Before the Request Arrives: Rehearsing Subject Access Retrieval Across a UK Cloud Estate
Photo: UK data protection officer reviewing cloud data access logs in modern office, via thinkerten.com
A subject access request is often treated as a legal process. In practice, it is an operational one. UK GDPR gives individuals the right to obtain a copy of their personal data, and the default response period is one month. The Information Commissioner's Office recognises that complex requests may need an extension, but the clock still starts on receipt. The uncomfortable question for many UK organisations is not whether they have a policy. It is whether the cloud estate can actually find the data.
Most estates now span Microsoft 365, several SaaS applications, line-of-business systems, unstructured file storage and backups. A request may arrive from an employee, a customer, a former supplier contact or a complainant. The data may sit in a shared mailbox, a Teams channel, a CRM record, a support ticket, a payroll export and a backup set. A rehearsal is the only reliable way to know whether those systems can be searched, exported and delivered within the statutory window.
Why a rehearsal beats an assumption
A policy describes the duty. A data map describes the intended estate. Neither proves that a named person can retrieve a specific subject's data from each service under time pressure. Tabletop exercises can reveal obvious gaps, but they rarely expose the delays that matter: a vendor ticket queue, an admin account held by a departed contractor, an export tool that cannot search chat history, or a backup restore that takes longer than the response period allows.
The rehearsal described here is deliberately modest. It uses a controlled test subject, runs for half a day and produces an issue log rather than a polished assurance statement. Its value lies in repetition. Once the organisation has a baseline, it can re-run the exercise after major migrations, new SaaS adoptions or changes in retention settings.
Choose a test subject that crosses boundaries
The first decision is who the rehearsal is about. A synthetic test identity is usually preferable. Seed it across the systems in scope with clearly marked test data, then attempt to retrieve that data as if a real request had arrived. This avoids processing a real person's information unnecessarily and makes it easier to judge whether the search is complete.
Where a synthetic identity is impractical, a consenting employee can act as the subject. Consent should be informed, documented and withdrawable. The exercise should be explained as a systems test, not a live subject access request, and the data used should be minimised. The organisation's data protection officer or legal adviser should agree the approach before any real personal data is processed.
The test subject should have a footprint in more than one service. A person who appears only in email will not test the estate. A person who appears in email, a SaaS CRM, a support platform, a shared file area and a backup set will reveal far more about retrieval capability.
Map the estate before searching
Before the clock starts, list the systems in scope. For each one, record the data types held, the search method, the export method, the system owner and the retention period. The list should include Microsoft 365 workloads such as Exchange, SharePoint, OneDrive and Teams, as well as any SaaS applications that hold customer, employee or contact data. Unstructured storage, archives, endpoints and backups belong on the list too, even if the initial conclusion is that retrieval is not proportionate or technically feasible.
The mapping stage often exposes the first problem: nobody is sure who owns a particular application. That is a useful finding. A system without a named owner is a system without a retrieval route. The rehearsal should record the time taken to identify the owner as well as the time taken to search the data.
Time the exports, not just the searches
A search that returns results is not the same as an export that can be delivered. The rehearsal should time each stage separately: locating the data, reviewing it for relevance and third-party information, exporting it in a usable format, applying any necessary redactions and preparing the response package.
Record the time in working days, not just hours. A native Microsoft 365 search may complete quickly, but a vendor-assisted export from a specialist SaaS platform may require a support ticket and several days of waiting. If a single service takes ten working days to return data, that is a material risk to the one-month response period. The issue log should capture the delay, the cause and the action required.
Delivery should also be tested. Secure links, encryption, recipient verification and file-size limits can all create friction. The rehearsal should confirm that the export can be delivered in a way that satisfies the organisation's security requirements without adding unnecessary delay.
Treat backups as part of the question
Backups are frequently omitted from subject access planning because they are difficult to search. That omission should be a conscious, documented decision rather than an untested assumption. The rehearsal should establish whether personal data in backups can be located, whether a restore to a staging environment is required and how long that process takes.
The outcome may be that backup retrieval is not proportionate in ordinary circumstances. That conclusion still needs an owner, a recorded rationale and a review date. The rehearsal provides evidence for that decision rather than leaving it to be discovered during a live request.
Record where ownership or tooling fails
Every rehearsal should produce a short issue log. For each failure, record the service, the failure mode, the impact on retrieval time, the action required, the owner and a due date. Distinguish between tooling failures, such as an export that cannot include chat messages, and process failures, such as an approval that nobody knew was required.
The log should be reviewed by the data protection lead and the IT owner together. Some issues will be fixed quickly. Others may require procurement action, retention changes or new licensing. The point is not to solve everything in one half-day. It is to replace assumption with a prioritised list of operational gaps.
Make the rehearsal repeatable
A single exercise is useful. A repeatable one is more valuable. Keep a template covering the test subject, systems in scope, timings, issues and sign-off. Re-run the rehearsal after significant cloud migrations, new SaaS deployments or changes to Microsoft 365 retention and eDiscovery settings. Compare the timings with the previous run.
The results can also inform wider cloud strategy. If a business unit has adopted a SaaS tool without a retrieval route, that is a procurement and ownership issue as much as a data protection one. If shared mailboxes or Teams channels are consistently difficult to search, the organisation may need to adjust its information management approach before the next request arrives.
The legal boundary
The rehearsal tests operational retrieval. It does not replace legal advice. The data protection officer or legal adviser should define the scope of any real request, decide whether an extension is justified and determine what exemptions or redactions apply. The rehearsal should not create a parallel subject access process or encourage staff to bypass the formal route.
What it can do is give the organisation something it currently lacks: evidence. Evidence that a named person can search a named system. Evidence that an export can be produced within a stated time. Evidence that a backup decision has been considered rather than ignored. Before the next request arrives, that evidence is worth more than another policy review.