Between February 4 and September 18, 2026, pages on this site appeared in Google results 10,414,199 times and were clicked 19,948 times, a click rate of 0.19%. We are a small Spanish school with a teaching blog, and we can see the whole ledger, so we are publishing it rather than guessing at it.
Most writing about zero click search is built from panel data or from someone else’s aggregate. This is the view from inside one site’s Search Console: every query, every impression, every click, with the method attached at the bottom so you can check the work or reuse the numbers.
Key takeaways
- 10,414,199 impressions produced 19,948 clicks over 227 days, an overall click-through rate of 0.19%.
- Of the 95,086 individual queries Search Console named for us, 92,541 (97.3%) produced no clicks at all.
- Ranking well did not rescue it. Queries where our average position was between 1.0 and 3.0 converted at 0.51%.
- Seventy-five queries with more than 10,000 impressions each returned under 0.1%. Together they carried 2,174,506 impressions and 581 clicks.
- The pattern is strongest on short translation lookups (“now in spanish”, “morning in spanish”), which is where Google shows its own answer above the results.
The short version
Visibility and traffic have come apart. A small site can hold page-one rankings on queries worth millions of impressions and still collect a rounding error in visits, because the queries it ranks for are the ones the search engine now answers itself. Ranking is no longer the finish line. Being the source the answer is built from is what is left.
What is already known, and what this adds
Zero-click search is well documented at the level of the whole web, and almost undocumented at the level of a single site. SparkToro’s 2026 study, built on Similarweb’s US clickstream panel for January to April 2026, found that 68.01% of Google searches ended without a click, up from 60.45% in 2024. Ahrefs, looking at 300,000 keywords in December 2025, found that an AI Overview correlates with a 58% lower click-through rate for the top-ranking page, with position-one click-through on those keywords falling from 7.3% in December 2023 to 1.6% two years later.
Those studies answer “how often does a search end without a click.” They cannot answer “which of my pages lost what, and on which questions,” because panel data does not know your Search Console. That is the gap this study fills: one site’s full query set, its own click rates by position band, and the specific searches where visibility and traffic came apart. Our top three positions convert at 0.51%, which is below even the AI Overview figure Ahrefs reports, and the reason turns out to be the shape of the question rather than the ranking.
What we measured
We pulled every row Google Search Console holds for the domain property speakbetterspanish.com, from the first day with data (February 4, 2026) to the last complete day at the time of the pull (September 18, 2026). That is 227 days. Site totals come from the unfiltered API response. The query-level analysis uses the 95,086 named queries the API returned, which together carry 7,025,331 impressions, or 67.5% of the site total. The gap is Google’s anonymity threshold, which hides rare queries, and it means the query-level figures describe our measurable demand rather than all of it.
How many of our queries got no clicks at all?
92,541 of 95,086 named queries, or 97.3%, produced zero clicks in seven and a half months. Those silent queries carried 2,298,069 impressions between them. The distribution is not a long tail of junk either. Seventy-five of them cleared 10,000 impressions each on their own.
The clicks we did get are concentrated in a narrow band of queries where someone needs more than a word: “no manches vs no mames”, “difference between tu and su”, “how to use hijole andale and orale”. People click when the answer is a judgment call. They do not click when the answer is a word.
Does ranking at the top still bring traffic?
Not here. On queries where our average position was between 1.0 and 3.0, our click-through rate was 0.51%. The band from 3.0 to 5.0 came in at 0.36%, and 5.0 to 10.0 at 0.09%. Position still helps, and the gradient is real, but it now moves us between fractions of one percent rather than between meaningful amounts of traffic.
| Average position | Queries | Impressions | Clicks | CTR |
|---|---|---|---|---|
| 1.0 to 3.0 | 46,512 | 628,901 | 3,187 | 0.51% |
| 3.0 to 5.0 | 12,928 | 855,628 | 3,110 | 0.36% |
| 5.0 to 10.0 | 15,163 | 4,858,866 | 4,243 | 0.09% |
| 10.0 to 20.0 | 10,327 | 582,347 | 203 | 0.04% |
| 20.0 and lower | 10,156 | 99,589 | 49 | 0.05% |
Read the impressions column next to the click rate and the shape of the problem shows up. Our biggest impression volume sits in positions 5 to 10, where almost nothing converts, and our best click rate sits on low-volume queries in the top three.
The twenty queries that cost us the most
Our largest single loss is “now in spanish”: 156,307 impressions, 45 clicks, average position 5.2. The pattern repeats down the list, and it is almost entirely made of short translation lookups where Google prints a translation widget at the top of the page.
| Query | Impressions | Clicks | CTR | Avg position |
|---|---|---|---|---|
| now in spanish | 156,307 | 45 | 0.029% | 5.2 |
| right in spanish | 111,646 | 17 | 0.015% | 9.5 |
| morning in spanish | 101,686 | 13 | 0.013% | 9.2 |
| nada meaning | 99,794 | 8 | 0.008% | 9.8 |
| me gusta | 94,596 | 29 | 0.031% | 8.5 |
| family in spanish | 90,059 | 31 | 0.034% | 8.6 |
| me gusta meaning | 81,814 | 18 | 0.022% | 7.1 |
| no problem in spanish | 74,577 | 11 | 0.015% | 9.0 |
| nothing in spanish | 65,213 | 12 | 0.018% | 6.7 |
| i know in spanish | 59,043 | 27 | 0.046% | 8.1 |
All seventy-five queries in this group are in the downloadable CSV. Together they account for 2,174,506 impressions and 581 clicks.
Why this is happening, and what we cannot prove
The pattern is consistent with Google answering these queries itself, but our data cannot prove causation. Search Console reports impressions, clicks and position. It does not report what else was on the results page, so we cannot see a translation widget or an AI overview in our own numbers. What we can say is narrower and still useful: the queries that lose almost every click are the ones whose entire answer fits in one line, and the queries that keep their clicks are the ones that need a paragraph.
That split is the finding. It holds across 95,086 queries, and it points somewhere practical, which is the subject of the next study in this series.
What we changed because of it
We stopped treating impression growth as progress. A page that climbs from position 9 to position 5 on “morning in spanish” earns us almost nothing, so it is not worth the work. Our writing now goes toward questions where the reader has to weigh something: two words that look identical, a phrase that is rude in one country and normal in another, a rule that textbooks state and Mexicans ignore. Those are also the questions our own students ask in class, which is a nice coincidence and not an accident.
How we counted this
Source: Google Search Console API, domain property sc-domain:speakbetterspanish.com, pulled September 20, 2026.
Window: February 4, 2026 (first day with data on this property) to September 18, 2026. 227 days.
Site totals: one unfiltered request with no dimensions. 10,414,199 impressions, 19,948 clicks, average position 7.3.
Query set: paginated requests on the query dimension returned 95,086 rows carrying 7,025,331 impressions, which is 67.5% of the site total. The remainder is hidden by Google’s anonymization of rare queries. Position bands use each query’s average position as reported.
Limits: one site, one language-learning niche, one property with seven and a half months of history. Search Console does not report what else appeared on the results page, so every statement here about why the clicks are missing is a hypothesis, labeled as one. Impressions and clicks are Google’s counts, not ours.
Data: CTR by position band (CSV) and the biggest zero-click queries (CSV). Free to reuse with attribution.
Questions people ask about this data
What is a zero click search?
A zero click search is one where the person sees the results page, gets what they needed from it, and never visits a website. In our measurement we use a stricter, checkable version: a query where our site was shown to searchers and received no clicks at all during the period. On 92,541 of our 95,086 named queries, that is what happened.
Does a number one ranking still get traffic in 2026?
On our queries, barely. Where our average position was between 1.0 and 3.0, we collected 3,187 clicks from 628,901 impressions, a rate of 0.51%. We would not generalize that to every industry. For short informational lookups that a search engine can answer in a line, a top ranking is worth a small fraction of what it used to be.
Is this just a small site problem?
Size affects the absolute numbers, not the shape. 10.4 million impressions is a real sample, and the click rates split cleanly by the type of question rather than by the page or the ranking. A larger site in the same niche would collect more clicks from the same tiny percentages.
How can I check these figures?
The two CSV files linked above contain the aggregate tables exactly as the article quotes them. Every number was produced by a script that reads the Search Console API directly, and we reran it the morning this was published. If you find an error, write to hannah@speakbetterspanish.com and we will correct the page and note the change.
Can I cite this study?
Yes, including the tables and the CSV files, with attribution to Speak Better Spanish and a link to this page. No permission needed.
References
- Fishkin, R. In 2026, Less than One Third of Google Searches Still Send a Click. SparkToro, using Similarweb US clickstream data, January to April 2026.
- Law, R. and Guan, X. AI Overviews Reduce Clicks by 58%. Ahrefs, 300,000 keywords, December 2023 compared with December 2025.
- Speak Better Spanish. Research data files, September 20, 2026.
Latest updates
- September 20, 2026, later the same day: Added the background section, the outside references and the data visuals. Figures unchanged.
- September 20, 2026: First publication. All figures pulled fresh from the Search Console API on the same day.
Recent Comments