Contents
01Why the numbers disagree
Almost every disagreement between an ad platform and a CRM comes from one of four causes, and only one of them is a tracking fault. Platforms count a conversion against the click they attribute it to, in their own window, including modelled conversions they estimate rather than observe. The CRM counts what a human entered. Browsers block scripts and shorten cookie lifetimes, so some real conversions never reach the platform at all. And the same enquiry often arrives twice, by form and by phone, and is counted twice.
Before buying anything, compare like for like: the same date range, the same definition of a conversion, the same deduplication rule.
02What server-side tracking actually fixes
Sending events from your server rather than from the browser repairs the losses, and it does so reliably.
- Events that never left the browser because a blocker or a privacy setting stopped the script.
- Attribution that expired because a cookie lived for seven days and your sales cycle takes six weeks.
- The gap between a form submission and a signed deal, which is the only conversion worth optimising on.
- Duplicates, by sending one event with one identifier from one place.
03What it does not fix
It does not recover people who declined analytics. Consent still governs what you may collect, and in the US that increasingly means state privacy laws with real enforcement, on top of what the ad platforms themselves require.
It also does not fix a disagreement about what counts as a lead. If marketing counts form fills and sales counts qualified opportunities, both numbers stay right and the argument stays unresolved. That one is settled in a meeting, not in code.
A tracking project that starts without a written definition of a qualified lead produces a cleaner version of the same argument.
04When the work pays for itself
Our tracking and attribution repair is $4,800, done once. The payback is straightforward: if better signal lets the platform optimise toward qualified leads rather than form fills, a 10 to 20 percent improvement in cost per qualified lead is a common outcome, and at $5,000 a month in spend that returns the cost within a few months.
Below that level, spend the money on the cheaper fixes first. Deduplicate the double counting, send offline conversions from the CRM, and agree the definitions. Those are days of work, not weeks, and they often close most of the gap.
05How to tell if yours is already broken
Three checks take an afternoon. Compare conversions in the ad platform with deals in the CRM for the same month and the same definition. Look at how many enquiries have no source attached at all. And submit a test enquiry from a phone with a tracking blocker enabled, then see whether it appears anywhere.
If the third test fails, you are buying traffic you cannot measure, and every decision after that is a guess with a chart attached.
Questions
Is server-side tracking a way around consent?
No, and anyone selling it that way is selling a problem. Consent decides what may be collected. Server-side changes where the event is sent from, not whether you were allowed to collect it.
Will my reported numbers go up or down?
Usually down at first and then truer. Deduplication removes conversions that were counted twice, which looks like a loss for a week and makes every later comparison honest.
Can we do this without touching the website?
Partly. Offline conversions from the CRM can be sent without front end changes, and that alone is often the highest value step. Full coverage needs a change on the site as well.
Where we can help