Zero-click search in South Africa
What 16 months of Google Search Console data from 110 South African websites tells us about clicks, AI Overviews and organic search.
For every 100 times these websites appeared in Google for a South African searcher, about 5 people clicked through. That's 67.3 million clicks from 1.43 billion appearances.
Why we ran this study
Most of what gets written about zero-click search comes from the US and Europe, built on panels of browser data. South African marketers get those numbers second hand and have to guess how much of it applies here.
We look after Google Search Console for more than 100 South African websites, from banks and retailers to manufacturers and schools. That gives us a first-party way to ask the question for this market: when a page shows up in Google for a South African searcher, how often does that person click, and is that changing?
So we pulled every day of data Google still holds for every one of those sites, 19 May 2025 to 28 September 2026, and compared like with like.
What this data can and can't tell you
Search Console shows how often our clients' pages appeared in Google and how often they were clicked. It doesn't show what a searcher did instead, whether they clicked a competitor, an ad or nothing at all. So this report measures clicks per appearance for the sites in the study, which is the part of zero-click search a business can see in its own data and act on.
Every figure is anonymised. No client, website or search term is named, a sector is only reported when at least 5 separate businesses sit in it, and the headline figures describe the typical business rather than a total the biggest sites would dominate.
- Appearance
- one time a page showed in Google's results for a searcher (Search Console calls it an impression)
- Clicks per 100 appearances
- the click rate, written as a count of people
- Typical business
- the median business in the group, with each business's websites combined first
The data behind the report
Every number in this report comes from one archive: daily Search Console data for South African searchers, for 110 websites, for every day Google still holds.
How we built it
Daily Search Console data for South African searchers, for every website, one day at a time.
Every search term labelled branded or non-branded, and by type: question, comparison, buying or navigational.
3 breaks in Google's own counting dated and kept out of the comparisons.
The same weeks a year apart for the typical business, with mobile and desktop read separately.
Each business's main page-one search terms checked for an AI Overview in South African results.
No names or search terms published, and a sector only appears with 5 or more businesses in it.
What we found
These are the 6 numbers that sum up the study, and each one is explained further down with the detail behind it.
19 May to 30 June 2026 against the same weeks in 2025, for the typical business. 82% of businesses fell.
Over the same weeks the typical business appeared 22% more often in Google and got 8% fewer clicks.
People searching for a business by name click through less. Branded appearances held level and branded clicks fell 14%.
4 100 of the 5 850 terms we checked with a South African results page captured from June 2026.
19 May to 30 June 2026 against 2025, for search terms on page one both years. Search terms that show an AI Overview today lost clicks faster for 66% of businesses.
The sector that relies most on branded search, 19 May to 30 June against the same weeks in 2025, with every business in the sector down.
What we set out to test
We went in with 5 working hypotheses, the things most people in the market assume about zero-click search. This is how each one held up against the data.
Down 29% across all devices and 21% on mobile for the typical business, with 91% of businesses down.
Search terms that show an AI Overview today lost 22% of their clicks per appearance in a year, against 9% for terms without one. But branded searches fell 16% too, so AI Overviews explain part of the fall and not all of it.
Question searches get the fewest clicks today, 1.8 per 100 appearances in the top 3. But the biggest fall in a year was in comparison searches, −34%, against −24% for informational ones.
Branded clicks per appearance fell 16% for the typical business, from 19.0 to 15.6 per 100, and 74% of businesses were down.
Clicks per appearance fell in all 7 sectors we can report on, by between 15% and 50% for the typical business.
3 breaks in Google's data
Search Console isn't a stable measuring tool. Google changed how it counted at least 3 times in these 16 months, and any one of those changes would make clicks look as if they had collapsed or recovered.
View the data
| Date | Average position | Clicks per 100 |
|---|---|---|
| 19 May 2025 | 6.7 | 4.67 |
| 26 May 2025 | 6.7 | 4.79 |
| 2 Jun 2025 | 6.8 | 5.25 |
| 9 Jun 2025 | 7.6 | 5.32 |
| 16 Jun 2025 | 7.6 | 4.46 |
| 23 Jun 2025 | 7.4 | 4.62 |
| 30 Jun 2025 | 7.4 | 4.59 |
| 7 Jul 2025 | 18.3 | 3.27 |
| 14 Jul 2025 | 16.8 | 3.20 |
| 21 Jul 2025 | 18.9 | 3.08 |
| 28 Jul 2025 | 17.3 | 3.66 |
| 4 Aug 2025 | 15.3 | 3.47 |
| 11 Aug 2025 | 17.5 | 3.14 |
| 18 Aug 2025 | 15.1 | 3.15 |
| 25 Aug 2025 | 16.1 | 3.59 |
| 1 Sep 2025 | 14.1 | 3.63 |
| 8 Sep 2025 | 16.3 | 3.60 |
| 15 Sep 2025 | 8.0 | 4.24 |
| 22 Sep 2025 | 7.5 | 4.47 |
| 29 Sep 2025 | 7.5 | 4.79 |
| 6 Oct 2025 | 7.3 | 4.41 |
| 13 Oct 2025 | 7.6 | 4.22 |
| 20 Oct 2025 | 7.2 | 4.29 |
| 27 Oct 2025 | 7.5 | 4.84 |
| 3 Nov 2025 | 7.0 | 4.51 |
| 10 Nov 2025 | 7.2 | 4.17 |
| 17 Nov 2025 | 7.1 | 4.22 |
| 24 Nov 2025 | 7.7 | 4.16 |
| 1 Dec 2025 | 8.2 | 3.94 |
| 8 Dec 2025 | 8.1 | 3.78 |
| 15 Dec 2025 | 7.9 | 3.24 |
| 22 Dec 2025 | 11.7 | 2.11 |
| 29 Dec 2025 | 7.8 | 2.98 |
| 5 Jan 2026 | 7.8 | 3.54 |
| 12 Jan 2026 | 8.2 | 3.44 |
| 19 Jan 2026 | 7.9 | 3.25 |
| 26 Jan 2026 | 8.4 | 3.00 |
| 2 Feb 2026 | 7.9 | 3.40 |
| 9 Feb 2026 | 7.8 | 3.45 |
| 16 Feb 2026 | 7.0 | 2.81 |
| 23 Feb 2026 | 7.5 | 3.06 |
| 2 Mar 2026 | 7.4 | 3.22 |
| 9 Mar 2026 | 6.1 | 3.27 |
| 16 Mar 2026 | 6.4 | 2.79 |
| 23 Mar 2026 | 5.9 | 3.05 |
| 30 Mar 2026 | 6.2 | 3.31 |
| 6 Apr 2026 | 6.3 | 3.08 |
| 13 Apr 2026 | 6.5 | 3.17 |
| 20 Apr 2026 | 6.9 | 3.53 |
| 27 Apr 2026 | 7.7 | 3.14 |
| 4 May 2026 | 7.5 | 3.61 |
| 11 May 2026 | 7.8 | 3.81 |
| 18 May 2026 | 7.0 | 3.33 |
| 25 May 2026 | 6.8 | 3.29 |
| 1 Jun 2026 | 6.8 | 3.47 |
| 8 Jun 2026 | 6.5 | 3.44 |
| 15 Jun 2026 | 6.7 | 3.26 |
| 22 Jun 2026 | 7.3 | 3.55 |
| 29 Jun 2026 | 7.6 | 3.57 |
| 6 Jul 2026 | 7.4 | 3.34 |
| 13 Jul 2026 | 7.4 | 3.59 |
| 20 Jul 2026 | 6.5 | 3.73 |
| 27 Jul 2026 | 7.0 | 3.74 |
| 3 Aug 2026 | 7.7 | 3.33 |
| 10 Aug 2026 | 9.4 | 2.79 |
| 17 Aug 2026 | 7.7 | 2.51 |
| 24 Aug 2026 | 7.2 | 3.91 |
| 31 Aug 2026 | 7.9 | 4.23 |
| 7 Sep 2026 | 6.8 | 3.57 |
| 14 Sep 2026 | 6.8 | 3.38 |
| 21 Sep 2026 | 6.7 | 3.43 |
| 28 Sep 2026 | 6.1 | 3.65 |
View the data
| Date | Share at position 1 |
|---|---|
| 19 May 2025 | 9.8% |
| 26 May 2025 | 9.6% |
| 2 Jun 2025 | 9.7% |
| 9 Jun 2025 | 9.8% |
| 16 Jun 2025 | 9.7% |
| 23 Jun 2025 | 10.0% |
| 30 Jun 2025 | 9.8% |
| 7 Jul 2025 | 8.7% |
| 14 Jul 2025 | 8.5% |
| 21 Jul 2025 | 8.3% |
| 28 Jul 2025 | 7.8% |
| 4 Aug 2025 | 8.3% |
| 11 Aug 2025 | 8.4% |
| 18 Aug 2025 | 8.7% |
| 25 Aug 2025 | 8.1% |
| 1 Sep 2025 | 8.3% |
| 8 Sep 2025 | 9.5% |
| 15 Sep 2025 | 10.3% |
| 22 Sep 2025 | 10.0% |
| 29 Sep 2025 | 9.3% |
| 6 Oct 2025 | 9.6% |
| 13 Oct 2025 | 9.5% |
| 20 Oct 2025 | 9.7% |
| 27 Oct 2025 | 9.8% |
| 3 Nov 2025 | 11.0% |
| 10 Nov 2025 | 10.4% |
| 17 Nov 2025 | 10.0% |
| 24 Nov 2025 | 9.5% |
| 1 Dec 2025 | 10.3% |
| 8 Dec 2025 | 11.1% |
| 15 Dec 2025 | 12.4% |
| 22 Dec 2025 | 11.9% |
| 29 Dec 2025 | 14.2% |
| 5 Jan 2026 | 14.4% |
| 12 Jan 2026 | 14.4% |
| 19 Jan 2026 | 17.4% |
| 26 Jan 2026 | 19.7% |
| 2 Feb 2026 | 20.4% |
| 9 Feb 2026 | 20.2% |
| 16 Feb 2026 | 22.8% |
| 23 Feb 2026 | 22.7% |
| 2 Mar 2026 | 22.4% |
| 9 Mar 2026 | 23.4% |
| 16 Mar 2026 | 22.5% |
| 23 Mar 2026 | 20.7% |
| 30 Mar 2026 | 23.4% |
| 6 Apr 2026 | 27.6% |
| 13 Apr 2026 | 23.5% |
| 20 Apr 2026 | 18.8% |
| 27 Apr 2026 | 10.1% |
| 4 May 2026 | 11.7% |
| 11 May 2026 | 11.5% |
| 18 May 2026 | 7.6% |
| 25 May 2026 | 8.0% |
| 1 Jun 2026 | 7.0% |
| 8 Jun 2026 | 7.5% |
| 15 Jun 2026 | 7.4% |
| 22 Jun 2026 | 7.8% |
| 29 Jun 2026 | 8.2% |
| 6 Jul 2026 | 8.0% |
| 13 Jul 2026 | 9.0% |
| 20 Jul 2026 | 8.3% |
| 27 Jul 2026 | 11.6% |
| 3 Aug 2026 | 11.5% |
| 10 Aug 2026 | 9.2% |
| 17 Aug 2026 | 11.6% |
| 24 Aug 2026 | 9.1% |
| 31 Aug 2026 | 8.2% |
| 7 Sep 2026 | 8.0% |
| 14 Sep 2026 | 6.9% |
| 21 Sep 2026 | 6.5% |
| 28 Sep 2026 | 7.3% |
Overnight, the typical site's average position moved from about 7.5 to about 20, and its click rate fell from about 4.6 to about 3 clicks per 100 appearances. The extra appearances sit deep in the results and bring no clicks, which points to automated tools loading results pages.
It reversed when Google stopped serving 100 results a page to rank-tracking tools. Average position went back to about 7 to 8 and has stayed there.
Google reported a growing share of non-branded appearances at position 1, from about 10% to 24%, taken from positions 3 to 10 while total appearances stayed flat. It unwound between 16 and 28 April 2026. We don't know the cause.
Appearances are up and clicks are down
We index every business to its own daily average for 19 May to 30 June 2025, then take the typical business each month. A large website and a small one count the same, so the lines show direction rather than size.
View the data
| Month | Appearances | Clicks | Clicks per 100 |
|---|---|---|---|
| May 2025 | 102 | 104 | 4.92 |
| Jun 2025 | 99 | 98 | 5.11 |
| Sep 2025 | 126 | 103 | 4.25 |
| Oct 2025 | 123 | 106 | 4.34 |
| Nov 2025 | 138 | 103 | 3.70 |
| Dec 2025 | 134 | 79 | 3.06 |
| Jan 2026 | 157 | 100 | 3.35 |
| Feb 2026 | 168 | 107 | 3.20 |
| Mar 2026 | 154 | 102 | 2.70 |
| Apr 2026 | 134 | 101 | 3.24 |
| May 2026 | 132 | 98 | 2.90 |
| Jun 2026 | 128 | 89 | 3.29 |
| Jul 2026 | 125 | 98 | 3.52 |
| Aug 2026 | 132 | 90 | 3.36 |
| Sep 2026 | 118 | 87 | 3.62 |
From November 2025 appearances climbed, reaching about 1.7 times the starting level by February 2026 before settling at about 1.3 times through the winter of 2026. Clicks stayed close to where they started for most of the year and drifted down to about 0.9 times by August and September 2026.
On mobile the climb was smaller and has mostly unwound. By September 2026 mobile appearances were back near the starting level and mobile clicks were about 8% below it.
Clicks per appearance fell for 9 in 10 businesses
The cleanest test we have is the same 6 weeks a year apart, 19 May to 30 June, with both years clear of Google's counting changes.
View the data
| 2025 | 2026 | Typical change | |
|---|---|---|---|
| All devices | 4.77 | 3.03 | −29% |
| Mobile | 3.98 | 3.1 | −21% |
| Desktop | 6.33 | 3.27 | −49% |
| Non-branded, page one | 1.96 | 1.27 | −31% |
| Branded | 19.04 | 15.59 | −16% |
| Appearances | Clicks | |
|---|---|---|
| All devices | +38% | −9% |
| Mobile | +22% | −8% |
| Desktop | +62% | −18% |
| Branded | 0% | −14% |
Typical business. Desktop appearances have run high since September 2025 in a way that looks like automated traffic, which never clicks.
Desktop shows the steepest fall, but desktop appearances have run high since September 2025 and the extra appearances don't bring clicks with them. About 4 in 5 appearances in this study are on mobile, so we lead with mobile.
The second window, 15 to 28 September, tells the same story on mobile: appearances −6%, clicks −17% and clicks per appearance −13% for the typical business.
Branded searches, where someone types the business's own name, still get far more clicks than anything else, 15.6 per 100 appearances in 2026. They are also falling: down 16% for the typical business, with 74% of businesses down.
What a page-one ranking is worth today
Clicks per 100 appearances on non-branded searches, June to August 2026, after Google's position reporting settled.
View the data
| Position | Typical business | Middle half of businesses | All websites combined | Businesses |
|---|---|---|---|---|
| Position 1 | 3.54 | 2.21 to 6.30 | 8.20 | 44 |
| 2 | 3.06 | 2.16 to 4.34 | 5.25 | 39 |
| 3 | 1.97 | 1.54 to 3.11 | 3.18 | 41 |
| 4 to 5 | 1.46 | 0.85 to 1.93 | 2.08 | 50 |
| 6 to 10 | 0.68 | 0.35 to 0.97 | 0.92 | 57 |
| 11 to 20 | 0.39 | 0.20 to 0.62 | 0.70 | 56 |
View the data
| Type of search | Clicks per 100 | Businesses |
|---|---|---|
| Questions | 1.77 | 31 |
| Comparison and buying | 3.41 | 31 |
| Other non-branded | 3.35 | 49 |
Branded searches are 13% of appearances and 75% of clicks for the typical business. Search Console only names the search terms enough people use, and branded terms are more often among them, so the real branded share of clicks is probably a little lower than this.
Question searches get about half the clicks of comparison and buying searches in the top 3, 1.8 against 3.4 per 100. Longer searches of 5 words or more get 2.5 per 100, against 3.1 for 1 or 2 words.
How often AI Overviews show
We took each business's main non-branded search terms where it ranked on page one between June and September 2026, up to 300 per business, and checked each one against a South African Google results page captured from June 2026 onwards.
AI Overviews show on buying and comparison searches about as often as on informational ones, 75% and 77% against 75%. Navigational searches, where someone is looking for a particular site, show one far less often, 35%.
View the data
| Stage | With an AI Overview | Terms |
|---|---|---|
| Learn (informational) | 75.4% | 2034 |
| Buy (transactional) | 74.5% | 2781 |
| Compare (commercial) | 77.0% | 871 |
| Navigate (navigational) | 34.9% | 627 |
View the data
| Sector | With an AI Overview | Businesses |
|---|---|---|
| Financial services | 78.3% | 9 |
| FMCG & beauty | 97.0% | 5 |
| Retail: home, DIY & furniture | 61.8% | 6 |
| Retail: fashion, health & general | 63.0% | 6 |
| Industrial & B2B services | 69.4% | 9 |
| Automotive | not reported | |
| Leisure, education & property | not reported |
Search terms with an AI Overview lost clicks faster
We compared the same search terms a year apart, 19 May to 30 June, for terms on page one in both years, split by whether they show an AI Overview today.
View the data
| 2025 | 2026 | Typical change | Average position 2025 | Average position 2026 | |
|---|---|---|---|---|---|
| AI Overview today | 1.56 | 1.35 | −22% | 6.2 | 5.6 |
| No AI Overview today | 1.4 | 1.33 | −9% | 5.9 | 5.8 |
View the data
| Position | With (typical) | Without (typical) | Within-business ratio | With (all combined) | Without (all combined) | Businesses |
|---|---|---|---|---|---|---|
| Position 1 | 0.96 | 4.17 | 0.73 | 6.48 | 20.43 | 12 |
| Positions 2 to 3 | 2.78 | 3.36 | 1.06 | 3.14 | 3.44 | 27 |
| Positions 4 to 10 | 0.73 | 0.76 | 0.85 | 1.14 | 0.92 | 46 |
At position 1 the gap is large however we measure it. With all businesses combined it's 6.5 clicks per 100 with an AI Overview against 20.4 without, about two thirds fewer. Within the typical business it's about a quarter fewer, a ratio of 0.73. Only 12 businesses rank first often enough on both kinds of term to compare, so read the size of the gap as a range.
At positions 2 to 10 we see no consistent gap. Terms with and without an AI Overview get about the same clicks there.
Clicks per appearance fell at every stage of the funnel
We grouped the search terms we checked into 4 stages using DataForSEO's search intent labels, and added branded searches from the full archive as a fifth.
Changes are 19 May to 30 June 2026 against the same weeks in 2025, for the typical business, on search terms that were on page one both years. *Branded searches: clicks per 100 across all positions in 2026, from the full archive. AI Overview share isn't measured for branded searches.
Comparison searches had the biggest fall in clicks per appearance, −34%. Their clicks held up (+5%) because they appeared 39% more often, so the extra appearances brought very few extra clicks.
Learn and navigational searches lost the most clicks, −18% each. Navigational searches are also the only stage where AI Overviews are uncommon, at 35%.
Branded searches still turn appearances into clicks far better than any other stage, and they carry about three quarters of the clicks for the typical business, so the 16% fall there matters most.
Every sector we can report on lost clicks per appearance
7 sectors have at least 5 businesses in them. Telecoms has 2, so it counts towards the overall figures but isn't reported on its own.
| Sector | Businesses | Websites | Clicks per 100, clean year | Change in clicks per appearance | Appearances | Clicks | Terms with an AI Overview |
|---|---|---|---|---|---|---|---|
| Financial services | 9 | 12 | 4.7 | −37% | +44% | −20% | 78% |
| FMCG & beauty | 6 | 14 | 2.1 | −33% | +60% | +8% | 97% |
| Retail: home, DIY & furniture | 8 | 9 | 2.4 | −29% | +47% | +13% | 62% |
| Retail: fashion, health & general | 7 | 9 | 3.8 | −15% | −8% | +3% | 63% |
| Industrial & B2B services | 25 | 33 | 2.9 | −29% | +48% | −10% | 69% |
| Automotive | 5 | 19 | 1.8 | −15% | +5% | −18% | not reported |
| Leisure, education & property | 9 | 11 | 2.6 | −50% | +47% | −23% | not reported |
Typical business in each sector. Changes are 19 May to 30 June 2026 against the same weeks in 2025. Clean year is October 2025 to August 2026.
The falls run from 15% in fashion, health and general retail and in automotive to 50% in leisure, education and property. Financial services, the sector that relies most on branded search, fell 37%.
Each sector has a page below with its own monthly trend, what a ranking is worth there, and what it means.
Financial services: clicks per appearance down 37% in a year
Financial services websites get more clicks per appearance than any other sector in the study, 4.7 per 100 across the clean year, and most of those clicks come from people searching for the brand by name. Branded searches bring 88% of clicks for the typical business, the highest share of any sector.
Over 19 May to 30 June the typical business appeared 44% more often in Google and got 20% fewer clicks, and every business in the sector lost clicks per appearance. Branded search took the biggest hit: branded clicks fell 28% while branded appearances rose 6%.
Non-branded searches fell too: clicks per appearance on page one dropped 34%. In the top 3, question searches bring 1.5 clicks per 100 against 2.5 for comparison and buying searches, and AI Overviews show on 78% of the non-branded page-one terms we checked.
View the data
| Month | Appearances | Clicks | Clicks per 100 |
|---|---|---|---|
| May 2025 | 103 | 102 | 8.55 |
| Jun 2025 | 99 | 99 | 8.69 |
| Sep 2025 | 163 | 103 | 5.43 |
| Oct 2025 | 165 | 107 | 5.16 |
| Nov 2025 | 198 | 95 | 4.17 |
| Dec 2025 | 182 | 65 | 3.22 |
| Jan 2026 | 209 | 94 | 3.88 |
| Feb 2026 | 214 | 105 | 4.15 |
| Mar 2026 | 212 | 108 | 4.50 |
| Apr 2026 | 175 | 97 | 4.40 |
| May 2026 | 162 | 79 | 4.92 |
| Jun 2026 | 140 | 77 | 5.70 |
| Jul 2026 | 140 | 76 | 5.11 |
| Aug 2026 | 137 | 64 | 5.50 |
| Sep 2026 | 118 | 60 | 3.97 |
View the data
| Position | Financial services | All businesses |
|---|---|---|
| 1 | 4.02 | 3.54 |
| 2 | 2.93 | 3.06 |
| 3 | 1.63 | 1.97 |
| 4 to 5 | 1.39 | 1.46 |
| 6 to 10 | 0.40 | 0.68 |
| 11 to 20 | 0.37 | 0.39 |
FMCG & beauty: AI Overviews on almost every search we checked
AI Overviews show on 97% of the non-branded page-one search terms we checked for the typical business, the highest share of any sector.
It's also the sector where appearances grew fastest. Over 19 May to 30 June the typical business appeared 60% more often and got 8% more clicks, so clicks per appearance fell 33%, with every business in the sector down. By September 2026 appearances were about twice their level of a year earlier.
Position 1 on a non-branded search brings 2.2 clicks per 100 appearances for the typical business, the lowest of any sector we can report on, and positions 6 to 10 bring 0.3. Branded clicks fell 17% while branded appearances rose 8%.
View the data
| Month | Appearances | Clicks | Clicks per 100 |
|---|---|---|---|
| May 2025 | 105 | 105 | 2.90 |
| Jun 2025 | 98 | 98 | 2.76 |
| Sep 2025 | 111 | 119 | 4.21 |
| Oct 2025 | 123 | 113 | 3.10 |
| Nov 2025 | 136 | 105 | 3.36 |
| Dec 2025 | 142 | 97 | 2.41 |
| Jan 2026 | 168 | 99 | 2.03 |
| Feb 2026 | 168 | 85 | 1.80 |
| Mar 2026 | 218 | 102 | 1.35 |
| Apr 2026 | 178 | 104 | 1.73 |
| May 2026 | 143 | 110 | 2.43 |
| Jun 2026 | 160 | 108 | 2.15 |
| Jul 2026 | 174 | 125 | 2.02 |
| Aug 2026 | 185 | 113 | 1.71 |
| Sep 2026 | 207 | 127 | 1.72 |
View the data
| Position | FMCG & beauty | All businesses |
|---|---|---|
| 1 | 2.23 | 3.54 |
| 2 | not reported | 3.06 |
| 3 | not reported | 1.97 |
| 4 to 5 | 0.69 | 1.46 |
| 6 to 10 | 0.26 | 0.68 |
| 11 to 20 | 0.15 | 0.39 |
Retail: home, DIY & furniture: more clicks, on far more appearances
Home, DIY and furniture retail runs on non-branded search more than any other sector. Branded searches bring 31% of clicks for the typical business, against 75% across the whole study, so the click rate on product and category searches decides how this sector does.
Over 19 May to 30 June the typical business appeared 47% more often and got 13% more clicks, which still meant clicks per appearance fell 29%, with every business in the sector down. The September window shows the same direction, down 28%.
Comparison and buying searches are worth the most here, 4.1 clicks per 100 in the top 3 against 1.7 for question searches. AI Overviews show on 62% of the non-branded page-one terms we checked.
View the data
| Month | Appearances | Clicks | Clicks per 100 |
|---|---|---|---|
| May 2025 | 101 | 101 | 3.66 |
| Jun 2025 | 100 | 100 | 3.67 |
| Sep 2025 | 126 | 106 | 3.81 |
| Oct 2025 | 124 | 110 | 3.60 |
| Nov 2025 | 142 | 135 | 3.27 |
| Dec 2025 | 146 | 107 | 3.04 |
| Jan 2026 | 192 | 135 | 3.03 |
| Feb 2026 | 193 | 129 | 2.80 |
| Mar 2026 | 180 | 130 | 2.53 |
| Apr 2026 | 164 | 121 | 2.71 |
| May 2026 | 157 | 121 | 2.82 |
| Jun 2026 | 143 | 112 | 2.73 |
| Jul 2026 | 147 | 118 | 2.77 |
| Aug 2026 | 118 | 86 | 2.73 |
| Sep 2026 | 118 | 87 | 2.58 |
View the data
| Position | Retail: home, DIY & furniture | All businesses |
|---|---|---|
| 1 | 4.19 | 3.54 |
| 2 | 3.98 | 3.06 |
| 3 | 3.10 | 1.97 |
| 4 to 5 | 2.00 | 1.46 |
| 6 to 10 | 0.97 | 0.68 |
| 11 to 20 | 0.71 | 0.39 |
Retail: fashion, health & general: the sector that held up best
Over 19 May to 30 June the typical business appeared 8% less often and got 3% more clicks. Clicks per appearance fell 15%, the smallest fall in the study alongside automotive, and 3 in 5 businesses were down.
On mobile the fall was 7%, and the September window shows almost no change (down 1%). Branded clicks held level.
Position 1 on a non-branded search is worth more here than anywhere else in the study, 6.1 clicks per 100 appearances for the typical business. AI Overviews show on 63% of the non-branded page-one terms we checked.
View the data
| Month | Appearances | Clicks | Clicks per 100 |
|---|---|---|---|
| May 2025 | 102 | 103 | 3.87 |
| Jun 2025 | 99 | 99 | 3.77 |
| Sep 2025 | 100 | 126 | 4.86 |
| Oct 2025 | 103 | 129 | 4.78 |
| Nov 2025 | 128 | 175 | 4.07 |
| Dec 2025 | 112 | 126 | 3.80 |
| Jan 2026 | 117 | 98 | 3.78 |
| Feb 2026 | 108 | 102 | 3.64 |
| Mar 2026 | 117 | 108 | 3.38 |
| Apr 2026 | 108 | 105 | 3.62 |
| May 2026 | 97 | 108 | 3.78 |
| Jun 2026 | 90 | 103 | 3.59 |
| Jul 2026 | 94 | 120 | 3.67 |
| Aug 2026 | 98 | 137 | 4.62 |
| Sep 2026 | 89 | 169 | 5.79 |
View the data
| Position | Retail: fashion, health & general | All businesses |
|---|---|---|
| 1 | 6.09 | 3.54 |
| 2 | 2.75 | 3.06 |
| 3 | 1.90 | 1.97 |
| 4 to 5 | 1.33 | 1.46 |
| 6 to 10 | 0.85 | 0.68 |
| 11 to 20 | 0.57 | 0.39 |
Industrial & B2B services: the biggest fall on non-branded page-one searches
The biggest group in the study: 25 businesses across manufacturing, equipment, facilities, recruitment, software and professional services. Most of their search clicks are branded, 83% for the typical business.
AI Overviews show on 69% of the non-branded page-one terms we checked, and non-branded page one had the biggest fall in clicks per appearance of any sector, 39%. Over 19 May to 30 June the typical business appeared 48% more often and got 10% fewer clicks.
The September window is the exception: clicks per appearance were up 9% and only a third of businesses were down, so the trend in this sector has been uneven.
View the data
| Month | Appearances | Clicks | Clicks per 100 |
|---|---|---|---|
| May 2025 | 103 | 108 | 6.92 |
| Jun 2025 | 99 | 96 | 6.18 |
| Sep 2025 | 130 | 92 | 4.37 |
| Oct 2025 | 139 | 104 | 4.68 |
| Nov 2025 | 136 | 90 | 3.96 |
| Dec 2025 | 111 | 61 | 2.92 |
| Jan 2026 | 155 | 99 | 3.16 |
| Feb 2026 | 166 | 117 | 3.61 |
| Mar 2026 | 155 | 102 | 3.54 |
| Apr 2026 | 134 | 91 | 3.64 |
| May 2026 | 127 | 99 | 3.95 |
| Jun 2026 | 119 | 88 | 4.25 |
| Jul 2026 | 154 | 100 | 4.98 |
| Aug 2026 | 156 | 90 | 4.25 |
| Sep 2026 | 118 | 97 | 4.60 |
View the data
| Position | Industrial & B2B services | All businesses |
|---|---|---|
| 1 | 5.67 | 3.54 |
| 2 | 3.41 | 3.06 |
| 3 | 1.54 | 1.97 |
| 4 to 5 | 0.88 | 1.46 |
| 6 to 10 | 0.44 | 0.68 |
| 11 to 20 | 0.35 | 0.39 |
Automotive: the fewest clicks per appearance in the study
Across the clean year the typical automotive business got 1.8 clicks per 100 appearances, against 2.9 for the typical business in the whole study.
Over 19 May to 30 June the typical business appeared 5% more often and got 18% fewer clicks, and every business in the sector lost clicks per appearance. On mobile, appearances were level and clicks fell 14%.
Only 4 of the 5 businesses have enough data for the deeper tables, below our minimum of 5, so we leave AI Overview, branded and position figures out for this sector.
Leisure, education & property: the biggest fall of any sector
A mixed group: entertainment, travel, education and property. Read it as a direction rather than a benchmark for any one business.
Clicks per appearance fell 50% for the typical business over 19 May to 30 June, with every business down. The typical business appeared 47% more often and got 23% fewer clicks, and the September window shows a similar fall, 45%.
Much of it is branded: branded appearances fell 45% and branded clicks 59% for the typical business. Too few businesses here had search terms checked for AI Overviews to report a figure.
View the data
| Month | Appearances | Clicks | Clicks per 100 |
|---|---|---|---|
| May 2025 | 102 | 107 | 5.05 |
| Jun 2025 | 99 | 97 | 7.41 |
| Sep 2025 | 129 | 91 | 2.90 |
| Oct 2025 | 121 | 93 | 3.00 |
| Nov 2025 | 130 | 90 | 2.52 |
| Dec 2025 | 128 | 72 | 2.11 |
| Jan 2026 | 225 | 93 | 2.32 |
| Feb 2026 | 237 | 80 | 2.02 |
| Mar 2026 | 154 | 68 | 2.25 |
| Apr 2026 | 141 | 90 | 2.36 |
| May 2026 | 132 | 90 | 2.67 |
| Jun 2026 | 136 | 73 | 2.44 |
| Jul 2026 | 110 | 78 | 1.91 |
| Aug 2026 | 127 | 65 | 1.99 |
| Sep 2026 | 95 | 58 | 1.97 |
What this means for organic search
These are 6 things we're changing in how we plan and report organic search for our clients, and any business can apply them to its own data.
A click rate on its own can't tell you whether you lost clicks or gained visibility. Most businesses in this study did both: over the same 6 weeks the typical business appeared 38% more often and got 9% fewer clicks.
Search terms that show an AI Overview lost clicks per appearance faster than terms without one, −22% against −9%, and position 1 under an AI Overview gets fewer clicks than position 1 without one. We track whether our clients are cited inside AI answers next to their rankings, and report both.
Branded searches bring about three quarters of clicks for the typical business, and branded clicks per appearance fell 16% in a year. What Google shows above the website for its own name is worth checking as often as any non-branded ranking.
Question searches bring 1.8 clicks per 100 appearances in the top 3, about half the rate of comparison and buying searches. Content that answers questions still earns visibility, so it needs to be judged on appearances and AI citations as well as clicks.
Comparison searches lost the most clicks per appearance, −34%, while appearing 39% more often, and 77% of them show an AI Overview. Comparison pages need clear, specific answers that an AI Overview can quote and link to.
Any comparison that crosses 1 July to 14 September 2025 will mislead, and so will a comparison by position across December 2025 to May 2026. Compare the same weeks a year apart and read mobile on its own.
What zero-click looks like in South Africa
Across 110 websites and 1.43 billion appearances, the pattern is consistent. South African businesses are showing up in Google more than they were a year ago and getting fewer clicks from it. On mobile, the typical business appeared 22% more often and got 8% fewer clicks over the same 6 weeks, and clicks per appearance fell for more than 8 in 10 businesses.
AI Overviews are a large part of it. They show on 70% of the non-branded page-one search terms we checked, and the terms that carry one lost clicks faster than the terms that don't.
Terms without an AI Overview lost clicks too, branded searches are losing clicks, and every sector we can report on is down, so AI Overviews explain part of the change and not all of it.
Organic search still delivered 67.3 million clicks to these websites over the 16 months. What has changed is how many appearances it takes to earn one, and that's what we now plan and report against.
How we did it, and the limits
- Data source
- Google Search Console Search Analytics for web search, filtered to searchers in South Africa, read through Algorithm's own Search Console access. One property per website: where a website had several, the one with the most South African appearances.
- Websites
- 110 websites run by 71 businesses in 8 sectors, which is every website we could read with any South African data. Each business's websites are combined before any figure is worked out.
- Window
- 19 May 2025 to 28 Sep 2026, 498 days. That's every day Google still held when we pulled the data on 1 October 2026; Google keeps 16 months.
- Search terms
- Search Console only names the search terms enough people use. For the typical business about 65% of appearances come with a named search term. Site totals include everything, so headline figures come from totals and search-term figures are used for the splits.
- Branded searches
- Matched against each business's brand and product names, including common misspellings, plus login and portal searches.
- Types of search
- Questions start with how, what, why and similar words. Comparison and buying searches contain words like price, best, buy, review or near me. Funnel stages use DataForSEO's search intent labels: informational, commercial, transactional and navigational.
- Typical business
- The median business. A business only counts towards a figure with at least 1 000 appearances in it (500 for the same-weeks comparisons and 300 per group for the AI Overview comparisons).
- Same weeks a year apart
- 19 May to 30 June and 15 to 28 September, 2025 against 2026, worked out per day so the windows match.
- Breaks in Google's data
- 3, described in section 05: 1 July to 14 September 2025 is left out of every comparison, and comparisons by position avoid December 2025 to May 2026.
- AI Overviews
- DataForSEO Labs Keyword Overview, location South Africa, language English, with results page data. Up to 300 non-branded page-one search terms per business, each with at least 30 appearances from June to September 2026. 11 811 terms checked, 9 419 in DataForSEO's database, 5 850 with a results page captured from June 2026 onwards; older captures are left out.
- Anonymity
- No business, website or search term is named. A sector is only reported with at least 5 businesses in it, headline figures are the typical business rather than totals, and sector figures are rates and changes, never volumes.
Limits
- These are the websites we work on, not every South African search, and the mix leans towards the sectors we work in.
- Appearances aren't searches. When 2 of these websites show on the same results page, that's 2 appearances from 1 search.
- Search Console doesn't show what a searcher did instead of clicking, so this report can't say how many searches ended with no click at all.
- Desktop appearances look inflated by automated traffic since September 2025, which is why we lead with mobile.
- Sector groups are small, from 5 to 25 businesses, so sector figures show direction rather than precise benchmarks.
- The AI Overview check is one results page per search term, not a history, and only covers terms in DataForSEO's database.
- Google changed its counting at least 3 times in the window. We've kept comparisons clear of each change, but there may be smaller ones we haven't caught.
Sources
- Google Search Console API, Search Analytics, web search, South Africa
- DataForSEO Labs API, Keyword Overview, South Africa, English
- Analysis by Algorithm, October 2026