Organic clicks are flat. Impressions are up substantially. Enquiries have increased & the sales team says prospects arrive better informed than they used to. Every one of those statements can be true simultaneously, & a click-based report will describe that quarter as a failure.
Measurement has not kept up with how search works. When a growing share of queries are answered in the results, & another share are answered inside an assistant that never sends a visit, a dashboard built entirely on sessions will mislead you. Here is a framework that does not.
Accept what you have lost
Start by being honest about the limits. You cannot know how many people read an answer that cited you & did not click. You cannot see the full query set an assistant used. Attribution for assistant-sourced enquiries is frequently absent entirely.
Pretending otherwise leads to worse decisions than accepting the gap. The right response is triangulation: several imperfect indicators, read together, rather than one clean number that measures the wrong thing.
A metric you can measure precisely but which no longer reflects the outcome is more dangerous than an estimate that does.
Layer one: impressions & query coverage
Search Console impressions remain useful, provided you interpret them correctly.
Track the number of distinct queries your site appears for over time. Growth in query coverage indicates expanding topical presence even when clicks are static. Track impressions by query group rather than in total, separating informational from commercial queries, because they are moving in different directions & an aggregate hides that.
Watch click-through rate by query type as a diagnostic rather than a target. Falling click-through rate on informational queries alongside stable rates on commercial ones is the signature of answer surfaces absorbing informational demand. That is a market condition, not a performance failure, & it should be reported as such.
Layer two: citation tracking in answer engines
This requires manual work & it is worth doing properly.
- Build a query set of twenty to forty questions a genuine prospect would ask, spanning awareness, comparison & decision stages.
- Run them on a schedule, monthly is usually sufficient, across the assistants & search surfaces your audience actually uses.
- Record four things per query: whether you were mentioned, whether you were cited with a link, which page was cited, & whether the description of you was accurate.
- Record your competitors in the same way, so you have a share-of-citation figure rather than an isolated count.
- Log inaccuracies separately & treat each one as a task. A model describing your services wrongly is usually reading something outdated that you can fix.
Over six months this produces a genuine trend line. It is laborious, it is currently the only reliable method, & it answers the question executives actually ask, which is whether the business appears when someone asks an assistant about the category.
Layer three: brand demand
When people encounter you in an answer & do not click, a portion of them search for you later. That makes branded search volume one of the better proxies for unmeasurable exposure.
Track branded query impressions & clicks in Search Console, direct traffic to your homepage, & branded search volume in whatever keyword data you have. Look at the trend rather than the absolute number, & compare it against periods when you were & were not visible in answer surfaces.
Rising branded demand with flat non-branded clicks is a strong indication that unmeasured exposure is converting into deliberate visits. It is not proof, but combined with citation tracking it is a defensible read.
Layer four: enquiry quality, not just quantity
This is the layer most reporting omits & the one closest to the business.
Add a source question to your enquiry form, phrased openly rather than as a dropdown, & read the answers. When people say they were recommended by an assistant, or that they found you through a specific article, that is attribution no analytics platform will provide.
Then track quality downstream. Qualified enquiry rate, average deal size, & sales cycle length. A shift towards better-informed prospects who need fewer explanatory calls is a real commercial outcome & it frequently accompanies strong content & answer visibility.
Build the report around decisions
A report exists to support decisions. Structure it that way rather than as a data dump.
- Visibility. Query coverage, impressions by query group, citation share across your tracked query set.
- Demand. Branded search trend, direct traffic trend.
- Engagement. Clicks & sessions, reported with context rather than as the headline.
- Outcome. Enquiries, qualified enquiries, & closed business where the sales cycle allows.
- Actions. What the data implies you should do next quarter.
Present visibility first & outcome last, with engagement in the middle where it belongs. Leading with sessions trains everyone to optimise for a metric that is decoupling from revenue.
Set expectations before the numbers move
The hardest part of this is not analytical, it is organisational. If a business has spent five years judging its marketing on sessions, changing the report mid-year looks like moving the goalposts.
Introduce the framework before the trend forces it. Explain the mechanism: answers appearing in results, assistants absorbing informational queries, & the resulting decoupling of clicks from influence. Show the citation tracking method so it is understood as evidence rather than assertion. Agree the new reporting structure explicitly.
Then hold the line on outcome metrics. Enquiries & revenue have not changed meaning, & they remain the final arbiter. Everything above them in the report is there to explain why they moved.
If you want this built for your business, see how we approach answer engine optimisation & generative engine optimisation, or ask us to set up a citation tracking baseline.