Tracking audit and specification
Every existing tag, event and conversion catalogued, then a written specification of what should be collected, in what shape, with which identifiers attached.
Build & Measure
Attribution that reconciles to the ledger.
Four dashboards that disagree is not a measurement problem, it is a decision problem. We build one server-side data layer joined to your CRM, and the numbers it produces survive a finance review.
- Collection model
- Server-side, first-party
- Source of truth
- Your CRM
- First reconciled report
- Week five
- Minimum engagement
- Six months
What the engagement covers
From event specification to one shared report.
This is the layer every other marketing decision rests on, and it is the one most often assembled by whoever had access to the tag manager that quarter.
Server-side collection
First-party endpoint under your own domain, with events validated before they are forwarded. Data quality problems get caught at the boundary instead of in a report six weeks later.
CRM and revenue joins
Sessions and campaigns joined to opportunity records and contract value, so a channel is judged on closed-won revenue rather than on form submissions.
Incrementality testing
Geo splits and holdout groups designed and run properly, which is the only method that answers whether a channel adds revenue or merely records it.
Reporting layer
One report covering spend, qualified pipeline, closed revenue and payback by channel, defined once and refreshed automatically. No manual assembly, no version disputes.
Consent and governance
Consent state respected at collection rather than filtered later, with retention periods, lawful basis and data flows documented for your privacy record.
Why this works
Four principles that make the numbers hold.
Measurement projects usually fail politically rather than technically. Somebody produces a number that contradicts a number a colleague has been reporting for a year, and the project quietly stops.
-
Events are specified before they are built
A written specification agreed by marketing, sales and engineering comes first: event names, properties, identifiers and when each fires. Tracking assembled ad hoc in a tag manager works until the person who built it leaves, and then nobody can explain what any metric means.
-
Server-side is about durability, not just blocking
Moving collection server-side does recover data lost to blockers and browser restrictions. The larger benefit is control: you can validate, correct and enrich events in one place, and a platform changing its client library no longer silently breaks a quarter of reporting.
-
Closed-won is the only outcome that counts
Form fills and trial signups are steps, not results, and optimising toward them reliably produces more of the wrong ones. We join everything to contract value in your CRM, which is when campaign reporting starts matching what your finance lead already believes.
-
Incrementality, not attribution models
Every attribution model is a set of assumptions about credit, and choosing one is choosing which channel to flatter. Holdout and geo testing measures what actually changes when a channel stops. It is slower, occasionally uncomfortable, and it is the only version that survives scrutiny.
What you get
Outcomes you can hold us to.
The first deliverable is a specification you own. Even if the engagement stops there, you keep a document that makes every future measurement decision cheaper.
Week 5
First reconciled report
Spend, pipeline and closed revenue by channel, agreeing with your CRM and your finance figures rather than approximating them.
Within 2%
Reconciliation acceptance gate
Reported closed revenue must reconcile to the finance ledger within the tolerance agreed in the specification before budget decisions use it.
One report
Numbers everyone shares
Marketing, sales and finance work from the same definitions, so review meetings discuss decisions instead of whose figure is correct.
Zero
Manual assembly steps
No spreadsheet in the middle of the process. Every figure traces back to a specified event or a CRM record you can open.
The detail
Terms, team and stack.
The data layer is built in your infrastructure and documented so your own engineers can maintain it. Nothing here depends on a tool only we know how to operate.
Discuss an engagement- Onboarding
- Tracking audit, event specification document, consent review
- Team assigned
- Analytics engineer, data analyst, front-end developer
- Stack
- Server-side tag management, first-party collection, your warehouse or ours
- CRM systems
- HubSpot, Salesforce, Pipedrive, or a custom system via API
- Reporting
- Weekly channel report, monthly economics review with finance
- Governance
- Consent handling, retention policy, documented data flows
Questions we are asked first
Before you send a brief.
If your question is not here, write to info@bari-buti.com and you will get a direct answer from the engineer who would build it.
Not at the start. For most companies under a few thousand opportunities a year, the CRM plus a well-built reporting layer is enough, and a warehouse adds cost without adding an answer. We will tell you when your volume genuinely justifies one, and it is usually later than a vendor would suggest.
It improves control compared with client-side-only collection, though nobody should promise complete data. The collection gap is measured during implementation rather than estimated in advance, so later trends can be assessed against a documented baseline.
You do. It runs in your infrastructure, under your domain, with the specification, queries and configuration handed over in writing. Measurement built so that only the agency understands it is a commercial tactic, and it is the reason so many companies are afraid to change agencies.
Usually, yes. Most of the work sits in collection and the joins rather than in your templates. Where a form or a funnel step genuinely cannot be instrumented as built, we say so and quote that change separately instead of quietly reporting an approximation.
Ad platforms report on themselves and each one counts a conversion it can plausibly claim, which is why the totals exceed your actual sales. Their numbers are useful for optimising inside a channel and unusable for deciding between channels. Our reporting answers the second question, and it will frequently show a lower figure than a platform dashboard.
Works well with
The disciplines this one makes accountable.
Measurement earns nothing by itself. It earns by changing where the budget goes, which means it is worth most alongside the channels it judges.
Next step
Give us read access, and we will show you where your reporting disagrees with itself.
The tracking audit is a fixed-fee engagement producing an event specification and a ranked list of what is currently being measured incorrectly.