OffroadRigHQ: 1,600 pages of tire and wheel research, run by a small team and AI agents
OffroadRigHQ is an independently operated tire and wheel review site that Xumulus designed, built, and runs. It went from an empty repository to roughly 1,600 pages in six months, and in its latest eleven weeks moved from page 7 of Google to page 3.
The problem: retailers lose the research stage
Off-road buyers research for weeks before they buy. They search for things like "what size is 295/70R17", "KO2 vs KO3", and "best mud tires for an F-150". A retailer's product pages rarely win those searches. The traffic goes to forums, review sites, and the large marketplaces, and the retailer meets the buyer only at the end, on price.
A store cannot fix this by calling its own pages neutral. The honest answer is a separate guide that does the research properly and sends ready buyers on to retailers. Two things usually stop that from happening:
- Cost. A useful guide needs thousands of pages of specs, sizes, and fitment. Written by hand, that is more than a retailer's content budget will carry.
- Trust. Affiliate sites that inflate reviews lose their readers, and they draw regulatory attention. The site is only worth building if it stays honest at scale.
What we built
A content platform on Payload CMS, Next.js, and Postgres where facts arrive from data feeds, original writing sits on top of them, and every buy link knows whether the retailer behind it has the product in stock.
- A catalog from licensed data. Specs, per-size load and speed ratings, warranties, and factory fitment for 370 tires, 1,052 wheels, 58 brands, and 47 vehicles come from a licensed industry data feed. The supplier's marketing copy is never published. Every public description is rewritten in the site's own voice.
- Live retailer stock on every page. A daily sync checks the storefront of the site's primary retail affiliate, National Tire & Wheel, and records which products it carries. A buy button appears only when the product is in stock.
- Pages generated from the data. 68 tire size pages, a size chart, and a tire size calculator are built from the sizes themselves, with nothing typed by hand. These were the first pages to rank.
- Editorial pages. 14 buying guides, 13 head-to-head comparisons, and 6 how-to guides, with named sources, a published editorial method, and a corrections page.
- Original research. One example ranks 12 all-terrain tires by price per 1,000 miles of the maker's warranty. Pieces like this exist so that other publications have a reason to link to the site.
- Measurement. Search Console and GA4 feed a warehouse. Clicks out to each retailer are tracked, with crawler traffic filtered from the count.
Nothing reaches a reader without passing through the CMS, where every draft waits for a person to publish it.
How the work is divided: agents do the legwork, people sign off
A small team runs a site of this size because AI agents do the repetitive work and people approve everything a reader sees. The rules are written into the codebase, so every agent and every new teammate works to the same ones.
- Agents import and sync data, draft copy and comparisons, check warranties against the makers' own sites, and build and test pages.
- People set priorities, review drafts, approve images, and publish.
The trust rules every page follows
- Never imply hands-on testing that did not happen.
- Retailer data supports price and availability only, never a verdict on a product.
- Sources are cited and linked directly, not through affiliate links.
- Every affiliate link is disclosed.
The site also has a voice of its own, plain-spoken and a little irreverent, so that a page built from a spec feed does not read like one.
Results so far
The site is seven months old and the numbers are early. They are reported here as they stand, for the eleven full weeks from August 24 to November 8, 2026.
- Search position: the average Google position went from 64 to 22, which is page 7 to page 3.
- Visibility: weekly Google impressions tripled, from 491 to 1,478.
- Page one, where it started: tire size pages reached page one for searches such as "what size is 295/70R17", and the F-150 mud tire guide reached an average position of 7.7.
- AI answers: the site appeared in Google's AI search results 107 times between July 26 and October 19, half of them in the final week of that period. The first visits referred by ChatGPT arrived shortly after.
Most of the site is not on page one yet, so traffic is still small. What the first seven months show is that the pages generated from structured data rank first and that the editorial pages follow them up.
What comes next
- More of what already ranks: vehicle pages for the Wrangler, Ram 1500, and Tacoma on the F-150 pattern, more comparisons, and more size pages.
- Links from other sites: the cost-per-warranty-mile study is going out to off-road publications.
- Titles written for the click: pages already on page one were winning fewer than 1% of searchers. New search titles went live on November 10.
- New ground: wheel buying guides to sit on top of the 1,052 wheel pages, and ATV and UTV tires.
What this means for a merchant with a large catalog
OffroadRigHQ is the same method Xumulus brings to client catalogs, run in the open. Product facts live in one structured source. Pages, tools, and feeds are generated from it. AI agents carry the volume, and people own the judgment and the publish button.
That is the working framework behind our fractional digital leadership and our large-catalog builds. If your product data could answer the questions your buyers are searching for, and today it does not, this is what the alternative looks like after seven months.
Find out what your catalog could answer
Send a note with your platform, rough SKU count, and where your product data is authored. You will get a direct answer from Daniel on what an approach like this would take for your catalog, including when it is not worth doing.