Tracking & Attribution Repair

Your ad platform and your bank account disagree. The gap is your tracking.

In the last DTC account we audited, the pixel recorded roughly 30% of real orders. Every bid, every budget shift, every "winning creative" call was made while the other 70% was invisible. We find that gap, close it server-side, and prove it closed.

Book a Strategy Call  → See the 8-point diagnostic

For Shopify and DTC brands spending €20K+/month on paid media

~30%
Of real orders the pixel actually recorded — audited DTC account
1.34×
True blended MER, while in-platform ROAS looked healthy
10 days
From first reconciliation to corrected measurement
8
Diagnostic points across pixel, CAPI, GA4 and consent
The Gap

Two systems. Two answers. Only one of them pays salaries.

Every brand we audit has this gap. Almost none of them know its size — because nobody reconciles the ad platform against the store, order by order.

Signal
What the platform reported
What the store actually recorded
Purchase events
Browser-side pixel only
~30% of real orders captured
Return on spend
In-platform ROAS reading healthy
1.34× blended MER across the account
Optimisation signal
Assumed full-funnel
Algorithm trained on a third of conversions

Findings from a live account audit of a global wellness brand, conducted by Prosperus. Figures are that account's — not a benchmark. Your gap will be a different size. The point is that it exists and is measurable.

Why It Happens

Nothing broke. The ground moved underneath the setup.

Most tracking stacks were correct the day they were installed. Then browsers, consent law and platform APIs changed around them — and nobody re-reconciled.

The result is a measurement layer that still reports confidently while quietly losing the majority of its signal. Confident and wrong is the expensive combination.
→
Browser privacy strips the client-side pixel ITP, ATT and tracking prevention drop a large share of browser-fired events before they ever reach the platform. Nothing errors. The events simply never arrive.
→
Consent banners fire before the tag does Under EU consent rules, events for non-consenting users are suppressed. Without Consent Mode configured properly, that traffic disappears instead of being modelled.
→
CAPI added without deduplication Server events bolted on without a shared event_id either double-count purchases or get silently discarded. Both corrupt the optimisation signal.
→
Weak match quality on the events that survive Events that do arrive often carry too few identifiers to be matched to a person, so they never make it into attribution at all.
→
GA4 is answering a different question Default models and channel groupings routinely hand credit to branded search for demand that paid social created three days earlier.
The Diagnostic

Eight checks that tell you exactly how blind you are.

This is the audit, in full. No black box — you receive the reconciliation spreadsheet and the findings, whether or not you continue with the rebuild.

1
Order-level reconciliation
Platform-reported purchases matched against real store orders for the same window. This produces the single number the rest of the audit hangs on.
2
Event match quality & parameter coverage
Which identifiers are being sent, which are missing, and how much matchable signal is being left on the table per event.
3
CAPI presence and deduplication integrity
Whether server events exist, whether they share an event_id with the browser event, and whether dedup is actually working.
4
Consent Mode configuration
Whether consent states are transmitted correctly so non-consenting traffic is modelled rather than silently lost.
5
Server-side hit ratio
What proportion of conversions arrive server-side versus browser-only — the durability measure of the whole stack.
6
GA4 model and channel grouping
Attribution model, lookback windows and custom groupings audited against how your customers actually buy.
7
Checkout hand-off continuity
Whether identity and campaign parameters survive the jump from site to checkout, or reset and orphan the session.
8
PII hashing discipline
Confirmation that everything leaving your servers is correctly normalised and hashed — accuracy and compliance in one check.
The Engagement

Ten working days. Reconcile, rebuild, verify.

Your team supplies access. We do the work. Nothing is handed over until the numbers reconcile against a control period.

Days 1–3
Reconcile
We establish the size of the gap before touching anything.
  • Order-level export vs. platform events
  • Full 8-point diagnostic run
  • Findings document with the gap quantified
Days 4–8
Rebuild
The measurement layer is rebuilt to survive browsers and consent.
  • Server-side event pipeline stood up
  • Deduplication keyed correctly end to end
  • Match-quality parameters widened
  • Consent states wired through properly
  • GA4 model and groupings corrected
Days 9–10
Verify
We prove the gap closed, then hand you the means to check it yourself.
  • Re-reconciliation against the same window
  • Before/after capture-rate comparison
  • Monitoring so drift surfaces early
  • Walkthrough and documentation
From A Live Audit

What this looks like when we run it.

~30%
Share of real orders the pixel was recording when we opened the account
1.34×
True blended marketing efficiency ratio across the audited period
The Engagement

A global wellness brand running paid social across multiple markets asked us to explain why in-platform performance and actual revenue kept diverging.

We reconciled platform-reported purchases against store orders at the order level. The pixel was capturing roughly 30% of real purchases. The optimisation algorithm had been learning from a third of the available signal for months — and every creative and budget decision made in that period rested on it.

Restoring the server-side conversion pipeline was named the account's first priority ahead of any spend or creative change, because no other lever can be measured reliably until it is fixed.

What We Will Not Claim

We are showing you what the audit found, not a revenue number attached to the repair. Any agency quoting you a guaranteed lift from a tracking fix is quoting you a number they cannot support. The honest promise is narrower and more useful: you will know the real size of your gap, and it will be closed.

Engagement

Fixed scope. Fixed price. No retainer required.

The audit stands alone. If the gap turns out to be small, you have paid for certainty and we will tell you so plainly.

Attribution Repair
€2,500 one-time

Ten working days, reconciliation through verification.

  • Full 8-point diagnostic
  • Order-level reconciliation, before and after
  • Server-side conversion pipeline built
  • Deduplication and consent wired correctly
  • GA4 model and channel groupings corrected
  • Findings document and walkthrough
Measurement Retainer
€1,500 / month

Optional. Tracking degrades — themes, apps and platform changes break it quietly.

  • Continuous capture-rate monitoring
  • Drift alerts before they cost a month
  • Event coverage extended as you add channels
  • Monthly reconciliation report
  • Tracking review on every site release
The repair fee is credited in full against the first three months if you continue onto any Prosperus system engagement.
Questions

Straight answers.

Our agency says tracking is fine. Why would this find anything?
Because "fine" usually means the pixel fires and the platform shows conversions. That is a liveness check, not an accuracy check. The only test that matters is reconciling platform-reported purchases against real orders — and in our experience that is almost never run.
What access do you need?
Read access to your ad accounts and analytics, an order export for the audit window, and collaborator access to your store and tag manager for the rebuild. No credentials are shared over email and access is handed back at the end.
Will this disrupt live campaigns?
The new pipeline is built in parallel and validated against a control period before anything is switched over. Campaigns keep running. You should expect the reported numbers to change after cutover — that is the fix working.
Does my in-platform ROAS go up or down?
Usually up, because conversions that were happening but never recorded start being counted. But the goal is not a better-looking dashboard — it is a dashboard that matches your bank statement, so the optimisation algorithm learns from something real.
We are on Shopify with a dozen apps. Does that complicate it?
It is the common case, and app sprawl is frequently where the signal is lost — checkout extensions, subscription apps and theme changes all break tracking silently. The reconciliation finds it regardless of cause, because it starts from orders and works backwards.
Is this GDPR-safe?
Server-side tracking done properly is more compliant than the browser-side default, not less: consent states are honoured explicitly and identifiers are normalised and hashed before leaving your infrastructure. Hashing discipline is check eight of the diagnostic for exactly this reason.
Next Step

Find out how much of your revenue your platform cannot see.

A thirty-minute call. Bring your ad account and last month's order count — we can usually estimate the gap before the call ends.

Book a Strategy Call  → See the full system