When answers absorb the journey, sessions stop measuring influence. The replacement is a question panel: the fifty questions that decide purchases in your category, scored monthly for citation share, accuracy, and displacement. Three numbers, tracked in a spreadsheet, that tell you more about future pipeline than any traffic chart.
Why your analytics understate you
A buyer asks an assistant who does the best ecommerce builds in Melbourne. The assistant names you, summarises what you do, and mentions your pricing. A week later that buyer emails you directly. In your analytics, this appears as "direct traffic", or doesn't appear at all. The most influential moment in the funnel happened off your property, on a surface you don't instrument, and your dashboard has no idea. This is the zero-click search reality, and it breaks the measurement most teams rely on.
The click was always a proxy for influence. It was just a convenient one, so we treated it as the thing itself. As answer engines absorb the informational half of search, informational traffic falls even for brands whose influence is growing, and teams that only read sessions conclude they're losing exactly when they're winning. Judged by clicks alone, the best possible outcome (the assistant answered so well the buyer came straight to you) looks like a marketing failure. The economics behind this are chapter three of Be the Answer.
Build the question panel
Write down the fifty questions that decide purchases in your category. They fall into predictable families: definitional ("what is answer engine optimisation"), comparative ("Webflow vs WordPress for a marketing site"), local ("best branding agency in Melbourne"), pricing ("how much does a Shopify build cost"), and evaluative ("is [your company] any good", "alternatives to [your company]"). Pull them from sales calls, support tickets, and the autocomplete suggestions the engines themselves offer. If a question has never preceded a purchase, it doesn't make the panel. Building Your Question Panel covers the method in full.
Then run the panel. Once a month, put every question to the major engines and record four things per answer: were you named, were you cited as a source, was what they said about you accurate, and who appeared instead of you. A spreadsheet and an hour a month is genuinely enough; the discipline matters far more than the tooling.
Traffic will keep falling for informational queries. Influence doesn't have to.
The three numbers to watch
Citation share: of your panel questions, in what fraction are you named or sourced? This is your new rankings report, except it measures presence in the answer rather than position on a page nobody scrolled. Watch it by question family; being absent from pricing questions is a different problem to being absent from best-of questions, and the fixes are different too.
Accuracy: of the answers that mention you, how many are right about what you do, who you serve, and what you charge? An assistant confidently repeating your 2022 service list is doing damage at scale, and you will never see it happen unless you're looking. Accuracy problems are usually source problems: somewhere on the quotable web, an old version of you is still the loudest. Auditing how AI currently describes your brand is where that clean-up starts.
Displacement: which competitors appear on questions where you don't? This is the sharpest number of the three, because it converts a vague anxiety ("are we visible in AI?") into a target list. Open the pages the engines cited instead of you and you are looking at exactly what the machine preferred. Usually it's structure and specificity, not brand size.
How to move the numbers
The levers are the citation signals themselves. Restructure the pages that should win each lost question so the direct answer sits in the first paragraph. Replace adjectives with numbers. Put named authors on everything. Keep your descriptions consistent across every directory and profile that mentions you, because corroboration decides whether an engine trusts the claim enough to repeat it. Then re-run the panel and watch which changes moved which questions. That feedback loop, change one thing and measure the drift, is the whole practice, and it's the same one behind why AEO is the new SEO.
What to tell the board
Report presence alongside traffic, and say plainly that informational traffic is expected to decline without that being a failure. A monthly line for citation share, a note on accuracy fixes shipped, and the displacement list of competitors gaining ground: that reads as control, not excuse. The teams in trouble are the ones still promising session growth in a market where the sessions are evaporating for everyone.
Start before it's obvious
You can only manage what you measure, and you can only measure from a baseline. The panel takes an afternoon to build and an hour a month to run. The brands that start now will know exactly what moves their numbers by the time everyone else starts asking how to measure this. Being early to the measurement is how you end up early to the result.
