Pricing the Answer: A Zero-Click Worksheet for US Publishers in 2026
Most publisher conversations about AI Overviews begin with a screenshot of a falling clicks line and end with a vague resolution to "do more video." That is an emotional artifact, not a business case. A chief financial officer cannot allocate against a screenshot. What follows is a way to turn the shift into something a US publisher can actually file: a worksheet, built from data you already own in Search Console and your analytics platform, that prices exposure to AI-generated answers instead of merely lamenting it.
The framing matters. If you treat zero-click as a traffic grievance, you will spend your reporting cycles arguing about attribution. If you treat it as a budgeting problem — a question of which impressions convert into sessions, which impressions convert into brand presence, and which simply evaporate — you end up with numbers that support decisions about staffing, format investment, and licensing.
Step One: Split Impressions by Query Intent, Not by Page
The default Search Console view lumps everything into a page-level total. That is the wrong unit of analysis for this problem, because AI Overviews do not treat all queries equally. They appear heavily on informational and comparison queries and much less reliably on branded, navigational, and highly specific transactional queries.
Export your Performance report with queries and pages, then segment into five buckets:
- Branded — queries containing your publication name or established column names.
- Informational head — broad "what is" and "how does" phrasing where answer engines are most aggressive.
- Comparison and evaluation — "best," "vs.," "worth it," "alternatives."
- Local and time-sensitive — weather, hours, scores, prices, election results.
- Long-tail transactional — narrow, specific, low-volume queries where a click is usually the only way to get an answer.
For each bucket, calculate a ratio: sessions divided by impressions, using a matched date range and the same country filter. US-only data is essential here; blended global figures mask the pattern badly.
Step Two: Build the Exposure Ledger
The worksheet has four columns per bucket. Keep it in a spreadsheet, not a slide; you will want to re-run it quarterly.
- Impressions — 90-day total.
- Expected sessions — impressions multiplied by a baseline ratio, taken from a pre-AI-Overview period you trust, or from a control segment (branded queries work well as a stable comparator).
- Actual sessions — from the same export.
- Gap — expected minus actual.
The gap is your exposure. It is not automatically a loss, because some of those impressions still produced a citation or a brand impression inside an answer panel. That distinction is the whole point of the exercise, and it is where most publishers stop too early.
Step Three: Assign a Value to Citation Visibility
A citation inside a generated answer is not a session, but it is not zero either. Three defensible ways to price it:
- Impression-value method. Apply a conservative display-equivalent value to the impression, discounted heavily to reflect that the reader did not choose to visit. Even at a fraction of a cent, at scale this produces a meaningful credit line.
- Branded lift method. Track branded search volume for publications that receive frequent citations. If branded impressions hold steady while non-branded sessions fall, part of the exposed traffic is being converted into future demand rather than destroyed.
- Assisted-conversion method. In GA4, examine sessions that arrive after a user has previously been exposed to your brand, and compare conversion rates for those sessions against cold sessions. The delta is rough but directionally useful.
Whichever method you choose, apply it consistently and document the assumption in one sentence. A number with a stated assumption survives scrutiny. A number without one does not.
Step Four: Find the Subsidy and the Cannibalization
Sort your buckets by gap size and by monetization rate. That two-dimensional view produces the only chart most executives need.
Subsidized categories are those with a large exposure gap but negligible revenue per session. Recipe-style explainers, definition pages, and evergreen backgrounders often land here. Losing these clicks hurts the vanity metrics and almost nothing else. Cutting investment here is rational, and reallocating the savings is the fastest win available to most US newsrooms in 2026.
Cannibalized categories are the dangerous ones: meaningful exposure gaps paired with high revenue per session. Buying guides, product comparisons, and service journalism in competitive verticals frequently belong in this group. These are the pages where an answer panel substitutes for a visit rather than merely preceding it, and where the loss shows up in affiliate and subscription revenue months later.
A third, easily missed category is the subsidy in reverse: pages that gain impressions inside answer panels and see modest but unusually high-quality session growth, because the readers who do click arrive pre-qualified. Flag these and protect them.
Step Five: Convert the Ledger into a Budget Request
The output of this exercise is not a grievance memo. It is three lines an editor can take into planning:
- "This category generated X sessions per thousand impressions a year ago and Y this quarter; at our current revenue per session, that is Z dollars annually, of which we credit W back to citation visibility."
- "The net exposure sits in four page types, listed here, which represent N percent of our editorial calendar."
- "We propose shifting M hours per week from category A to category B, where answer engines still require a click to complete the task."
That last point deserves emphasis. Formats that resist summarization — original data, interactive tools, calculators, local service directories, anything requiring a logged-in state or a transaction — remain click-dependent. The worksheet exists to prove, with your own numbers, which of those formats deserve the hours you are currently spending on pages that an answer engine can absorb whole.
A Caution on Precision
Search Console impressions are estimates, analytics attribution is imperfect, and AI Overview presence is not directly reported. Do not present the ledger as an audit. Present it as a model with stated inputs, re-run on a fixed cadence, and used comparatively — this quarter against last, this category against that one. Consistency, not decimal places, is what makes the argument hold up when someone in the room asks how you know.