Get your Attribution Consulting tracking to 95%+ accuracy guaranteed in writing.
This is the meeting where the media buyer trusts Meta, finance trusts the order table, and analytics trusts GA4, and nothing gets decided. You're not looking at a bug — you're looking at systems answering different questions, with some genuine signal loss mixed in. Our job is to tell you which part of the gap is definitional and permanent, which part is broken and fixable, and which single number your budget decisions hang on.
Free30 minutesNo obligation
Attribution Consulting · what breaks
Why your platforms will never agree on their own — Attribution Consulting
- 01
Every platform counts on a different clock
Meta reports a conversion back on the impression date and restates history as it learns. Google Ads books it on the click date. GA4 books it on the conversion date. Before anyone deduplicates anything, the same week is already three different weeks.
- 02
View-through is counted by some and ignored by others
Meta's default includes a 1-day view, YouTube counts engaged views, GA4 counts none at all. Whether an impression can earn credit is a policy decision, and if you haven't made it, three vendors have each made it for you.
- 03
Credit is claimed, not allocated
Meta and Google each report the sale they believe they influenced, in full. Add platform-reported revenue together and you'll exceed the money that reached the bank. That isn't fraud — it's two ledgers with no shared user identifier between them.
- 04
Real data loss hides inside the disagreement
Part of the gap is definitional and will never close. Part of it is a purchase event that stopped firing on one payment method or a match key that was never sent. Most teams spend a year arguing about the first and never go looking for the second.
Attribution Consulting · the fix
What changes when one number runs the budget — Attribution Consulting
Three numbers with three defined jobs
One for allocating budget, one for feeding in-platform optimization, one for diagnosing what happened. Nobody argues about which dashboard is right, because each one is finally being asked the question it can answer.
The gap gets split into two piles
A written reconciliation showing how much of the variance is window and view-through policy, and how much is signal we can actually go recover. The second pile is the one worth spending money on.
MER is anchored to margin, not to a benchmark
Your blended spend-to-revenue target is derived from contribution margin and payback period, so it reconciles to the P&L instead of to a number someone quoted on a podcast.
Incrementality tests replace opinions
A geo holdout or platform lift design you can genuinely run, with the sample, duration and read-out rules agreed before the money goes in rather than argued about afterward.
Attribution Consulting · architecture
The decision model: three numbers, three separate jobs — Attribution Consulting
This engagement doesn't build a container. What it installs is a hierarchy that tells each number which question it's allowed to answer.
- 1
Tier 1: the ledger that settles disputes
Orders in your store, charges in Stripe or closed-won in the CRM — the one system that reconciles to the bank and outranks every dashboard.
- 2
Tier 2: the metric budget moves on
A single blended number, usually MER read against contribution margin, defined once with its window, timezone and currency written down.
- 3
Tier 3: platform numbers, for bidding only
Meta, Google Ads, TikTok and Microsoft figures are treated as optimization fuel and directional read, never summed into a revenue total.
- 4
The variance ledger between the tiers
Every gap is written down and labeled either definitional attribution difference or recoverable signal loss, and only the second kind gets a budget.
- 5
The override rule, agreed in advance
A written condition for when a platform-reported number is overruled by the ledger, and the named person who is allowed to make that call.
Attribution Consulting · events
What actually gets measured, and where each number comes from — Attribution Consulting
Advisory work produces a measurement inventory rather than a tag spec, so this is the list and its provenance.
| What we measure | Definition we hold it to | Where the number comes from | Decision it drives |
|---|---|---|---|
| Net revenue | Paid orders minus refunds, on order date, one timezone | Store order export or Stripe charges | The denominator every other number reconciles to |
| Blended MER | Total paid media spend divided by net revenue, same period | Platform spend exports plus the ledger | Whether total spend goes up or down |
| Platform-reported ROAS | Value each platform claims under its own window and view policy | Meta Ads Manager, Google Ads, TikTok, Microsoft | In-platform bidding and creative calls, not budget splits |
| Window and view policy | Click and view windows chosen per platform, applied everywhere | Ad account settings, recorded in the charter | How much variance is expected before anyone investigates |
| New vs returning revenue | First-time buyer flag taken from the customer record | Store or CRM customer object | Whether acquisition is growing or repeat is carrying it |
| Signal delivery rate | Share of backend orders that reached each platform with match keys | Server container logs against the order table | Whether the gap is policy or genuine data loss |
| Meta Event Match Quality | Match key coverage score per event, trended monthly | Meta Events Manager | Whether match keys need rebuilding before modeling is trusted |
| Incrementality read | Geo holdout or lift result against pre-agreed decision rules | Test design plus the ledger | Whether a channel's claimed credit survives being switched off |
Attribution Consulting · prove
How we check the reconciliation actually holds up — Attribution Consulting
A reconciliation you can't reproduce is an opinion with a spreadsheet attached. Everything here is designed so your own team can re-run it without us.
Source of truth
Your order table or CRM closed-won — order date, one timezone, one currency
- Export the same 30-day period from Meta, Google Ads, TikTok, GA4 and the order table, normalized to one timezone and currency.
- Restate Meta on conversion date rather than impression date before it is compared with anything else.
- Model each definitional difference — click window, view-through, reporting date — and record how much of the gap it closes.
- Trace the unexplained remainder: sample twenty real orders and check whether each one reached each platform at all.
- Confirm someone outside the project can reproduce every figure in the charter from the exports alone.
- Re-run the whole reconciliation a month later; a moved explained-share means policy changed or something broke.
Attribution Consulting · free audit
What we check in the free audit
- Every platform counts on a different clock
- View-through is counted by some and ignored by others
- Credit is claimed, not allocated
- Real data loss hides inside the disagreement
Checked against
Your order table or CRM closed-won — order date, one timezone, one currency
Free30 minutesNo obligation
Attribution Consulting · when not to buy
When an attribution engagement is the wrong spend
This is advisory work, and advisory work is easy to buy too early.
- One paid channel, one product, same-day purchase. That's a tracking problem, not an attribution problem — fix the events instead.
- Your purchase event is visibly broken. Rebuild the signal first; reconciling numbers you know are wrong just confirms it expensively.
- Nobody in the room can actually move budget on the answer. Without that authority the charter becomes a document everyone ignores.
- Spend is small enough that misallocating it costs less than the engagement. Blended MER in a spreadsheet is genuinely enough there.
Attribution Consulting · what we deliver
What an attribution engagement produces
Cross-platform reconciliation model
Meta, Google Ads, TikTok, Microsoft, GA4 and your order table placed side by side for the same period, same timezone, same currency, with each definitional difference named and quantified.
Attribution window and view-through policy
A per-platform decision on click and view windows, written down once and then applied consistently inside the ad accounts and in every report that leaves the team.
The measurement charter
Two pages naming the source of truth, the decision metric, who owns each number, and what has to be true before anyone moves budget on the basis of it.
Signal-loss audit
Browser, server, consent and CRM signals checked against backend orders using the Source-of-Truth Reconciliation Method, so recoverable loss is separated from the gap that no implementation will ever close.
MER and channel target model
A contribution-margin-based blended target with per-channel guardrails, plus the explicit condition under which a platform-reported ROAS number gets overridden.
Incrementality test plan
Geo holdout or conversion-lift designs, sized and sequenced by which channel your reconciliation says is most likely overstating itself, with the decision rules written before the test runs.
Written guarantee
95%+ tracking accuracy
Measured against Your order table or CRM closed-won — order date, one timezone, one currency. Miss it and we keep working at no cost.
FAQ
Attribution Consulting questions, answered
No, and anyone who says otherwise is selling you something. Meta, Google and GA4 are measuring different things over different windows with different rules about views. The realistic goal is explained variance: you know why each number differs, by roughly how much, and which one to act on. Identical numbers would actually be evidence something is wrong.
No, and this distinction matters more than almost anything else here. The orders still happened — they're in your store, your bank and your ledger. What you lost is the platform's ability to learn from them and your ability to allocate budget correctly. The cost is real, but it's misallocation and worse optimization, not missing sales.
MER is the only number that reconciles to the bank, so it belongs in the budget conversation. But it can't tell you which channel to cut, and it moves for reasons that have nothing to do with media. Platform numbers overlap and overstate, but they're what the bidding algorithms actually consume. Use both, for the two different jobs they're each good at.
Usually not yet. MMM wants years of history and real variation in spend across channels to say anything trustworthy, and it costs accordingly. For most mid-market advertisers a reconciliation model plus two or three geo holdout tests answers more of the actual questions for far less. We'll tell you honestly if you're at the scale where MMM starts earning its keep.
Meta's modeling, Google's Consent Mode modeling and GA4's behavioral modeling are all estimates filling gaps left by consent and platform restrictions. They're reasonable in aggregate and directionally useful. What you shouldn't do is reconcile a modeled number to your order table row by row, because there are no rows underneath it.
Either. The deliverables are deliberately vendor-neutral so your in-house team or your existing agency can execute them. If you'd rather we rebuild the signal layer ourselves, that's a separate scope with its own price, and you're under no obligation to take it.
If you run one paid channel, sell one product, and people buy the same day they click, you don't have an attribution problem — you have a tracking problem, and it's cheaper to fix the events than to hire someone to think about windows. This engagement earns its money when there are multiple channels, a consideration period, and a real argument happening in your budget meeting.
More on Attribution Consulting
- GuideWhy Every Platform Shows a Different ROAS (And Which Number to Trust)
- GTM consultingA tag manager container someone can actually maintain
- Call trackingAttribute the calls that turn into your best customers
- Measurement StrategyMost tracking rebuilds get quoted before anyone has decided what the conversion …
Free · 30 min
Bring us the four numbers that don't match
WhatsAppFree30 minutesNo obligation
