Measuring organic when attribution is already broken
Zero-click results, assistant answers, and privacy changes have made last-touch attribution a poor description of reality. That is an argument for better measurement, not for measuring less.
What the gap actually is
Someone reads your comparison page on a work laptop, thinks about it for two weeks, asks an assistant a related question, sees you mentioned, and eventually types your brand name into a browser. Your analytics records a direct visit from a returning user. The organic work that produced it is invisible in the report.
This is not a tracking bug to be fixed with better tagging. It is a structural mismatch between how buying now happens and what session-based attribution can observe. Teams that respond by demanding tighter attribution get more precise measurement of a shrinking observable slice.
This is the organic-specific case of a general problem. For the wider set, see the marketing metrics worth reporting.
The gap is widest in B2B SEO, where a nine-month sales cycle can outlast the reporting window entirely.
Three layers worth measuring separately
- Presence. Are you in the index, in the considered set, in assistant answers, on the queries that matter. Measurable with rank tracking, impression data, and periodic assistant sampling. Leading, noisy, cheap.
- Engagement. Do the people who arrive do something meaningful. Scroll depth is weak; tool usage, downloads, and return visits within a fortnight are much stronger.
- Demand. Branded search volume, direct traffic to deep URLs rather than the homepage, and self-reported source on forms. Lagging, imprecise, and the closest honest proxy for whether the whole thing is working.
Read them together. Presence rising while demand stays flat means you are visible on queries that do not convert. Demand rising while presence is flat usually means something outside organic is doing the work, which is worth knowing before you claim the credit.
If you need to put a defensible number on the demand layer before the pipeline data exists, the SEO ROI calculator models it from your own inputs.
Presence is also the layer most often confused with progress, which is part of why audit backlogs feel productive without compounding.
The self-reported question earns its place
One free-text field on the form asking how someone heard about you outperforms most of the attribution stack, for one reason: it captures the influence that never produced a click.
It is messy. People misremember, and some answer “Google” for a referral they got from a colleague. Read it as a distribution rather than as records. When a channel that your analytics shows as negligible turns up repeatedly in free text, that is a real finding and your analytics is the thing that is wrong.
Hold-outs, where you can get them
The cleanest available answer to “did this work” is still a hold-out. Deliberately do not publish for one segment, one region, or one product line, and compare against a segment where you did. The same problem in its sharpest form, where there is no rank to track at all, is covered in tracking brand mentions in AI search.
For most organic programmes this is impractical, because you cannot un-publish and the effects take too long. Where it does work is with generated page families and with content targeting geographic or vertical segments, precisely because those come in discrete blocks you can withhold. If you are building one of those, decide the hold-out before the build, not after somebody asks for proof.
If an agency is doing the work, agree the hold-out before the engagement starts. Choosing a SaaS SEO agency covers what else belongs in that contract.
Measure presence, engagement, and demand as three separate layers and read them together. A single attribution number describing a channel that mostly operates without clicks is a number describing the part you can see.
I put more weight on one free-text form field than on most of the analytics stack, and I am aware of how that sounds. It is the only instrument that captures influence which never produced a click, and in B2B that is the majority of it.
When presence rises and demand stays flat, I stop adding content and go look at which queries we are winning. Usually we have become visible on terms nobody with budget is searching. That is a targeting problem, and more publishing makes it worse rather than better.
Frequently asked
Why does my analytics show organic traffic falling but leads steady?
What is a realistic organic conversion rate for B2B?
Should we use last-touch attribution for organic?
How do I prove SEO caused a revenue increase?
I take on a small number of engagements at a time. If you are past product-market fit and want to talk through what this would look like for your situation, the calendar is open.
Book a call