and where these workloads are.
Where (and what) are the records in Microsoft 365?
If you are a records manager then there is a reasonably good chance that you have very little access to, or visibility of, all the content stored across Microsoft 365.
You may have access to one or more SharePoint sites, but unless you are a SharePoint Admin or Site Collection Admin on every site, your visibility will be very limited.
Most of the records in Microsoft 365 will be stored in Exchange, SharePoint, OneDrive for Business, or MS Teams.
Emails created and sent by users are stored in Exchange mailboxes. There may also be public mailboxes. Unless there is a plan (or third-party app) to copy these (or some of these) emails out of Exchange (e.g., to SharePoint), most email records will probably remain in user’s mailboxes.
Records that, in the past, would have been saved to a network file share (or EDRMS) will now be in SharePoint Online (corporate content) or OneDrive for Business (ODfB) (personal/working content).
Chat messages in MS Teams are stored in a hidden area of the Exchange mailbox of each user who participates in the chat. Any documents shared in this chat area are stored in the OneDrive for Business of the person who shared the document.
Channel-based Team chat messages in MS Teams are stored in a hidden area of the Exchange mailbox of the Office 365 Group linked with the Team. Any documents shared in this chat area are stored in the SharePoint site of the Office 365 Group linked with the Team.
So, fundamentally, records are stored in two primary workloads: Exchange mailboxes and SharePoint/OneDrive for Business.
What are the retention options?
There are two retention options in Microsoft 365. Both are configured in the Compliance portal of Microsoft 365. Access to this portal requires special privileges, which may not always be granted to records managers.
The two options are:
Retention labels published as retention policies and then applied to the various workloads (Exchange email, SharePoint, OneDrive, Office 365 Groups (Exchange/SharePoint content)). These are sometimes described as ‘explicit’ policies because they are visible to end users. Organisations with an E5 licence can extend the way these labels are applied and retention managed.
Retention policies that are applied directly to the various workloads (Exchange email, Exchange public folders, SharePoint, OneDrive, Office 365 Groups (Exchange/SharePoint content)). These are sometimes described as ‘implicit’ policies because they are not visible to end users. These policies automatically delete content at the end of a retention period, without any review possible.
Records managers will need to determine how to ‘translate’ each records retention class into one of the two options above, and how and where it will be applied in Microsoft 365.
Some of the options may also require the creation of new records retention classes – for example for the chat element in Microsoft Teams.
A suggested first model
Your IT probably already has some form of back-up regime (‘archive’) for mailboxes, used for disaster recovery and investigation purposes.
It might be worth creating two policies for mailboxes:
All end-user mailboxes could have a single ‘implicit’ retention policy (e.g., 7 years).
Mailboxes for specific staff (e.g., senior managers) could have a second, longer, ‘implicit’ retention policy. This policy will take over when the first one expires, but just for the mailboxes identified.
The use of retention policies in this way can replace the need for mailbox backups. No emails will ever actually be deleted while the retention policy is in place and all content can be retrieved via the Content Search option in the Compliance Portal.
Content Searches can also be used to retrieve and export emails.
OneDrive for Business
As with end-user mailboxes, OneDrive for Business accounts are generally inaccessible to records managers. To ensure that the content in those accounts is not deleted, a single Microsoft implicit retention policy of, say, 7 years could be applied to all ODfB accounts. This policy will create a hidden (to the user) ‘Preservation Hold’ library on the ODfB account.
Anything ‘deleted’ by the end user during the retention period will be moved to the Preservation Hold library, which is visible to the Global Admins and SharePoint Admins from this URL – /_layouts/15/viewlsts.aspx?view=14
In addition the OneDrive settings include the option (under ‘Storage’ in the ODfB admin portal) to retain OneDrive accounts for a period of time after they are inactive.
All content in these locations is accessible from a Content Search.
SharePoint is likely to be the most complicated in terms of retention policies if there is a requirement to keep content for different periods of time in accordance with the retention schedules/records disposal authorities.
There are likely to be three main options in relation to SharePoint content:
One or more implicit retention policy/ies applied to one or more sites. When applied to a SharePoint site, a ‘Preservation Hold’ library retains anything that is ‘deleted’ by end users.
One or more explicit label-based retention policies applied to one or more sites. When applied to a SharePoint site, the option to apply it appears for each document library on the site. Once applied (manually), end users cannot delete anything and if the library is synced to File Explorer, the File Explorer view of the library will be read only.
A combination of implicit and explicit retention policies.
The decision to apply what policy to what site will depend on your SharePoint architecture and the content stored in each site. For example:
A SharePoint site that only stores records that map to one records retention class could have either a single implicit policy (if there is no requirement for disposal review) or a single explicit policy that is applied manually to each library.
A SharePoint site that contains records that map to multiple retention classes, but for one business function and also ‘working papers’ could have (a) one implicit policy to cover the working papers and (b) one label-based retention policy with multiple labels – one for each class. This means, for (b), that a specific retention label can be applied to each library as required.
SharePoint sites linked with Office 365 Groups and Teams. Depending on the content in the site, it may be possible to apply a single retention policy for all M365 Groups (which covers both the SharePoint site and the mailbox), or a similar policy created for a Group of SharePoint sites (which excludes the mailbox).
As noted above, the chat content in MS Teams is stored in Exchange mailboxes – (a) the mailbox of each participant for one-to-one chat, and (b) the mailbox of the Office 365 Group for channel-based chat.
You may consider having a relatively short-term retention period for one-to-one chat. The retention period for the channel based chat will depend on the subject matter and should – ideally – be the same as for the linked SharePoint site. For example:
A Team set up for a specific business function and activity (or activities) will have channel based chat and a linked SharePoint site. Both should be subject to the same retention period.
A Team set up for low-level discussion about a subject that may be not be covered by any retention period could be subject to a general retention policy for the chat and the SharePoint content.
Bringing it together
As noted at the beginning of the post, if you are going to use retention policies in Microsoft 365 you need a plan and you need to document it. It doesn’t matter too much if the environment is already active.
However, you will need to have discussions with your Microsoft 365 Global Admins, Compliance Admins and SharePoint Admins and know where the content is stored.
The Global Admins can give you a list of every Office 365 Group and Team in MS Team (these are connected – every Team is based on an O365 Group).
The SharePoint Admins (or Global Admins) can give you a list of every SharePoint site.
There are some potential ‘quick wins’, such as agreement with IT regarding Exchange mailboxes, OneDrive for Business accounts, and MS Teams.
The more complex requirement is to map the classes in your records retention schedules/disposal authority to content stored in SharePoint, including for standard sites (not linked with Microsoft Groups), communication sites, and sites linked to Office 365 Groups.
You can start to do this by having a list of all the sites exported from the SharePoint Admin portal. This should allow you to see how many sites exist, how much content they hold, and if they are active or not.
It is probably a good idea for the records manager to be included as a Site Collection Administrator, including by being a member of a Security Group added to every SharePoint site. This will help the records manager gain visibility of the content of each site, however they should be very careful about browsing the content as everything is recorded in audit logs.
Document and plan
The outcome of all these actions should be one or more documents that describe (a) where records are stored and (b) the retention policy and action that will apply to those records.
For Exchange mailboxes, OneDrive for Business accounts, and MS Teams, this may be a single line for each policy.
For SharePoint, there should be a listing of every site and the retention policy or policies that apply to that site.
Additionally, for SharePoint sites where an explicit label-based retention policy is applied, the listing should show which libraries this has been applied to. If a disposal review option has been selected, there should be a process to ensure that the metadata of the library where the records are stored is exported and stored in a different location. The original library may then be deleted.