Decisions are only as good as the numbers under them
Most marketing data degraded quietly over the last few years. Reports still look stable while the signal underneath them rots, which produces confident decisions in an unknown direction.
- Server-side collection with correct deduplication
- Consent-aware, so the numbers are defensible
- Geo holdout tests that establish causation
- One revenue figure everyone agrees on
Typical outcomes
- Conversions recovered on average
- +34%Conversions recovered on average
- Reported vs actual revenue gap
- -41%Reported vs actual revenue gap
- Typical rebuild to first clean month
- 6 wkTypical rebuild to first clean month
- Of clients keep the implementation
- 100%Of clients keep the implementation
Averages across Digital Squad client accounts. Individual results vary — we show you a realistic range for your situation on the audit call.
Companies we've done analytics & measurement for
Why your dashboard is lying quietly
Broken tracking announces itself. Degraded tracking does not — it keeps reporting, keeps looking plausible, and keeps moving budget in the wrong direction for months.
Browser collection misses a large share of conversions
Ad blockers, cookie lifetime caps and mobile privacy controls now suppress a meaningful proportion of events. Platforms fill the gap with modelled estimates, so the number still appears — it just is not a measurement any more.
Every platform claims the same sale
Sum what Meta, Google and TikTok each report and the total exceeds your actual revenue, often substantially. Budget decisions made on that sum are allocating against a figure that does not exist.
Attribution cannot answer the question you are asking
Every model distributes credit among touchpoints that were observed. None of them can tell you what would have happened without the touchpoint, which is the only thing that matters when deciding whether to spend more.
Nobody trusts the reporting, so decisions get made on instinct
The most expensive symptom. When marketing and finance hold different numbers, the argument stops being about strategy and becomes about whose spreadsheet is right — and the loudest opinion wins.
Measurement built to be argued with
We rebuild collection so the data is trustworthy, then add the one method that establishes causation rather than correlation. The output is a number your CFO will defend in a board meeting.
+34%
Conversions recovered on average
-41%
Reported vs actual revenue gap
6 wk
Typical rebuild to first clean month
100%
Of clients keep the implementation
Everything in the engagement.
No line items appearing on month-three invoices. This is the full scope, written down.
Measurement audit and event map
Every event, where it fires, what it means and whether it correlates with revenue. Most accounts are optimising toward an event chosen because it was easy to implement rather than because it predicts money.
Server-side collection with deduplication
Events sent from your own infrastructure with a shared identifier so browser and server transmission cannot double-count. Running both without that identifier inflates conversions by a large margin, and teams celebrate it.
Consent-aware implementation
Collection that respects consent state by design rather than by a banner bolted on afterwards. This is a legal requirement in several markets and, done properly, recovers more usable signal than the non-compliant version most sites are running.
One revenue number, from your own systems
Blended efficiency calculated from your actual revenue and total spend. It cannot be double-counted or inflated by a platform, which is why it is the only paid metric that has not degraded.
Geo holdout test design and analysis
Switch a channel off in matched regions, run it past the purchase cycle, compare against control. It costs real revenue in the test markets and it is the only method here that answers the causal question.
Reporting a finance team will sign
A single view reconciling marketing-reported performance against booked revenue, with the methodology written down so nobody can quietly change models between quarters.
How this actually runs.
You always know what's happening this week, what's next, and who's doing it.
Audit and reconcile
Week 1–2We map every event and compare summed platform-claimed revenue against what your finance system actually recorded. The size of that gap is usually the moment the project gets prioritised.
Rebuild collection
Week 2–4Server-side transmission, deduplication, consent handling, and re-pointing optimisation at the event that correlates with revenue rather than the one that fires most often.
Validate against reality
Week 4–5A parallel run comparing new collection against your order data, because a tracking implementation nobody reconciled is a hypothesis rather than a measurement.
Design the holdout
Week 5–6Matched test and control regions, a run length set by your purchase cycle, and a pre-registered analysis so the result cannot be reinterpreted afterwards.
Read the result and reallocate
Post-testThe incremental figure usually differs sharply from the attributed one, most often on brand search and remarketing. That difference is the budget decision.
Hand over the documentation
OngoingEvent map, methodology and test protocol written down and yours. If you leave, the implementation does not stop making sense.
Real projects. Real numbers.
Analytics & Measurement engagements where we can show you exactly what moved and why.
Real prices, on the page.
Making you book a call to learn the number wastes your time and ours. Here's what it costs.
Audit
Teams who suspect the numbers are wrong and need to know how wrong.
Scoped to your brief
- Full event and tag audit
- Platform vs actual revenue reconciliation
- Consent and compliance review
- Prioritised remediation plan
- Findings session with your team
- Documentation yours to keep
Rebuild
Companies acting on data they cannot currently defend.
Scoped to your brief
- Everything in Audit
- Server-side collection implementation
- Deduplication and consent handling
- Conversion event re-mapping
- Parallel validation against order data
- Blended efficiency reporting
- 60 days of post-launch support
Measurement partner
Accounts where testing needs to be continuous rather than a project.
Scoped to your brief
- Everything in Rebuild
- Quarterly geo holdout programme
- Ongoing reconciliation and QA
- Incrementality-informed budget guidance
- Board-ready reporting pack
- Named analytics lead
All engagements start with a free audit. Prices exclude media spend, which you pay directly to the platforms on accounts you own.
Analytics & Measurement questions, answered straight.
Including the ones where the honest answer is “you probably don't need us for that.”
Firing is not the same as measuring. The questions that matter are whether events are deduplicated, whether collection survives consent rejection, whether the optimisation event actually correlates with revenue, and whether the total reconciles with your finance system. In most audits we run, at least two of those four are wrong — and none of them produce an error anywhere.
Often paired with
Ready to fix this properly?
Most marketing data degraded quietly over the last few years. Reports still look stable while the signal underneath them rots, which produces confident decisions in an unknown direction.
Our commitment: The audit is fixed-fee and the findings are yours regardless — including if the conclusion is that your measurement is fine and you should spend the money elsewhere.