The board asks for one figure: how much of what we sold this quarter came from digital marketing. Marketing brings twelve dashboards, one per platform, and none agrees with the others or with the CRM. The CFO nods, takes a note, and next year's budget gets decided without that figure. What follows explains why the dashboards will never answer the question, and where the answer does come from.
Twelve dashboards, twelve truths
Each platform measures its own success by its own rule. The search engine counts clicks. The social network counts reach. The ad platform counts conversions it defines itself, and claims them even when the buyer had already decided. Add up the twelve dashboards and you get more sales than the company made. The CFO knows that, which is why the CFO stops listening.
So far the usual problem. What is new is that in 2026 part of the buyers arrive after asking an AI assistant, and that path shows on no dashboard. The buyer who decided with ChatGPT and then typed the company name into Google is recorded as direct or brand traffic. The dashboard is not lying: it does not know.
The figure the CFO does accept
There is one figure a CFO accepts without argument: the one that comes out of the company's own system. In a company that sells to other companies, that figure is the quote request. How many arrived, from which market, through which channel, and how many turned into orders. That figure lives in the CRM, in the site form and on the phone line, not in the platforms.
So why does almost no company have it by channel? Because the quote form does not ask how the buyer got there, the CRM does not store the origin, and the phone call is not logged. You have seen why the dashboards fail and which figure works. What is missing is the step almost everyone skips, and it is the cheapest of all.
The one-line question that changes the report
Here is the step. Before buying an attribution platform, before hiring anyone, one required question goes on the quote form: how did you find us? With a short list and one option that cannot be missing in 2026: AI assistant (ChatGPT, Gemini, Perplexity), with a follow-up asking which one. The same question is asked on the call and stored in the CRM next to the order.
With ninety days of that question the company holds something no dashboard gave it: quote requests by channel and by market, with closed orders beside them. That is the table the CFO reads. And it is the table that closes every monthly report of my analysis: the point of the whole plan is the quote-request column, not the traffic.
| Channel | What the platform dashboard measures | What the CFO's table measures |
|---|---|---|
| Organic search | Clicks and impressions | Non-brand quote requests, with the order beside them |
| AI assistants | Nothing | Requests that declare they found you on ChatGPT, by market |
| Paid ads | Conversions the platform claims for itself | Real cost per request, against the CRM |
| Direct and brand | Visits | Requests that came through a referral or an assistant that went unrecorded |
In the illustrative case of my sample, the baseline was nine requests a month with no attribution; the plan starts with the question on the form and ends with the table by channel. Sample figures. Yours come out of your form as soon as the question exists.
Do I need an attribution platform?
Not yet. First the question on the form, the CRM and the call; with ninety days of data you decide whether anything more is needed. The analysis designs that path for your operation.
What about sales that never touch the site?
They are logged in the CRM with the same question. The CFO's table joins site, call and salesperson in one view by channel.
Does the analysis do this, or do I?
The analysis delivers the measurement design and the baseline. If your team wants to build it, it can; if you want me to fix the form and the flow, that is quoted separately, case by case.
