Definition
An integration register is a single document, kept for as long as the stack exists, with a row for each route by which data passes from one marketing tool to another. The method does not decide whether a route counts: a native connector, middleware, a file sent on a timer and a colleague who exports a list and uploads it by hand each get a row. The tutorial on documenting marketing integrations builds the register and then draws the stack's data flow diagram from its rows, so the picture is a view of the record.
It is standing: it covers every connection, not only those one project touched, and a connection that is switched off stays on it as retired, dated. And each row carries enough detail for a colleague who never saw the connection set up to judge whether it still runs and what depends on it.
What each row records
- What moves and why. A short name, the job the connection does, and the objects and fields that cross.
- Which way and by what method. The source and destination tools, the direction and the mechanism.
- What it matches on. The identifier the destination uses to recognise a record, such as an email address or a customer id, and any key used to remove duplicates.
- When, and with what effect. The schedule, and whether the destination adds what arrives to what it holds or replaces it, as that tool's own documentation describes.
- What is held back. The consent filter, the field it reads and who decided it.
- Whose access it runs on. The login, service account or key, and who in the business holds it.
- Who answers for it. One accountable owner, named with a role so the row survives a reorganisation.
- How a failure shows. The check that would catch a stopped run, who sees the alert, and the last change.
A gap in a row is a finding in its own right. The tutorial marks connections with no owner or no failure check, and moves any that run on an agency's or a former employee's access onto credentials the business controls, since such a connection stops on the day that access is withdrawn.
In practice: change control, retention, consent, sync changes and outages
- Change control. A connection only counts as live once its row is filled in and its owner has agreed it; a changed connection updates its row, and a retired one keeps it.
- Retention. The guide to retention across marketing tools reads every row as one path by which personal data is copied into another tool, so the search for copies starts from the register, and each new connection adds a copy for the retention schedule to cover.
- Consent. The consent mapping tutorial uses the register to confirm that every tool receiving personal data appears somewhere on its map, and reads the consent filter column to learn what each connection leaves out.
- Sync changes. The CRM sync change checklist starts by completing the connection's row, owner included, before a connector is switched on or altered, and ends by writing the failure check and the alert's recipient into the same row.
- Outages. The guide to preparing for a marketing tool outage treats any connection into or out of the tool that is down as a job stopped with it, and checks whose credentials each connection uses before blaming the supplier.
A tile retailer finds whose login a sync ran on
Hourvain Tiles, an invented tile retailer, noticed in March that its welcome emails had gone quiet: none of the 214 first-time customers from the previous 9 days had reached the email platform. Both the shop and the email platform reported themselves healthy. The ecommerce manager opened the integration register the team had started the year before and found the row for the customer sync: a native connector, run hourly, matched on email address and authenticated with the login of a merchandiser who had left at the end of February. Her account had been closed on her last day. That afternoon the connector was reconnected under a service account the business controls, the 214 customers were sent across in one batch after a check against opt-outs, and the row gained a volume check that alerts its owner if a day passes with no new customers arriving.
Related terms
Change ControlThe routine a proposed change to a live system follows: written down first, reviewed, tested, signed off and recorded before it reaches customers or data. keeps the register true, because each new, changed or retired connection passes through it with its row. The Stack RecordThe live register of every marketing tool: owner, deputy, cost, notice date, users, use cases and status, kept current by onboarding, reviews and retirement. lists the tools themselves; the register lists what passes between them, and one tool can sit in many of its rows. A 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. is organised by purpose rather than by connection, and it borrows each row's consent filter to follow a withdrawal between tools. The identifier column does for a connection what a Join KeyThe shared field, such as date and channel, a campaign ID or a click ID, on which rows from two data sources are matched in a blend or a query. does for two sources in a report: it names the field a record on one side needs to find its partner on the other, so recording it lets a mismatch be seen before it creates duplicates. The Source of TruthThe system whose figure is treated as authoritative for a given number, such as billing for revenue or the CRM for pipeline, with other tools reconciled to it., also called the system of record, is one place the retention guide lets a deletion start, which then reaches the other copies along the routes the register records.
Common confusions
- A data flow diagram does the same job. An arrow between two boxes says the tools are linked. It does not say what they match on, whose login carries the link or how a failure would be seen.
- It is a tag inventory. A tag inventory lists the tags a site loads, each with its consent category. The register covers data moving between tools.
