Skip to content
FlatRelay
How it worksServicesBlueprintPricingResourcesGet your AI Visibility Score

How it works · The Score & the method

A measurable method for getting found, understood, and quoted by AI.

We don't guess whether AI "likes" your business. We test it with the exact queries your buyers use, score it the same way every time, and prove the change. Here's precisely how.

Get your AI Visibility ScoreSee what the Audit includes

Retrieval, not guesswork.

We optimize for what's measurable: live retrieval and citation. Instead of hoping to influence future training data, slow and impossible to prove, we test what AI engines find, parse, cite, and recommend right now, against a fixed set of buyer queries. Anything we claim, we can show you in a live answer.

The FlatRelay AI Visibility Score.

A 0 to 100 score across five weighted pillars. The weights reflect what actually decides whether you make an AI shortlist: being found and understood matters most, so those pillars carry the most weight.

Findability

30%

Are you surfaced or cited for real buyer queries across AI engines?

How it's scored: Cited in N of M model responses across a fixed query set

Readability

20%

Can AI parse your offers, specs, services, and constraints?

How it's scored: Structure & clarity checklist on key pages

Structured data

20%

JSON-LD, entity/product graph, llms.txt, crawlability.

How it's scored: Presence, correctness, completeness

Trust signals

15%

Reviews, certifications, third-party citations, authority.

How it's scored: Checklist of trust markers AI can retrieve

Recommendability

15%

When cited, are you framed positively against competitors?

How it's scored: Rank / sentiment inside AI answers

How the pillars map to the offer: Findability and Readability are AI Visibility and AI Readability, the core wedge. Structured data and Trust signals support both. Recommendability is where AI Transactability starts to matter, because a confident recommendation depends on commercial proof AI can retrieve.

We test the questions your buyers actually ask.

Before any testing, we define roughly ten buyer queries with you, the real prompts a customer would type when shortlisting a supplier. Good queries are specific, not generic:

Generic (weak)

aluminium supplier

Real buyer query (what we test)

Best ISO-certified aluminium extrusion supplier for low-volume prototyping in Southeast Asia

We also define the competitors to benchmark against, so the result is a like-for-like picture of who AI recommends in your space.

Across every engine your buyers use.

We run each query across the major AI surfaces, because answers differ by engine.

  • ChatGPT
  • Gemini
  • Perplexity
  • Claude
  • Copilot
  • Grok
  • Google AI Overviews

For each query and engine we record whether you were cited, ignored, misunderstood, or out-ranked, and capture the actual prompt and answer as evidence. That record is what makes the Score concrete: you can see the exact moment AI did or didn't pick you.

The proof is the delta.

We score you before implementation and re-score with the same queries afterward. The change in citation rate, answer quality, and how you're framed against competitors is the proof the work paid off. Because AI answers drift over time, monitoring re-runs the same Score monthly, so the proof stays current and you catch competitors gaining on you.

Sample query · before / after

Before

Not cited

0 / 10 responses

After

Cited

7 / 10 responses

Illustrative.

How an engagement runs.

  1. 01

    Audit & Score

    Define queries, test across engines, score the five pillars, deliver the gap analysis and roadmap.

  2. 02

    Implement

    Fix what the Audit found: structure, content, spec and FAQ pages, entity/product graph, trust signals, llms.txt, and structured-data markup. Visibility and Readability first.

  3. 03

    Monitor

    Re-run the Score monthly, track competitor drift, and keep recommendations current as engines change.

Sequencing: the Audit earns the implementation; the implementation earns the monitoring. Each step proves the next is worth it.

The fixes, in plain terms.

Most of the work isn't a redesign, it's making your existing presence machine-readable. The four levers:

Content actions

Restructure key pages into direct answers and definitions AI can lift and cite.

Learn the fix

Content structure

Add the spec, comparison, and FAQ pages that answer real buyer questions, with clean semantic headings.

Learn the fix

llms.txt

Publish a machine-readable guide to your site so AI engines know what you offer and where to look.

Learn the fix

Structured-data markup

Add JSON-LD (Organization, Product, Service, FAQ, Article) and an entity/product graph so AI reads your catalog and credentials correctly.

Learn the fix

Questions, answered.

Why weight Findability highest?

If you're not surfaced at all, nothing else matters, there's no recommendation to improve. Visibility is the gate; readability and trust raise quality once you're through it.

Isn't AI output random? How can a score be repeatable?

Individual answers vary, which is why we use a fixed query set across multiple engines and measure citation rate across many responses, not a single reply. The pattern is stable enough to score and to re-test.

Can you really influence what ChatGPT says?

We influence what it can retrieve and cite. Engines increasingly pull from live, crawlable, well-structured sources. Make your business the clearest, best-structured source for a query and your odds of being cited rise, measurably.

How often should the Score be re-run?

Monthly. AI engines update, competitors change, and citations drift. Monthly monitoring keeps the picture and your position current.

Put a number on it.

Run the method on your business and see your AI Visibility Score across all five pillars.

Get your AI Visibility Score