The problem. Google reports forty conversions. Meta reports thirty five. Your CRM recorded fifty enquiries in total. Everyone in the room can see the arithmetic does not work, and the meeting turns into an argument about whose numbers are right.
They are all right, and they are all describing something different. Attribution stopped being reliable some years ago and is not coming back. The useful question is how to allocate budget anyway.
What it costs the business
Indecision, mostly. When nobody trusts the numbers, budget gets allocated by whoever argues most persuasively, or it stays frozen in last year split because changing it cannot be justified.
The sharper cost is cutting the wrong channel. Upper-funnel activity rarely takes credit for the sale under any attribution model, so it looks expensive and gets cut first. Enquiries then fall a month or two later, by which point the decision looks unrelated.
Why the numbers disagree
Each platform counts conversions it feels responsible for. If someone saw a Meta ad and later clicked a Google ad, both will claim it. Neither is lying, because each is answering the question of whether it contributed.
Attribution windows differ, and view-through counting differs even more. A platform crediting itself for an impression nobody clicked is measuring something quite unlike a click.
Tracking genuinely breaks. Privacy controls, blockers, people switching from phone to laptop, and the very common Indian pattern of seeing an ad and then messaging on WhatsApp, which severs the digital trail entirely.
That last one matters more here than in most markets. A meaningful share of enquiries arrive through a channel that no attribution model can see.
Decide without waiting for certainty
Pick one source of truth and use it for decisions. Almost always your CRM, because it counts real business events rather than platform-defined ones. Platform numbers become directional, useful for optimising within a channel and not for comparing across them.
Ask people how they heard about you. An unglamorous open field on your enquiry form, or a question your sales team asks. The data is messy and incomplete. It is also the only input that captures the WhatsApp path, word of mouth, and the ad someone saw three weeks ago. Compare its pattern to your platform reports and treat large disagreements as information.
Watch total blended cost per acquisition. All marketing spend divided by all new customers, over a month. It attributes nothing and cannot be gamed, and its trend line tells you whether the whole operation is getting more or less efficient.
Test by turning things off. The most reliable method available and the most avoided. Pause a channel in one region for three weeks and watch total enquiries, not that channel enquiries. If the total holds, you learned something expensive-to-learn any other way.
How to hold the budget conversation
Stop asking which channel produced this sale. Ask what happens to total enquiries if we move fifteen per cent from here to there, then move it and observe. Directional experiments beat precise attribution, because the precision was never real.
Two guardrails. Move budget in increments small enough to reverse. And give each change long enough to cover your sales cycle before reading the result, because a two-week read on an eight-week cycle is noise.
What not to do
Do not buy attribution software expecting it to resolve this. Better tools narrow the gap; they do not close it, and the confident number they produce can be more dangerous than the honest uncertainty it replaced.
And do not let each platform mark its own homework. If your reporting is the sum of what Google and Meta each claim, you are counting some customers twice and planning against a number that does not exist.
The short version
Attribution is broken and the correct response is to lower your dependence on it. One source of truth, a how-did-you-hear question, blended cost per acquisition as the trend, and deliberate on-off tests when a decision genuinely matters.
Our performance marketing page covers how we report across channels, or show us your conflicting dashboards and we will tell you which number to plan against.
