The year measurement got harder
Between platform privacy changes, the ongoing move away from third-party cookies, and consent rates that fell as soon as banners became genuinely enforced, a lot of teams ended this year measuring less than they did at the start of it.
The reaction I've seen most often is to treat this as a crisis and go looking for a replacement — a better attribution tool, a modelling vendor, a first-party data project that will restore the numbers.
I think that reaction misdiagnoses what was lost, because most attribution reporting was never doing the job people believed it was doing.
Two jobs, permanently confused
Attribution gets used for two very different purposes, and they have almost nothing in common.
Scoreboard attribution answers: which channel deserves credit for this revenue? It settles budget disputes, justifies headcount, and populates the board slide. It is fundamentally a political instrument.
Decision attribution answers: if I move the next £10k, where should it go? It is directional, comparative, and only has to be right enough to rank options.
The first requires precision it has never actually had. The second requires far less than most teams assume. What broke this year was mostly the first.
Why the scoreboard was always fictional
Consider a deal that closed last month. The buyer read a comparison post 14 months ago, forgot about you, saw a colleague mention you in a Slack community, searched your brand name, clicked a paid brand ad, and converted.
Last-click gives everything to paid brand search. First-click gives everything to a blog post from over a year ago. A linear model splits it evenly across whatever happened to be trackable, which excludes the Slack conversation entirely — the step that probably mattered most.
None of these is correct. They're three different fictions, and the organisation picks whichever one supports the decision it wanted to make.
Once you've seen that clearly, losing 20% of tracking coverage looks less like losing the truth and more like losing a comfortable illusion.
What still works for deciding
You don't need perfect attribution to allocate budget well. Four things have held up for me this year, none of which depend on complete tracking.
1. Self-reported source, at the point of conversion.
One optional open-text field: "How did you hear about us?" Messy, biased toward whatever's top of mind, unusable in a dashboard.
Also the only place channels like podcasts, communities, word of mouth, and events show up at all. In several accounts this year, the qualitative answers pointed at a channel that the analytics platform valued at almost zero.
2. Holdout and geo tests.
Turn a channel off in one region for four weeks and watch total pipeline, not that channel's attributed pipeline.
This is the closest thing to a real answer available, it requires no tracking whatsoever, and it's the single most underused technique in B2B growth measurement. The main cost is nerve.
3. Closed-won by source cohort, at low resolution.
Not "which touchpoint," but: of deals closed this quarter, what proportion had any interaction with paid, with content, with events? Directional, robust to tracking gaps, and enough to notice when something has stopped contributing.
4. Sales conversation intelligence.
Ask reps what prospects say on first calls. "We've been reading your stuff for a while" is attribution data. It's just stored in a person rather than a platform.
The uncomfortable implication
If you accept this, budget allocation stops being a calculation and goes back to being a judgement, informed by imperfect evidence from several directions.
That's harder to defend in a board meeting than a dashboard that assigns £847,000 of pipeline to LinkedIn with two decimal places of false confidence. But it produces better decisions, and it doesn't collapse the next time a platform changes its rules.
What I'd actually do in Q1
Run one holdout test on your largest channel. Add the self-reported field if you don't have it. And stop reporting attributed revenue by channel to three significant figures — replace it with a ranked view and a stated confidence level.
The precision you're giving up was never real. What you get back is a measurement approach that survives the next privacy change, which on current evidence is not far away.
Hilal Tasdan
B2B SaaS Growth Marketing Consultant & Fractional CMO. Partner in Growth.