Attribution that survives a board meeting
If 19% of your closed deals can be traced to a channel, your budget is being allocated by argument. Here's the measurement stack we build before touching a campaign.
A financial services CMO showed us her quarterly board deck. Six channels, $240,000 in quarterly spend, and a single slide attributing closed revenue.
19% of closed deals were attributed to a channel. The other 81% were "direct" or "unknown."
She wasn't being sloppy. That is what almost every default analytics setup produces once you account for consent banners, iOS privacy changes, long B2B sales cycles and the fact that most buyers touch a brand six times before they ever fill in a form.
The consequence is not merely that reporting looks bad. It's that budget allocation stops being an evidence-based decision and becomes a debate — and debates are won by whoever presents most confidently, not by whoever is right.
What "unknown" is hiding
When 81% of revenue is unattributed, you are not missing 81% of the picture evenly. The gap is systematically biased against exactly the channels that matter most in a considered sale:
- Organic search loses attribution because a buyer researches for weeks, then converts via a direct visit.
- LinkedIn and paid social lose it because click-through happens on mobile and conversion on desktop.
- Content and email lose it because they influence a decision without ever being the last touch.
Meanwhile branded search — where someone types your company name because another channel already did the work — is credited generously and looks like your best performer.
It is the channel doing the least original work and it gets the most credit. Budget follows. This is how companies end up quietly defunding the thing that was actually generating demand.
The stack that fixes it
There is no single tool. There are four layers, and they have to be built in order.
Layer 1 — Server-side event collection
Client-side tags are blocked by ad blockers, capped by ITP, and dropped by consent tooling. Move conversion events server-side so collection isn't dependent on the browser cooperating.
This alone typically recovers 15–30% of previously missing events.
Layer 2 — A durable, first-party identifier
A first-party cookie or an authenticated user ID that survives across sessions and devices where the user is logged in. Not a third-party cookie — those are gone and are not coming back.
Layer 3 — Source carried into the CRM
This is the layer almost everyone skips, and it is the one that makes the difference.
The original source must be written onto the lead record at creation and carried through to closed-won. Not stored in analytics, where it lives in a separate system that speaks a different language from your revenue data — written onto the opportunity itself.
Once that is true, you can ask the only question that matters: which channels produced revenue, rather than which produced form fills.
Layer 4 — One agreed definition of a qualified lead
Marketing counts form submissions. Sales counts opportunities. Finance counts closed revenue. All three report a different number for the same quarter, and everyone concludes the data is untrustworthy.
Write one definition down. Get sales to agree to it. Report against it exclusively, including when it makes marketing look worse.
What it produced
For the client above, attribution coverage moved from 19% to 81% of closed deals over about four months.
The findings were not comfortable:
- Two channels consuming 22% of budget had produced no traceable closed revenue in eighteen months. Both were cut.
- Organic search was producing 3.4× the pipeline previously credited to it. Its budget nearly doubled.
- The highest-converting single asset was a comparison page nobody had updated in two years.
Blended cost per opportunity fell 44% — on a smaller total budget.
The uncomfortable requirement
This only works if you are willing to act on results that contradict what you have been telling the board.
The two channels that got cut had been described in previous quarters as strategic investments. Presenting evidence that they produced nothing required someone to say so out loud.
Attribution isn't a reporting problem. It's an accountability problem wearing a reporting problem's clothes. The measurement is the easy half.
If you are not prepared to defund something on the evidence, you don't need better attribution. You need a better argument, and those are cheaper.
The full programme is written up in the Meridian Capital case study. If your board deck has an 81% shaped hole in it, let's talk.
We do this for a living
If any of the above describes your situation, a 30-minute call will tell you whether it's worth doing anything about — including if the answer is no.