An article we wrote outranks the client’s own product page

We write content for an AI model evaluation platform: venture-backed, selling enterprise contracts to large engineering teams. Three months in, one of our articles ranks #1 on Google for the exact search its buyers run, above the company’s own product page.

Measured externally: Ahrefs via third-party gateway + live Google search captures. No client analytics.

The result

The result: a #1 ranking where it matters most

We wrote content for an AI model evaluation platform: a venture-backed company that sells enterprise contracts to large engineering teams. Three months in, one of our articles ranks #1 on Google for the exact search its buyers run, ahead of the company’s own product page for that query.

Site signal	Jun 2026	Sep 2026	Change
Domain Rating (DR)	43	42	steady
Referring domains	353	500	+147
Ranked keywords, top-10 (Ahrefs sample)	35	31	-4
Indexed pages (Ahrefs)	48	30	-18

What these numbers mean: Domain Rating (DR) is Ahrefs’ third-party score of a website’s authority on a 0-100 scale. Ahrefs is an independent SEO data provider, not Google and not the client. All figures below are its estimates. The +147 referring domains include a share of low-quality link networks; the honest signal is the handful of legitimate new domains and a DR that held steady against that noise. We make no keyword-growth claim: the sampled keyword count fell.

What we did

Ten articles live, plus eight on the CEO’s public profile

Ten blog articles delivered and live on the client’s site (July-September 2026), covering how to benchmark LLMs (large language models), evaluate output quality in production, and govern and monitor production AI
Every URL verified live (HTTP 200) on 21 September 2026
Eight Medium pieces published on the CEO’s public profile, titles matching what we delivered
The flagship article, “How to Benchmark LLMs for Your Specific Business Use Case,” ranks #1 for its target search and outranks the company’s own product page on the same domain

What we did not claim

LinkedIn posts from the client’s team are live, but a text-overlap check against our drafts found no match. They are their own writing, so we do not claim them.
Quora was dropped: no live pages were found for the answers we wrote.
Article-level traffic is not measurable from outside. It needs the client’s own analytics, which we do not use.

The client

Anonymized, by industry and stage

We describe the client by what it is rather than its name, because the story is the shape of the company, not the brand.

Attribute	Profile
Industry	AI model evaluation: helps companies test, benchmark, and trust large-language-model output in production
Stage	Over $1M raised (prior funding round on the public record); merged with a global human-judgment network (BLEND) in January 2026 to scale edge-case AI decisioning
Backers	Co-founded by a serial founder with a prior exit and a revenue-lead co-founder. Full names redacted; see evidence file
Team	Tracxn records 26 employees (2026). Annual revenue is not publicly disclosed
Status	Self-published customer logos on the company’s own /about page (seven Fortune-500-adjacent logos). Three live AWS Marketplace listings
Proof	AWS Marketplace HERO Platform: $0.12 per processed resource with a 200,000-unit minimum (~$24,000+ minimum deployment). GenAI Model comparison: $750 one-time comparison or $1,000/year annual license

What the pricing tells you about the deal

AWS Marketplace HERO Platform: $0.12 per processed resource with a 200,000-unit minimum (~$24,000+ per deployment floor, sourced from the AWS listing)
AWS Marketplace GenAI Model comparison: $750 one-time comparison or $1,000/year annual license for up to 2 comparisons/month
A #1 rank on an LLM-benchmarking search is exactly what an enterprise platform team reads before evaluating a vendor in this category

AI visibility

Showing up in the answers AI gives

Signal	Count
Google AI Overview citations	14 (across 7 pages)
Assistant-engine citations (ChatGPT, Gemini, Perplexity)	7
Direct buyer-question probe	Client domain not yet cited in AI answer text (honest baseline, first measurement)

“AI Overview citations” means the number of times an AI answer engine (Google AI Overviews, ChatGPT, Gemini, Perplexity) mentions the client’s domain. These are domain-level counts, not proof that our specific articles are the cited pages. We re-run the same measurement monthly and compare.

How this was measured

The honest version

Where the data comes from: Ahrefs, an independent third-party SEO data provider. Its figures are estimates, not Google’s own data and not the client’s.
Rank positions: live captures of the Google search results on a single date (21 September 2026), not a tracked history.
No client analytics: we did not use the client’s sessions, conversions, or revenue numbers. Those are client-side and we will not present them as ours.
Declines are reported as declines: where a number fell in the window, it is shown as measured, not smoothed over.

Want this for your business?

Content that ranks for the exact searches your buyers run, measured from outside your analytics. A 30-minute call to see the method on your market.

Book a free 30-minute diagnostic