What this playbook covers and when to run it
The playbook covers marketing's day-to-day use of Personal DataAny information relating to an identified or identifiable living person, such as a name, email address, location data or an online identifier.: where permission or another basis is captured for each purpose, how tags and tools act on it, how withdrawals, objections and suppressions travel, how long data is kept and how deletion reaches every copy, how new tools and new uses are reviewed before real data arrives, and which records show the work was done. It does not decide any of it. The lawful basis for each purpose, retention periodsHow long a kind of personal data is kept for a stated purpose before deletion, set with whoever owns data protection and applied in every tool holding a copy. and how individuals' requests are answered belong to whoever owns data protection in the business, with legal advice where needed. The marketing technology stack hub frames the consent question, and each phase below hands off to the guide that owns its method.
Run it when nobody can say which system holds the consent record for each purpose; when an unsubscribe or objection has been found acting in one tool but not another; when personal data has turned up in analytics reports or a prompt; when retention has been decided on paper but never run; or when new tools, forms and uploads arrive faster than anyone reviews them.
It sits beside two other playbooks rather than repeating them. The marketing stack management playbook owns the life of each tool, including access reviews, onboarding and retirement; this playbook owns what those tools do with personal data. The measurement governance playbook owns the tracking plan and releases; this playbook owns the consent rows those releases have to respect. The lifecycle email programme and the inbound lead handling playbook both depend on its suppression and consent records. Like the rest of the playbooks library, it is a starting shape to adapt.
Roles and responsibilities
- Privacy lead. Whoever owns data protection. Decides the Lawful BasisOne of the six grounds the GDPR accepts for processing personal data, such as consent, contract or legitimate interests, chosen for each purpose. for each purpose, sets retention periods, decides what a new use of data needs before it starts and approves the records. Owns the decisions, not the routine.
- Marketing operations. Runs the routine: keeps the consent map, runs the monthly suppression check and the quarterly retention runs, keeps the records current and calls the privacy lead in when a change needs a decision.
- Analytics owner. Owns the banner's behaviour in the tag container, Consent ModeA Google tag feature that adjusts what tags collect according to the visitor's consent choices., the checks that tags respect each choice, and the sweep for personal data in analytics.
- CRM and email owner. Owns the consent fields, the sync between the CRM and the marketing platform, the Suppression ListA list of addresses or contacts excluded from every send, such as unsubscribes, hard bounces and spam complaints. and the customer lists that leave for ad platforms.
- Tool owners. Answer for how each tool reads consent, applies suppression and runs deletion, and tell marketing operations before anything about that changes.
- Legal adviser. Consulted by the privacy lead on questions the business cannot settle itself, such as a new kind of data, a new jurisdiction or an unusual use.
A RACI matrix is a useful way to write these roles down per phase. The commonest failure is a marketing team deciding the basis itself because the routine needed an answer that day; keep the decision with the privacy lead and record it in the map.
Cadence, inputs and outputs
The rhythm below is a suggested starting point. Adjust it to the number of tools, how often they change, how much personal data they hold and the risks the privacy lead identifies; it is not a legal requirement.
| Cadence | Owner | Inputs | Outputs |
|---|---|---|---|
| Per change | Marketing operations, analytics owner, tool owner | A new or changed tool, tag, form, upload or banner setting | The consent map updated; the banner and tags rechecked; the change recorded |
| Monthly | CRM and email owner, marketing operations | Marked test contacts, the suppression list, the latest syncs and uploads | Confirmation that withdrawals and suppressions reached every tool, or a fix logged |
| Quarterly | Marketing operations, tool owners, analytics owner | The retention schedule, the register of copies, an analytics sweep | Deletion runs logged; personal data found and removed; vendors' confirmations filed |
| Annually | Privacy lead, marketing operations | The consent map, the retention schedule, the year's records | Every row and period confirmed or changed; owners reconfirmed; gaps assigned |
Each cadence feeds the next. Changes recorded per change tell the monthly check which routes are new; the monthly check shows which tools drift; the quarterly runs show whether retention holds in practice; and the yearly review reads all of them to decide what the privacy lead needs to look at again.
The playbook in phases
Map consent and lawful basis by purpose
Start with a consent map: one row for each purpose the stack processes personal data for, recording where the permission or other basis is captured, which system holds the record, which tools read it, how a withdrawal or objection travels, how tags behave, the test for the row and its owner. The tutorial on mapping consent across the stack builds it and notes that the GDPR expects a basis to be chosen for each purpose before processing starts, with the ICO's direct marketing guidance explaining when electronic marketing needs consent; the map records the privacy lead's decision rather than making it. Several later phases test against, update or file the consent mapOne table with a row per purpose the stack uses personal data for: where permission is captured, which system holds it and how a withdrawal reaches each tool., which is why it is drawn first. For email, the guide to list building and consent asks each list to hold the source, date and wording behind every permission. For measurement, the privacy and consent guide sets out what is and is not collected without consent and how measurement is designed around it.
Check the banner and tag behaviour
The banner and the tags are tested against the consent map whenever either changes. The consent banner QA checklist is run before a banner or CMPSoftware that collects, stores and passes on visitors' consent choices for cookies and tracking. goes live and again after any change to its wording, categories, vendors or tag configuration: the visitor gets a real choice, the tags respect it, consent mode carries it to Google's tags and the measurement drop that follows is expected and annotated. The consent mode guide covers the implementation as Google's consent mode page describes it: the page sets a default state before any tag runs, the banner updates it when the visitor chooses, and each tag adjusts its behaviour; its section on verifying consent signals is the test the analytics owner records against each tag row of the map.
Keep personal data out of places it should not be
Personal data turns up where nobody meant to put it. The guide to personal data in analytics covers the routes by which it arrives, a sweep of reports and exports that finds it, fixes at the source with the container as a safety net, and the limits of HashingConverting an identifier such as an email address into a fixed string with a one-way function, so platforms can match records without the raw value.: a hashed email address is still personal data. The tutorial on minimising data before prompting reduces an export or transcript to what a task needs before anything reaches an AI tool. The guide to test environments and test data uses invented test records that cannot be mistaken for customers, so tests do not copy real people into new places. The analytics owner runs the sweep quarterly, as a starting rhythm, and after any form or site change that could put personal details in a URL.
Make withdrawals and suppressions travel
An unsubscribe or objection recorded in one tool has to reach every tool and list that acts on it. The guide to syncing the CRM and the marketing platform makes consent the group of fields that travels in every direction, with the most restrictive value winning, and keeps a seeded test contact that has unsubscribed and objected in both systems. The email list hygiene SOP updates suppression and checks consent records as part of its routine. The guide to customer lists across ad platforms puts a consent and purpose check before any upload, including removing people who asked not to be marketed to. The same check applies to lead outcomes sent back to an ad platform, whether by offline conversion import, enhanced conversions or a server-side conversions API: confirm the purpose, the lawful basis (and consent where it is needed) and the platform's own terms before the data leaves, and remember that hashing an email address or phone number before sending does not stop it being personal data. Each month, check that the marked test contacts are still excluded from the latest send, sync and customer list file; a failure is fixed before the next upload.
Set and run retention
Retention is decided by purpose and run in every tool. The guide to data retention across marketing tools finds every copy, sets a period for each purpose with the privacy lead (it gives no periods, and neither does this playbook), keeps suppression entries and consent records on purpose, puts each rule into each tool and makes deletion reach every copy in an order that stops a sync from restoring what was just removed. It also recommends trying each new rule on a small sample first. Each quarterly run is logged with the rule, the tool, the date and the number of records affected, without the personal data itself, and vendors' written confirmations of deletion are filed with the contracts.
Review new tools and new uses before data arrives
The cheapest time to get privacy right is before real customer data is used. The marketing tool onboarding SOP requires the data protection lead to approve before any personal data reaches a new tool. The guide to disclosure and data privacy in AI sets out which tools may see personal data at all and how vendor terms, retention and deletion are checked for AI uses. For research, the participant information and consent tutorial writes what participants are told and agree to into the study's records. Every one of these reviews ends with the consent map updated, so a new tool or use has its row before it goes live.
Keep the records that show it was done
The GDPR expects a controller to be able to show compliance, as the retention guide notes, so the routine leaves records. The integration register shows what connects to what, which fields travel and who holds the credentials; every new integration is a new route for consent and a new copy to delete. The access review tutorial keeps every account matched to a person, supplier or integration; the review itself belongs to the stack management playbook, and this playbook reads its result. Keep the consent map with each row's last test result, the retention schedule and deletion log, vendors' confirmations and the change records, in one place the privacy lead can read.
Triggers that reopen the consent map
The consent map tutorial lists the changes that require the map to be updated before they go live. The table turns them into triggers with an owner and a check, so the map changes with the stack rather than once a year.
| Trigger | Who raises it | What reopens | Check before it goes live |
|---|---|---|---|
| A new tool, widget or integration | Tool owner, through onboarding | A new row, or new readers on existing rows | Onboarding privacy review; marked test record through the new route |
| A new sign-up form or opt-in wording | CRM and email owner | The capture point and record for that purpose | Source, date and wording recorded; test contact excluded after opting out |
| A change to the banner, its categories or vendors | Analytics owner | The tag rows for every affected category | Consent banner QA checklist; consent signals verified |
| A new customer list upload or segment | CRM and email owner | The advertising row and its suppression route | Consent and purpose check; suppressed test contact absent from the file |
| A new channel, such as text messages | Marketing operations | A new purpose row | Basis decided by the privacy lead; withdrawal route written and tested |
| A new AI or research use of customer data | Requester, through the privacy lead | A new purpose row or a changed one | Data minimised; vendor terms checked; participant information recorded |
| A change of basis by the privacy lead | Privacy lead | Every row that purpose touches | Tools and records updated; test rerun |
| A tool leaving the stack | Tool owner | Rows that tool read or held | Exports with consent and suppression intact; deletion confirmed in writing |
A childrenswear retailer keeps its consent records current
Ezmolde Childrenswear, an invented online childrenswear retailer, runs its first monthly suppression check with a marked test contact who unsubscribed from email. The contact is excluded from the next send and the CRM, but appears in 2 of the 3 customer lists uploaded to ad platforms, because those files are built from an export that ignores the consent fields. The CRM and email owner rebuilds the export to filter at the point the file is made, and the check joins the monthly routine; the following 4 monthly checks pass. In the spring, marketing wants a product review widget that would receive order details. The tool owner raises it through onboarding, the consent map gets a new row, the privacy lead decides the basis and the widget goes live a week after its test passes. In the autumn, marketing operations runs the first retention deletion across 6 tools: it tries the rule on 200 records first, then deletes about 18,000 records whose period has ended, keeps the suppression entries, logs each run without personal data and files written confirmation from each of the 4 vendors that hold copies. At the yearly review, 1 row has no owner and is reassigned.
Measuring the playbook
The routine works when every purpose has an owner and a tested route, and when failures are found by the checks rather than by customers. Operating signals show whether the rhythm is holding: whether each change that should reopen the map did, whether the monthly suppression check runs and records its result, whether each quarterly retention run and personal data sweep is completed and logged, and whether the yearly review confirms every row and period.
Outcome signals show whether consent holds in practice: rows on the map with a passing test and a named owner, suppression failures found by the monthly check rather than by a complaint, personal data found in analytics or prompts and removed at the source, and deletion runs confirmed in writing by every vendor that holds a copy. Report them to the privacy lead in the same pack as the stack and measurement reviews, so privacy work sits beside the changes that create it.
Common mistakes
- Treating the banner as the whole of consent. Tag choices are one kind of permission; marketing permissions, objections and suppressions live in other systems and need their own rows and checks.
- Deciding the basis in marketing. The map records the privacy lead's decision. A routine that needs an answer that day is a reason to call the privacy lead, not to choose one.
- Separate suppression lists in each tool. They drift, and an objection recorded in one tool is ignored by the next. Name one system that holds the record and make the others read from it.
- Building upload files from exports that ignore consent. Filter where the file is built, from the system that holds the record, and check with a marked test contact.
- Deleting suppression entries with everything else. A suppression entry exists so that a person who opted out stays out when their details arrive again. Keep it on purpose.
- Retention on paper only. A schedule that never runs is not retention. Run it, log it and get vendors' confirmations.
- Reviewing new uses after data arrives. A trial, prompt or study that uses real customer data is a real use. Review it first.
