Marketing Attribution Platform Governance: Who Owns the Numbers and How Often to Review Them
Good governance of a marketing attribution platform comes down to three things: one named owner for the numbers, a fixed review rhythm, and a written log of every change to models and windows. Most teams skip all three. The platform gets set up once, people drift in and out of it, and six months later nobody can say why last-click was switched on or why the dashboard no longer matches the one in the board deck. This post is the operating manual for after the tool is live: roles, cadence, change control, and what to do when two sources disagree.
Who should own the numbers in a marketing attribution platform?
One person should own them, and that person should be accountable for definitions and settings, not just for reading the dashboard. Split the work into three roles:
Controls the attribution model, lookback window, conversion definitions, UTM conventions, and integrations. Signs off on every change and keeps the decision log. At a brand this is typically the head of growth or performance lead. At an agency, name one owner per client account.
The media buyer or analyst who checks spend against attributed revenue and flags anomalies. They read and annotate. They do not edit settings.
The founder, CFO, or client lead who looks at a short set of numbers monthly. They need stable definitions more than detail, and should hear about any definition change before the report, not after.
Agencies should add a standards owner who sets the defaults every client account starts from. This week: write the owner's name at the top of the decision log and tell the other two roles.
How often should you review attribution data?
Review on four clocks: daily for tracking health, weekly for campaign decisions, monthly for channel mix, and quarterly for the model itself. Each clock has a different question, and mixing them is how teams end up rewriting their model after one bad Tuesday.
Does attributed order count look sane next to store orders? Did spend or revenue by channel jump? You are checking that tracking is alive. AdBeacon syncs about every 10 to 15 minutes, with a manual refresh, so this check sees near real-time data. Do not change settings.
Campaign and ad set performance, platform-reported versus first-party numbers, and any gap that moved. This is where budget shifts get decided.
Channel mix, MER trend, first-click versus last-click comparison, and a read-through of the decision log. Prepare the executive summary here.
Is the model and window still right for how customers buy? Audit permissions. Decide whether it is time for an incrementality test, which checks attribution against a different kind of evidence.
Put all four on the calendar as recurring invites. A cadence that lives in someone's head does not survive a vacation.
What belongs in an attribution decision log?
A decision log is a running record of every change to attribution settings, with the reason and the expected effect. It is the single most useful governance artifact, and it can live in a shared spreadsheet. Each entry needs:
- Date and who made the change
- What changed (model, window, conversion definition, integration, UTM rule)
- Why, in one or two sentences
- A snapshot of the key numbers before the change
- The date you will check the result
A hypothetical entry: "Nov 3: Moved weekly review view from last-click to first-click for prospecting campaigns. Reason: average days to purchase lengthened after the holiday offer launched. Before: Meta prospecting 0.8x click-based ROAS. Check on Nov 24." Without the log, nobody can tell whether customers changed or the settings did.
| Date | What changed | Why | Before | Check on |
|---|---|---|---|---|
| Nov 3 | Weekly review view moved from last-click to first-click for prospecting campaigns | Average days to purchase lengthened after the holiday offer launched | Meta prospecting 0.8x click-based ROAS | Nov 24 |
| Next change | One setting at a time | One or two sentences | Key numbers snapshot | Review date |
Add a "Who" column with the named account that made the change. Hypothetical entry.
What should trigger a change to your attribution model or window?
A change should be triggered by something structural, never by one disappointing week. Legitimate triggers:
- A platform changes its definitions. Meta narrowed click-through attribution to link clicks and added engage-through for other interactions in its March 2026 update, which shifted reported numbers without any change in customer behavior. Google has similarly stopped supporting first click, linear, time decay, and position-based models in Google Ads. Our own breakdown of Meta's click definition change covers what moved.
- Your purchase cycle shifts. If days from first click to purchase stay longer or shorter for a month or more, the window should follow. Google Analytics 4 lets you set key event lookback windows of 30, 60, or 90 days for most events, so know what yours is set to. Google Ads has its own conversion windows, which default to 30 days for click-through on Search and Display.
- A new channel or a large spend change. A new channel needs its own tracking checks before you trust its numbers.
- A tracking break. Fix the break and annotate the gap rather than modeling around it.
Not triggers: a platform rep's recommendation, a stakeholder wanting a bigger number, or a single week of noise. When a real trigger arrives, change one setting at a time, log it, and keep the previous view available for comparison.
What do you do when platform-reported and first-party numbers disagree?
Expect them to disagree, then work out why before deciding which one to act on. The usual cause is a difference in definitions, not a tracking failure. Ad platforms count view-through and engage-through conversions. A click-only first-party view does not.
Run this sequence each week:
- Match the attribution window and conversion event on both sides.
- Reconcile each source's total orders and revenue to the store.
- Track the ratio between the two numbers over time. A stable gap is a definition difference. A sudden jump is something broken.
- Write down which number governs which decision.
Counts view-through and engage-through conversions. Use it inside Meta or Google to steer delivery.
Every credited order has a click behind it. Use it to decide where the next dollar goes.
That last step is the governance part. A workable rule: platform numbers guide in-platform optimization, and first-party click-based numbers guide budget allocation across channels. AdBeacon's Optimize report puts AdBeacon revenue and platform-reported revenue side by side, which makes the gap easy to track. Neither figure is a typo. They count different things, and the team needs a rule for which one sets the budget. AdBeacon's attribution is click-only and first-party, so every order it credits has a click behind it. For more on why the gap exists, see view versus click attribution.
How do you keep permissions clean in an attribution platform?
Treat access like a quarterly audit item: every user should map to a role, and every role should have the least access it needs. Attribution settings change the numbers everyone else reads, so edit rights deserve more care than view rights.
- Limit edit access to the owner and one backup. Daily users and executives get view access.
- Use named accounts only. Shared logins make the decision log meaningless. Agencies pay a flat rate with unlimited seats on AdBeacon, so seat cost is no reason to share a login.
- Add offboarding to the checklist: when someone leaves, a contract ends, or a freelancer wraps up, remove access the same day.
- Review connected accounts (ad platforms, store, email) quarterly and remove any that are no longer in use.
Agencies carry extra risk because contractors and former client teams accumulate. Make the audit a standing quarterly task per client, and log each change.
Attribution governance is not glamorous, but it is what keeps a dashboard credible after the person who built it moves on. Name the owner, put the four reviews on the calendar, and start the log with today's settings. If you want to see how first click and last click views by channel can sit side by side on the same data while you work through this, book a live AdBeacon demo.
FAQ
How often should a marketing attribution platform's model be reviewed?
Quarterly is a sensible default for the model and window, with daily and weekly checks for tracking health and campaign decisions. Review sooner if a platform changes its definitions or your average time to purchase shifts for a sustained period.
Who should own attribution data inside a brand?
One accountable person, usually the head of growth or performance marketing, who controls settings and keeps the decision log. Media buyers use the data daily and executives read a monthly summary, but neither should be changing definitions.
Why do platform-reported ROAS and first-party ROAS differ?
Ad platforms include view-through and engage-through conversions, while a click-only first-party view counts conversions that follow a click. Align windows, reconcile orders to the store, and watch the gap over time rather than expecting it to close.
What is an attribution decision log?
It is a dated record of every change to attribution settings, including the reason, who approved it, and the numbers before the change. It lets anyone separate a real shift in customer behavior from a shift caused by a setting.