Data as of 27 August 2026 · domain registered about 10 months ago, current site live for about 5.5 months and ranking for about 3 months
Who this case is for
If you run a product review site, a car information site, or an affiliate site that earns from product recommendation links, and you wonder why dozens of articles still leave your traffic concentrated on a handful of pages, this case is the closest match.
It also suits owners of sites about high-priced products with a payment calculator page. The thing that moved rankings fastest here was fixing wrong figures on the page people use to decide whether to buy.
At the end there is a checklist for car information and product review sites that you can use on your own site right away.
We do not name the site, the brand, the car model or competing sites.
1. The starting point: traffic was arriving, but concentrated on 3 pages
This is a Thai-language information site about one popular hybrid SUV model. It covers prices and trims, specs, colours, a payment calculator, a finder for 455 service centres, an accessories page with 743 affiliate links to online stores, and around 54 articles, 77 pages in total.
Its visitors are Thais researching before buying a car: how much the monthly payment is, which colour looks best, what a used one costs. The site earns when people click an affiliate link and buy.
Site age at the time of the data
- The domain was registered in November 2025, about 10 months before the data date used in this article (27 August 2026).
- The current version of the site was rebuilt in early March 2026, about 5.5 months earlier.
- Ahrefs started estimating Google traffic in late May, and the site started ranking in June, about 3 months before the data date.
The site still had old WordPress-style URLs, and 428 domains linked to it, all DR 0 (Ahrefs, 15 August 2026). So the domain was probably used by someone before. This is an inference from the data, not a confirmed fact.
How we worked
We started on 31 July 2026 and worked in 3 rounds.
- 31 July – 1 August: audited all 77 pages, upgraded the site’s build system, fixed speed and security, then redeployed with a backup of the old files.
- 15 August: fixed wrong data on the payment table page, merged duplicate articles and built topic hub pages.
- 25–27 August: pushed pages ranking 11–20, wrote new articles, made 15 new cover images and installed GA4.
Each round began with Search Console and Ahrefs data to see which pages people were already searching for, before deciding what to fix.
2. What we found, and what we did
The picture at the start
The three pages getting almost all the traffic were the colours page (about 180), the payment table page (about 100) and the homepage (about 89). These are Ahrefs traffic estimates from 15 August 2026. In other words, the dozens of other articles were doing almost nothing for the site.
The biggest problem: wrong figures on the money page
The payment table page is where people are closest to deciding to buy. On 15 August we found several errors there.
| What we found | Effect on readers and Google |
|---|---|
| The table header said “out of stock 20%” instead of “20% down payment”, in 17 places | Confusing for readers, and it looks untrustworthy |
| The page title showed the monthly payment as “9,xxx” instead of a real figure, in 8 places, including a call-to-action button in all 63 articles | People saw it in search results and didn’t want to click, because the page looked unfinished |
| The payment formula used reducing-balance interest, while car finance in Thailand uses a flat rate | The payments shown were about 8% lower than reality |
| No payment table for the special-colour trim | People searching for that trim found no answer |
| The legal page still had a sample phone number that was never replaced | It hurt the credibility of the whole site |
What we fixed:
- Switched the formula to flat-rate interest so the figures match what finance companies actually charge.
- Fixed all 17 table headers, and made the page title show the lowest monthly payment as a real figure.
- Added the missing tables: the special-colour trim, a comparison of 1.99% and 2.99% interest promotions, and a low down payment table.
- Removed the sample phone number from the legal page.
Wrong car data in articles
Beyond the payment table page, we found incorrect information in about 7 more articles. For example, one said the engine was a 3-cylinder when it is actually a 4-cylinder, and some mentioned ventilated seats and genuine leather seats this model doesn’t have. Another 4 articles had unreadable Thai text mixed with characters from other scripts.
We fixed all of it and checked the specs against the manufacturer’s official Thai website, producing a comparison table of all 4 trims. For a car information site, accuracy is the only reason people will trust it and come back.
Duplicate articles competing with each other
The site had 8–9 pairs of articles on the same topic. When two pages answer the same question, Google doesn’t know which one to send people to, and both end up ranking lower than they should.
We merged 11 duplicate articles into the main articles, set up 301 redirects from the old pages to the new ones, and tested every route (9 of 9 routes in the first round, 14 routes in total). We also fixed another 93 internal links that pointed to the old pages.
Topic hub pages: a path to the forgotten articles
At the start, 13 articles had only 1 internal link pointing to them from elsewhere on the site. Google treats articles with almost no inbound links as unimportant.
We grouped the articles into 8 topics, such as price, payments, specs and accessories, then built 8 topic hub pages with CollectionPage Schema. Every article in a topic links to the others and back to its hub.
New content, and pushing pages close to page one
- 4 new articles: the 2025 model year, how the hybrid system works, a colour ranking, and accessories.
- 4 articles fully rewritten: used car prices, the special-colour trim, the top trim, and pros and cons.
- Added FAQPage Schema to 6 articles.
- 15 new cover images for 7 articles and the 8 hub pages.
In the 25 August round, we fixed pages ranking 11–20 before writing anything new. Google already sees these pages as relevant, and they only need a little more to reach page one, so they are worth more effort than starting articles from scratch.
What readers can’t see, but Google can
The audit of all 77 pages on 31 July 2026 found things most site owners never look at.
- The security and cache config file had never worked, because it was written for a different type of host, not the one actually in use. We rewrote it in full with CSP, HSTS, security headers, 1-year caching for images and scripts, HTTPS on every page, and redirects for the old WordPress-style URLs.
- No measurement had ever been installed. The privacy policy said the site used Google Analytics, but there was no code on the site, and the CSP was blocking Google’s scripts anyway.
- The service centre finder page was very heavy. The HTML file was 2,171 KB with 10,648 elements on the page, and tasks that froze the page for 368 ms and 137 ms.
- 1,399 images had no dimensions set, so pages jumped while loading. There was no skip-to-content link for keyboard users, and the meta descriptions were too long.
The service centre finder page was restructured and no longer has tasks that freeze it while loading. The accessories page shows the first 24 products immediately and loads the rest as you scroll. Fixing the font loading order cut the main image load time on mobile (LCP) from 2.7 seconds to 2.3 seconds.
Lighthouse results on the live site after the fixes (31 July – 1 August 2026, our own tests): on desktop, a speed score of 97–99 with LCP of 0.6 seconds; on mobile, 76–95 depending on the run, with LCP of 1.8–2.3 seconds. Accessibility, best practices and SEO scored 100 on both, and CLS was 0. We didn’t record Lighthouse scores before the fixes, so we can only compare against the list of problems found in the audit.
Measurement that counts what earns money
On 27 August we created GA4, linked it to Search Console, set data retention to 14 months, and set it to count 2 things: clicks on affiliate links out to online stores, and contact clicks. We then tested in Realtime that both were really being counted.
3. Results
Results from Google in 3 months
These figures come from Google Search Console for the latest 3 months to around 24 August 2026, read on 27 August. The trend line rose from almost zero in late May to around 70–80 clicks a day.
To be clear, these 3 months include time before we started. The site had been growing on its own since June as it got older, and our work started on 31 July. So the 2.42K clicks reflect both the site’s age and our work, not our work alone. We didn’t export daily data, so we can’t split the Search Console figures into before and after.
Weekly estimated traffic
This chart shows two things. First, the site grew on its own to some degree before we started. Second, after steady growth, the figure held flat for three weeks, then moved up again after the payment table fixes and the duplicate article merge. One week isn’t a trend yet; it needs several more weeks of watching.
Measured results by page
| Page | Before | After | Period and source |
|---|---|---|---|
| Payment table page, this model’s payment table query (about 2,300 searches a month) | Position 17 | Position 10 | Ahrefs 10–24 Aug 2026 |
| Payment table page, estimated traffic | 99 | 144 (+45%) | Ahrefs 10–24 Aug 2026 |
| Colours page, estimated traffic | 171 | 189 | Ahrefs, same period |
| Price topic hub (new page), model name + “price” query | No page yet | Position 18 within 10 days | Ahrefs 25 Aug 2026 |
| 7-seater article | Not ranking | Position 7 | Ahrefs 25 Aug 2026 |
| Engine specs article | Not ranking | Position 10 | Ahrefs 25 Aug 2026 |
| Pros and cons article (rewritten) | Not ranking | Position 19 | Ahrefs 25 Aug 2026 |
A page to watch: over the same period, the homepage dropped from 127 to 91 (Ahrefs estimate), because the brand + model query fell from position 20 to 27. We logged it as work to follow up in the next round.
What doesn’t have numbers yet: GA4 was only installed on 27 August. As of the data date, the only events were our own Realtime tests, so there are no figures yet for affiliate link clicks or contact clicks from real users, and we don’t yet have a count of pages Google has indexed.
Expected results from here
Most competitors on page one for the main queries are dealer sites with DR 1–21 and no links from other sites at all. One site has DR 2 but an estimated traffic of about 13,574. That means these queries are won with content and site age, not links.
Next steps: get the remaining articles ranking, spread traffic beyond the 3 main pages, follow up on the homepage ranking drop, and start reading GA4 data to see which articles actually lead people to click affiliate links.
Checklist for car information and product review sites
Open your site and answer each point.
Money pages
- Page titles that show in Google display real figures, with no “xxx” or leftover sample text
- Payment table headers use the correct words, e.g. “down payment”, not a typo
- The car payment formula uses flat-rate interest, as Thai finance companies do, not reducing balance
- There is a table for every trim, including special editions
- Every spec has been checked against the manufacturer’s official website
Content structure
- No two articles answer the same question; if they do, merge them and add a 301
- Every article has at least 3 internal links pointing to it from other pages
- There are topic hub pages, such as price, payments and specs, that link to every article in that topic
- You fix pages ranking 11–20 before writing new topics
Back end
- Security headers and caching actually work on your host, not just exist in a config file
- GA4 code is really on the site, and the CSP doesn’t block Google’s scripts
- GA4 counts affiliate link clicks as a separate event
If you can’t tick more than half, your site is probably still losing a fair number of clicks it should be getting.
Questions car information site owners often ask us
Does a car information site need to buy links to rank? This case bought no links at all, and many page-one competitors have no links from other sites either. In a market like this, accurate data, internal link structure and site age matter more. Check the competitors for your own queries before deciding to spend on links.
I already have lots of articles, but traffic goes to only a few pages. Should I write more? Not yet. First check for duplicate articles, and how many internal links point to the articles that don’t rank. In this case, 13 articles had only 1 inbound link. Merging duplicates and building topic hub pages gave forgotten articles a path for Google to reach them, and a new topic hub page reached position 18 within 10 days.
Do small errors on a page really affect rankings? In this case, after fixing the page title showing “9,xxx” and the table headers, the payment table page moved from position 17 to position 10 within two weeks. We can’t say which change Google weighted most, but a title that looks unfinished puts people off clicking, and fewer clicks affect rankings.
How far can I trust Ahrefs traffic figures? They are useful for trends and comparing with competitors, but they are not actual visitor counts. The numbers to base decisions on are Search Console for clicks from Google, and GA4 for what people do on your site, such as clicking affiliate links.
We do SEO for information and review sites like this
If your site has dozens of articles but traffic is concentrated on a few pages, or you have price and payment calculator pages you want on page one, start with a free website audit or see the details of our SEO service.