Marketing Attribution for a Mid-Market B2B Company: The One Table the CFO Will Trust.
The marketing dashboard says the paid search campaign produced forty percent of pipeline. The CFO asks how that was calculated and the answer involves a model: last touch, or first touch, or a multi-touch weighting the platform chose. The CFO asks what the other model would say and the answer is a different number. The meeting ends with the budget flat, because the only evidence for it was something finance could not reconcile.
Attribution in a mid-market B2B company is not a modelling problem. It is a bookkeeping problem, and the table that survives the budget meeting is built the way finance builds everything: from recorded facts, joined on a key, with an honest line for what could not be assigned.
Why the models fail the finance test.
A multi-touch attribution model assigns fractional credit to each contact a buyer had with the company before buying. The weights are assumptions. Forty percent to the first touch, forty to the last, twenty spread across the middle, or some decay curve a vendor chose. Change the assumption and the answer changes, which is why two tools on the same data disagree.
Finance has a rule for numbers like that: they are estimates, and estimates do not go in the ledger. A controller will accept an estimate for a provision, with a note. A controller will not fund a channel on one. The problem is not that the models are wrong. It is that they cannot be checked, and a number that cannot be checked is a number that gets cut first.
There is a second problem specific to B2B. The buying cycle is long and the buyer is several people. The person who clicked the ad in March is not the person who signed in September, and the model that connects them is guessing. Multi-touch attribution was built for consumer purchases with one person and a short window. It is being sold to companies whose sales do not look like that.
The three recorded facts.
The table the CFO accepts has three inputs, and every one of them is a fact somebody recorded rather than a weight somebody assumed.
The source on the lead. When a lead is created in the CRM, two fields are set from where it came: the source (paid search, organic search, email, referral, event) and the campaign, carried in from the tracking parameters on the form or entered by the person who logged the call. This is a fact about how the lead arrived, and it is set once. We treat the lead record as the ledger entry for marketing, and its source fields are the account code.
The qualification status, set by a person. A lead becomes a qualified conversation when a salesperson reviews it and sets the status: sales ready, or whatever the company calls it. Not when a form is filled, not when a score crosses a threshold, when a person decides. This is the only count marketing should call a result, because it is the only one a human vouched for.
The spend, from the invoice. Per channel, per month, from what was actually paid: the ad platform's invoice, the agency fee, the tool licence. Not the budget, not the platform's own spend report. The number finance already has.
Three facts. None is a model.
The table.
Join them on the lead, and group by source. The table has one row per source and five columns: leads created, qualified conversations, spend, cost per qualified conversation, and the month. One more row at the bottom, labelled unattributed, for the leads whose source fields were empty.
That last row is the one that earns the CFO's trust, because it is honest. A model distributes every lead across channels by assumption and never shows a gap. The table shows the gap, and the gap is a measurable operational problem: the forms that do not carry tracking, the salespeople who log calls without a source, the website that drops the parameters on the way to the CRM. Fix the doors and the row shrinks. Finance understands a reconciling item; it does not understand a weight.
Our own version of this table runs every day from the CRM and the ad platform, and we publish the result in the piece on what it costs to run a marketing function as a system. The method is the same for any company whose leads land in a CRM: the fields exist, the status exists, the invoices exist. What is usually missing is the discipline of setting the fields at the door and the join that puts the three together.
What the table does not do.
It does not credit the second touch. If a buyer found the company through a search ad in March, read three articles, and filled in a form after an email in September, the table credits the email, because that is the source on the lead that converted. The search ad's contribution is invisible.
That is a real limitation and the honest answer is to accept it rather than model around it. The invisible contribution is seen another way: when the search ads are paused, the email channel's qualified conversations fall two months later, and that is an experiment with a result rather than a model with a weight. A company that cannot afford the experiment cannot afford the model either, because the model's answer would not have been trusted.
The table also does not measure brand, content reach, or the articles the buyer read in between. Those are measured on their own terms, as reach and engagement, and reported beside the table, not inside it. Mixing them in is how dashboards end up with pipeline numbers nobody can reconcile.
Building it in the CRM.
The work is small and most of it is a door-by-door audit. Every form that creates a lead carries the source and campaign fields and the tracking parameters reach them. Every imported list sets the source on import. Every manually created lead requires the source field before it can be saved, which is one validation rule. The qualification status is a picklist a person sets, with no automation allowed to set it. Spend is loaded monthly from the invoices into a small table with channel, month and amount.
Then one report, grouped by source, with the qualified count and the spend joined by channel and month. In Zoho CRM that is a custom report and one lookup; it is a morning's configuration once the fields are right.
The result is a table finance can audit: pick any row, trace the qualified conversations to named leads with a human-set status, trace the spend to an invoice. That is the standard the rest of the company's numbers meet, and the budget meeting goes differently when marketing's numbers meet it too.
Discovery is no-risk: we audit the doors, build the table, and you pay only if you go ahead.
Find out what one connected Zoho system would change in the way you run.
In a no-risk discovery we look at your CRM, finance and operations systems and who owns each part, and show what a connected system would do differently. You pay only if you proceed. Or see how we approach it.
More on the same problem:
Comments