What actually counts as a zero-click search
Three different things get filed under the same label, and conflating them is how people end up panicking about the wrong number.
The first is a satisfied zero-click search: Google answered it on the page. Exchange rate, PIN code, opening hours, an AI Overview summarising something you'd have skimmed anyway. The second is a redirected one: the click went to a Google property — Maps, Flights, Shopping, YouTube — so it counts as a click for Google and nothing for you. The third is abandonment: the searcher didn't find what they wanted and reformulated or left.
Only the first two are structural. The third is often a ranking problem wearing a scary costume.
How the rate has moved since 2019
There are two well-known public measurements, both from SparkToro, and it's worth quoting them precisely because they get misrepeated constantly.
In 2019, SparkToro's analysis with Jumpshot clickstream data put US zero-click searches at about 50.3%. In 2024, SparkToro's follow-up with Datos put the figure near 58.5% in the US and just under 60% in the EU. Both are clickstream panels of US and European desktop and mobile users — not India, not your industry, and not a Google-published statistic.
So the honest reading is: it was already about half before generative search existed, and it has climbed since. Anyone quoting a single dramatic percentage for "the death of organic traffic" is either citing one of these two studies badly or making it up.
Which query types resolve without a click
Exposure isn't evenly spread. Sorting by query shape gets you 90% of the way to a plan.
| Query shape | Example | Exposure |
|---|---|---|
| Single-fact lookup | "gst rate on packaged food" | Very high — Google answers it directly |
| Definition | "what is a canonical tag" | High — snippet or AI summary covers it |
| Quick comparison | "da vs dr" | Medium — summary answers, detail still earns clicks |
| Research with stakes | "seo agency pricing india" | Low — buyers want to see who's saying it |
| Transactional | "book seo audit" | Very low — the click is the point |
| Tool, template or download | "seo audit checklist xlsx" | Very low — the asset is on your site |
The triage: two piles, one afternoon
Open Search Console, pull the last twelve months of queries, and export the top 200 by impressions. Then put every query into one of two piles.
Pile one: answerable in a line. Anyone can satisfy this query with a sentence, so Google will. These will keep generating impressions and steadily fewer clicks. You still want a page for them — they build topical authority and they're what AI systems pull from — but stop forecasting traffic against them.
Pile two: needs a page. Comparisons where the reader wants to see reasoning. Pricing where they want to see who's committing to a number. Anything involving a decision they'll have to defend to someone else. These clicks are defensible because a summary can't finish the job.
Then move the budget. If 70% of your content plan sits in pile one, you're funding impressions. Rebalance toward pile two and toward pages that end in an action — a quote, a booking, a download. What SEO costs in India is a pile-two query. "What is SEO" is not.
What impressions without clicks are still worth
Something, but not what a dashboard implies. If your brand name appears in the answer a buyer reads, that's the same mechanism as a billboard: unmeasurable, non-zero, and impossible to invoice against.
The proxy worth tracking is branded search volume. If impressions on unbranded queries are flat or falling while people searching your brand name is rising, the unbranded work is doing something even though the click count says otherwise. If both are flat, the work isn't landing and no amount of zero-click framing changes that.
This is also why our guarantee is written against qualified leads from organic, not traffic. Traffic is now a partly Google-controlled variable. Leads aren't.