The Meerkat Index, edition 1 (cars). Method and corrections. Observations 6 and 7 November 2026. Page last updated 10 November 2026.
The Meerkat Index records two things. The first is who shows up on page 1 of Google for a fixed list of customer searches. The second is what happens when you open a company’s website on a phone.
RhinoBerry publishes the Index. Cobus van Vuuren runs the checks. On this page, “we” means the two together. It explains how we do both, what we leave out, and how to tell us we got something wrong.
What does the Index measure?
Two boards, kept apart, with no combined score. A company can lead one board and trail on the other. We’d rather show you that than blend it away.
Board 1: sampled search visibility
We ran 30 fixed customer searches on Google, set to South Africa, desktop and English, on two days: 6 and 7 November 2026. For each company we count how many of the 30 put its own website on page 1 on both days. We call this sampled visibility. It isn’t market share. Companies are ranked within their own peer group, and equal results share a rank, shown with “=”.
Board 2: automated phone-size checks
On a phone-sized screen our script follows the same three steps on every site: the home page, one offer page and one contact action. We record whether each step loads and shows its content. We don’t submit an enquiry or check that one would arrive. This board isn’t ranked. Sites are listed by steps passed, then alphabetically.
Context, never part of a rank
Beside the boards we show estimated monthly search visits from DataForSEO. These are modelled figures, not the companies’ own numbers. DataForSEO builds them from the searches a site already ranks for and how often those are searched, so they aren’t independent of Board 1.
We also show Lighthouse’s mobile speed score for each home page and a short evidence panel. For the ten sites in the hub’s two rounds, we add a phone picture of the first screen. The panel lists dated facts from the tested pages. That covers analytics and ad tags, contact links (tap-to-call, WhatsApp and forms), business structured data and robots.txt rules for AI crawlers.
What doesn’t it measure?
We observe selected public search results and website journeys. We do not measure actual visitors, leads, conversion rates, advertising returns, sales or revenue.
Reviews, backlinks, mentions in AI assistants and social media counts are out of edition 1. We never submit fake leads, call a sales team or buy anything to test a site.
Who is in edition 1?
Twenty brands in two peer groups. Search ranks are given within each group. Some descriptive figures, like the page-1 shares and the audience correlation, combine all 20.
| Peer group | Brands |
|---|---|
| Marketplaces and classifieds (7) | AutoTrader; Auto Mart; CarFind; Cars.co.za; ChangeCars; Surf4Cars; Truck & Trailer |
| Dealer groups and independent dealers (13) | Auto Emporium; Auto Pedigree; Bidvest McCarthy; CFAO Mobility; CMH Group; Fuzion Motor Group; Hatfield Motor Group; King Cars; NMI Motor Group; Rola Motor Group; SCW Mobility Group; SMH Group; Westvaal Motor Group |
They come from the list of companies RhinoBerry tracks in cars and mobility, and each website serves customers in South Africa. That’s not every car business in the country. We chose the first brand list and the searches before collecting any data. During review we corrected the brand groups, how subdomains count and a duplicated second sample. Every change is dated below.
Which 30 searches did we use, and how were they chosen?
RhinoBerry drafted the list, and Cobus approved it before any search ran. The rules were simple. Use searches a customer types, with no brand name of a company we rank and no “near me”. Name an area where it fits. It’s a list we chose, weighted equally, not a sample of how often people search for each thing. The list stays the same in later editions, so changes compare like with like.
The mix matters when you read the boards. Sixteen searches are about buying used, six about selling or valuing, five about dealers or new cars, and three about finance and other routes. When we drew it up, we expected the used-car searches to suit marketplaces and the dealer searches to suit dealer groups. Keep that in mind when you read the dealer board.
| Customer need | Searches |
|---|---|
| Selling or valuing a car (6) | sell my car online; sell my car for cash; how much is my car worth; car valuation south africa; trade in my car; sell my car johannesburg |
| Buying used, by place (8) | used cars for sale johannesburg; used cars for sale cape town; used cars for sale pretoria; used cars for sale durban; used cars for sale bloemfontein; used cars for sale port elizabeth; second hand cars for sale; buy a car online south africa |
| Buying used, by type and budget (8) | used bakkies for sale; used suv for sale; cheap cars for sale; used cars under r100000; used cars under r200000; used toyota hilux for sale; used ford ranger for sale; used vw polo for sale |
| Dealers and new (5) | car dealerships johannesburg; car dealers cape town; new cars for sale south africa; certified pre-owned cars south africa; toyota dealer pretoria |
| Finance and other routes (3) | car finance south africa; bank repossessed cars for sale; car for sale by owner |
How does the search board count a result?
Only ordinary (organic) results on page 1 count. Ads, map results and AI answers aren’t counted on this board. “Page 1” means the ordinary results on Google’s first page, which doesn’t always hold ten. We use “page 1” everywhere in the Index, not “top 10”.
Each search was run on two days, 6 and 7 November. A company counts only if its own website was on page 1 on both days. If it was there on one day and not the other, that search doesn’t count for it.
We use two days because a single day overstates how settled page 1 is. Between 6 and 7 November, no search kept the same page 1, even where a brand’s daily total barely moved. On 6 November there were 200 pairs of a website and a search on page 1, leaving out Facebook and Reddit posts and counting a brand’s subdomains as the brand. On 7 November, 88 of those pairs (44%) were on page 1 for the same search. So 56% changed hands between the two days.
A Facebook or Reddit post never counts for a company’s website, including a result that links to facebook.com. A subdomain of a company’s own website counts for that company. A separately registered domain never counts, even with a similar name. So dealer groups whose dealerships run separately registered websites are counted on their main site only. That’s a real limit. The count may understate a group’s combined visibility where separately registered dealership sites also show up.
How are the page-1 shares on the hub worked out?
The shares are given for each day. Page 1 doesn’t always hold ten ordinary results, so the 30 searches gave 274 page-1 results on 6 November and 272 on 7 November, not 300. One site can hold more than one spot on the same search, so the shares count spots, not searches. On 6 November, 223 of the 274 were matched to a website from the result’s address line, and 4 of those were facebook.com, which we count as posts. One came with its website given directly, and 50 were Facebook or Reddit posts we couldn’t match to a website. So 54 count as posts.
We sort each day’s results into four buckets. On 6 and 7 November, other websites outside our list (excluding Facebook and Reddit) took 58% and 54%, and Cars.co.za and AutoTrader together 14% and 19%. Our other 18 brands took 8% both days, and Facebook or Reddit posts 20% and 19%. The list of outside sites next to it uses the two-day rule. It ranks them by how many searches they held on both days.
How does the phone check work?
The same three steps on every site. The offer page and the contact action are picked by a written rule, the same for every site, never by hand. A step meets our criteria when it loads and shows its content on a phone-sized screen. The contact step meets them when its destination opens and we see a form, a tap-to-call link or a WhatsApp link there. Meeting the criteria isn’t a completed enquiry. We never submit one.
Every site gets two passes. A step is marked “not observed” only when both passes agree that we didn’t see it. If a bot check or a timeout blocks us, or the two passes disagree, the step is “not checked”. A contact step that sits behind an offer page we couldn’t check is “not checked” too, not judged.
On 7 November we re-checked six sites twice more: Bidvest McCarthy, Cars.co.za, ChangeCars, CMH Group, Rola Motor Group and Surf4Cars.
What does “not observed” mean?
It means we didn’t see it on the tested pages, on the date shown. It doesn’t mean the business doesn’t have it. A WhatsApp link on a branch page we didn’t test is still a WhatsApp link.
Why can a company be “partly checked”?
Because some sites turn automated checks away. A bot check, an access refusal (error 403) or a page that doesn’t finish loading stops our script. It tells you nothing about what a person sees.
When a step can’t be checked in either pass, we mark the site “partly checked”. We list it with the reason and don’t put it in the phone board’s steps-passed list. Its search result still counts. Edition 1 has four: Cars.co.za, ChangeCars, CMH Group and Surf4Cars.
How are speed and the first screen measured?
Speed is Lighthouse’s mobile performance score for each home page. Every tested site (19 of 20) has 5 retained runs, and we report the median with the lowest and highest run. We used Lighthouse version 13.6.0 for all 19. It’s a lab score under our test conditions, not a measure of real visitors. Lighthouse counts 90 and above as good. Surf4Cars couldn’t be tested, so 19 sites have a score.
The first-screen picture covers the ten sites in the hub’s two rounds, not every brand. It’s one load of each home page in a Chromium browser at phone size, 390 by 844 pixels, with an iPhone browser identity. We took it 4 seconds after the page loaded and noted what was on the first screen by eye. One load is a snapshot. Banners can depend on the time, on cookies and on location.
Who pays, and who gets told first?
No company pays to be included, for a better result, or to have a valid finding removed. Results are decided before any outreach, and sales never changes a result.
Once the Index is live, we email every named company its own figures before we share the Index anywhere. They have 5 working days to comment, and any comment is added next to their finding. Corrections are logged on this page.
The Index is research, based on what we observed. RhinoBerry sells the kind of work the Index looks at: getting found on Google and turning website visitors into enquiries. Some of these companies may become RhinoBerry clients. If they do, their findings stay exactly the same. RhinoBerry clients are marked as clients next to their name, and none of the 20 companies in edition 1 is a client.
How do you correct us?
Email [email protected] with the company, the page and what you think is wrong. We re-check within a week and log every change below, with the date and the reason.
If your website changed after our check, we’ll say so. A later change to a site is different from an error in what we observed, and the log keeps the two apart.
What changed after hand review?
Before naming anyone, we checked the boards by hand on 7 and 8 November 2026. These changes are in the data and on the boards.
| Company | Before | After | Why |
|---|---|---|---|
| All search counts | One sample on 6 November, recorded twice | Two samples, 6 and 7 November | Our first attempt at a second sample was a cached copy of the first: a tool on our side replayed the same result instead of asking Google again. We caught it in review and took a genuine second sample on 7 November. Under the two-day rule every non-zero count fell (Cars.co.za 15 to 10, AutoTrader 14 to 9), and Hatfield Motor Group replaced CFAO Mobility as the dealer leader. |
| Hatfield Motor Group | Subdomain pages not counted | Subdomain pages counted | vwbryanston.hatfieldgroup.co.za, a subdomain of Hatfield’s own site, was on page 1 for a search the main site wasn’t. The rule since then: a subdomain of a brand’s own website counts for that brand. |
| Bidvest McCarthy | Phone check 2 of 3 (offer page not observed) | 3 of 3 | Re-checked twice on 7 November. Both passes found and loaded the “Buy a new or used car” page. The 6 November miss didn’t repeat. |
| Cars.co.za (8 November) | Contact action “not observed” | Contact action “not checked” | The offer page refused our check, and the contact check then fell back to the home page. A contact step behind an offer page we couldn’t check is now “not checked”, not judged. Only Cars.co.za changed. |
| Blue Chip Dealers | On the dealer board | Moved out, not ranked | bluechipdealers.co.za is a dealers’ association. It doesn’t sell cars, so it now sits with industry associations. This also made the cohort 20 brands. |
| Four sites’ speed scores | Fewer than 5 Lighthouse runs | 5 runs each | CMH Group, Bidvest McCarthy, NMI Motor Group and Hatfield Motor Group had runs time out. We re-ran them from scratch to 5 successful runs. Their medians changed: CMH Group 7 to 5, Bidvest McCarthy 14 to 18, NMI Motor Group 45 to 38, Hatfield Motor Group 21 to 12. |
One change came earlier, on 6 November, before the boards were built. Auto Pedigree moved from marketplaces and classifieds to dealer groups and independent dealers, because its own website describes a used-car dealership.
We also checked these, and left them as they were:
- Cars.co.za stays partly checked. Its offer page refused our automated check (error 403) in both passes on 7 November. That’s an honest status, not a site failure.
- CMH Group and ChangeCars: the contact step still couldn’t be checked in either pass.
- Surf4Cars still showed our check a bot check or an empty page, in both passes.
- On Rola Motor Group’s tested contact page, which loads, our two checks didn’t detect a form, tap-to-call link or WhatsApp link, so it stays “not observed”. We didn’t test every contact route on the site.
- westvaalselectcars.co.za was on page 1 on both days for one search. The name matches Westvaal Motor Group, but neither site states the link and westvaal.co.za doesn’t link to it. Not counted.
- kingofcars.co.za (Boksburg), on page 1 for one search on 6 November, is a different business from King Cars (kingcars.co.za, Cape Town). Not counted.
What are the known limits?
This is a dated picture from two days. Page 1 changes over time, and our own two samples show how much it can move between two days. The 30 searches are a list we chose, in English, on desktop. A company at 0 may have been on page 1 on one of the days, or may rank for its own name or for searches not on the list.
Dealer groups with separately registered dealership websites are counted on their main site only. NTT’s ntttoyota.co.za is an illustration from outside this cohort. Motus and Supergroup were excluded from the cohort. We recorded motusselect.co.za on page 1 on both days for one search, so they’re candidates for edition 2. The phone screenshots use the same browser identity as our checker.
Corrections log
| Date | Company | What changed | Why |
|---|---|---|---|
| No corrections since publication yet. |
The Meerkat Index is published by RhinoBerry and produced by Cobus van Vuuren. RhinoBerry is a Growth & Systems Partner that helps businesses win more clients, consistently.
Want a read on your own business instead? The free Business Momentum Score takes about 4 minutes and names your weakest area, from being found to winning enquiries. It’s separate from the Index and has no effect on any result. [https://score.rhinoberry.com/]