Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Marketing Technology › The Marketing Stack

Deciding How Long Marketing Tools Keep Personal Data

How to decide how long marketing tools keep personal data: where copies sit, periods by purpose, suppression exceptions and deletion that reaches every copy.
Guideadvanced10 min readUpdated

Why personal data outlives its purpose

Nobody decides to keep a lead's details forever. It happens because every tool keeps by default and deleting is nobody's job. A trial-pass request becomes a CRM contact, an email subscriber, a row in an ad audience upload and a line in a spreadsheet someone made for a campaign review. The campaign ends; the copies stay.

The GDPR's principles, as the GDPR glossary entry describes them, are the frame for this page. Storage limitation means personal data is kept in identifiable form no longer than its purpose needs. Data minimisation means holding only what is adequate, relevant and limited to that purpose. And the controller must be able to show that it complies. The definition of personal data is broad, and pseudonymised data still counts, so a hashed email address in an upload file is in scope just as the plain address in the CRM is.

Two things make retention a stack question rather than a tool setting. Copies move: integrations push data from the system that collected it into others, each with its own defaults. And purposes differ: the same address can be held to send a newsletter, to exclude existing customers from ads and to recognise someone who opted out. The automation governance guide says retention for contact records, engagement logs and message histories is a policy decision made with legal and privacy colleagues, which the platform then implements. This guide extends that to every tool the data reaches.

Finding every copy

Start with the kinds of personal data the marketing stack holds, not the tools: enquiries and trial requests, subscribers, customers and their order history, engagement records, audience upload files, event and behaviour data. For each kind, follow it from where it is collected to every place it lands. The integration register is the map for this: each row is a route by which a copy is made. Where no register exists, the same search of each tool's integration settings and scheduled exports finds the routes. Once found, those routes are worth recording as rows of an integration register, so the next retention review starts from a list instead of a search.

Then look for copies no integration explains. Exports saved to shared drives, spreadsheets built for one analysis, files attached to tickets, upload files kept after the upload, and warehouse tables created for a project that ended. Ask the people who run reports where their data comes from and where it goes afterwards. The guide to keeping personal data out of analytics covers a special case: personal data that reached the analytics property by accident, through URLs, page titles or form fields, which has no purpose there at all.

Where copies sitHow they usually get thereWhat to record
CRMForms, imports, syncs from the shop and the email platformWhich kinds of data, from which sources
Email and messaging platformsSign-ups, CRM syncs, importsSubscribers, engagement logs, message history
Ad platformsCustomer list uploadsWhich lists, how often, from which system
AnalyticsTags and, by mistake, URLs and form fieldsUser-level data and any personal data that leaked in
WarehousePipelines from every sourceTables holding personal data and their purpose
Spreadsheets and shared drivesManual exports and scheduled reportsWho made them, why, and whether they are still needed
Vendors' own systemsEvery tool that processes data for the businessWhat the contract says about retention and deletion

Backups the business takes on purpose belong on this list as well; the guide to backing up marketing tool data and settings gives each set of copies an owner, an expiry and a tested restore.

Setting periods by purpose

A retention period attaches to a purpose, not to a tool. Newsletter subscribers who stop engaging, enquiries that never became customers, customers' order history used for marketing, engagement logs, audience upload files: each is a purpose with a reason for keeping the data and a point after which that reason no longer holds. List the purposes, with the copies found for each, and take the list to whoever owns data protection in the business. They set the period for each purpose, taking legal advice where needed, and record the decision and its reasoning. This guide does not suggest periods, because none would hold across businesses, purposes and jurisdictions.

Write down, for each purpose, what starts the clock (the last purchase, the last engagement, the end of a campaign), what happens when the period ends (deletion, or reduction to a minimal record) and who owns the rule. Some records exist for reasons outside marketing, such as finance or service, and are held in other systems under other rules; the marketing copy of the same person does not inherit those reasons. The data operations and taxonomies guide lists archive and deletion on the retention schedule as a routine owned by the data owner with legal input; this step writes the schedule that routine follows.

Suppression and other records you keep on purpose

Some records are kept deliberately after the rest of a person's data goes, and the most important is suppression. A suppression list exists so that people who unsubscribed or objected are recognised and blocked if their details arrive again from an import, an event list or a sync. Deleting the suppression entry with everything else makes them reachable again. The ICO's direct marketing guidance, as the suppression list entry explains, is the reason opt-outs are kept on a suppression list rather than deleted.

Keep each suppression entry with only what is needed to recognise the person and honour the choice: the identifier, the reason and the date. That follows minimisation while keeping the protection. Consent records are the other kind of record often kept on purpose: the source, date and wording behind a permission someone gave, which the guide to list building and consent asks every list to hold, so the business can show what people agreed to. Record each kept-on-purpose category in the retention schedule with its reason, so that nobody mistakes it for data that escaped the rules. The academy checkpoint on consent maps, retention and privacy operations tests which records are kept on purpose after deletion, beside retention by purpose and deletion that reaches every copy.

Putting the rules into each tool

A schedule nobody implements is a document. For each tool on the list of copies, decide how the rule will actually run: a setting in the tool where it offers one, a scheduled workflow that deletes or reduces records, a pipeline job in the warehouse, or a recurring manual task with an owner and a date. This guide makes no claim about what any particular tool offers; check each tool's documentation and record what it says.

Analytics is one place where the vendor documents limits. Google documents retention limits for user-level data in GA4, as the GA4 glossary entry states, and the GA4 setup checklist includes choosing the retention option as a decision rather than a default. Check the help centre for the current values rather than relying on a remembered figure.

Vendors keep data too. Retention and deletion on termination are contract terms, as the guide to AI disclosure and data privacy says for AI vendors and the choosing and evaluating tools guide says for every tool; if a vendor's documented deletion period is longer than the business's policy, the policy is in practice the vendor's. New tools should arrive with their retention rule decided: the marketing tool onboarding SOP asks the data and privacy review to record how long the tool will keep the data it receives.

Deletion that reaches every copy

Deletion has two triggers: a retention period ending and a person asking. Both have to reach every copy. A rule that deletes old contacts in the email platform but not in the CRM is undone by the next sync, which sends them back. Work through the list of copies for each purpose and make sure the deletion runs in each, in an order that stops a sync from restoring what was just removed: switch off or filter the route, then delete downstream, then upstream, or delete in the system of record and let the deletion travel along the routes the register shows.

Objections need the same reach. The GDPR's right to object to direct marketing is absolute, as the GDPR entry states, and an objection should stop marketing use of the person's data in every tool, with the suppression entry kept so it stays stopped. The consent map records how an objection travels between tools; the retention schedule records what happens to the rest of the data. A single request to stop marketing is handled by the marketing opt-out request SOP, which keeps the suppression entry and passes any deletion request to the data protection owner. The academy exercise on setting retention and deletion across the stack for a wine club practises finding every copy of member data, setting periods by purpose with the suppression exceptions and designing deletion that reaches each copy.

Deletion cannot be undone, which makes it a one-way door in the sense of the one-way and two-way doors framework. Try each new rule on a small sample first, check that it removed what it should and nothing else, and check that suppression entries and consent records survived. Only then run it on everything. When a whole tool leaves the stack, the tutorial on retiring a marketing tool handles the exports, the deletion request to the vendor and its written confirmation.

Showing it was done

The controller must be able to show compliance, so retention leaves a record. Keep the schedule, with each purpose's period, owner and the date it was decided. Keep a log of each deletion run: the rule, the tool, the date and the number of records affected, without the personal data itself. Keep vendors' written confirmations of deletion with the contracts. And review the schedule at each stack review, with the list of copies updated from the integration register, because every new integration is a new copy.

A gym chain sets retention across its tools (illustrative)

Illustrative example — not a real result

Yarvelune Fitness sets one rule per purpose

Yarvelune Fitness, an invented gym chain with 7 sites, starts with one kind of data: the details people give when they ask for a free trial pass. Following them through the stack, the marketing manager finds 5 copies: the CRM, the email platform, monthly ad audience uploads, a shared spreadsheet the sites use to chase trial visitors, and the warehouse. The spreadsheet holds 4,200 rows going back 3 years, and nobody can say who still uses it.

She takes the list of purposes to the chain's data protection lead, who sets one period for each: trial follow-up, the newsletter for those who opted in, and ad exclusions for current members. The lead records the reasoning for each decision. Everyone who unsubscribed or objected keeps a suppression entry holding the email address, the reason and the date, and nothing else.

The rules go in one tool at a time. The spreadsheet is retired and its follow-up job moved into the CRM. The CRM runs a scheduled workflow that deletes trial contacts past their period unless they joined or subscribed, and the deletion syncs to the email platform. Upload files are built from the CRM, so the next upload carries the change, and the warehouse pipeline drops the same records on its next run. The first run is tested on 50 records: 48 are removed everywhere, and 2 that had joined as members are rightly kept. The full run follows, and the log records the date and the count.

Common mistakes

  • Setting retention per tool instead of per purpose. The same person sits in many tools for different reasons; each purpose needs its own rule, applied in every tool.
  • Forgetting the copies no integration explains. Spreadsheets, saved exports and old upload files hold personal data long after the campaign that made them.
  • Deleting suppression entries with everything else. People who opted out become reachable again the next time their details arrive.
  • Deleting downstream while the sync still runs. The next sync restores what was removed; stop or filter the route first.
  • Running a new deletion rule on everything at once. Deletion is irreversible; test on a sample and check what survived.
  • Picking a period from somewhere else. Periods are set for this business's purposes by its data protection owner, not copied from another company or a vendor default.
  • No record of what was done. Without the schedule, the run log and vendor confirmations, the business cannot show that its rules were followed.

Frequently asked questions

How long should we keep marketing contacts?

This page does not give a period, and nor should a general guide. The GDPR's storage limitation principle says personal data is kept in identifiable form no longer than the purpose needs, which means the answer depends on the purpose, the business and the jurisdiction. The period for each purpose is set and recorded by whoever owns data protection in the business, with legal advice where needed. What this guide covers is making whatever they decide actually happen in every tool and every copy.

If we delete people, how do we stop mailing them again?

By keeping a suppression entry. A suppression list exists so that people who opted out or objected are recognised and blocked if their details arrive again from an import or a sync, which deletion alone cannot do. The entry is kept with only what is needed to recognise the person and the reason, and it is the one record that is kept on purpose when the rest of the person's data is deleted.

Is the vendor's own retention our problem?

Yes. A tool that holds personal data on the business's behalf keeps it under the vendor's terms, so retention and deletion on termination are contract terms to agree at signing and to check at renewal. The AI disclosure and privacy guide makes the point for AI vendors: if the vendor's documented deletion period is longer than the business's policy, the policy is in practice the vendor's. The same holds for every marketing tool.

Terms used on this page

Frameworks

Where this page is used

Part of these learning paths

Sources

  1. Regulation (EU) 2016/679 (GDPR) — EUR-Lex
  2. Direct marketing guidance — Information Commissioner's Office (UK)

Related