Technology · Methodology

How we measure whether AI names you

No gut feeling, no single screenshots. A repeatable pipeline across web-searching AI assistants, with the same question set before and after - so the progress is verifiable.

Start free scan →
01The pipeline

Four phases from measurement to proof

Every engagement runs the same loop. It is deliberately simple - because a result is only worth something if it can be repeated with the identical test.

01

Probe

We put the real buyer questions of your industry to web-searching AI assistants - ChatGPT, Perplexity, Gemini, Claude and Grok. Not "Do you know company X?", but the open questions a buyer actually types: "Who can help me with … in … ?" That shows who the AI names on its own.

02

Score

Every answer is rated on four signals - visibility, accuracy, structure, authority. From these we build a transparent overall score that shows exactly which of the four levers is stuck.

03

Optimize

The score becomes action: finished building blocks (llms.txt, structured data, quotable content) plus a prioritized roadmap with an installation guide. Your website stays in your hands - we deliver the parts, not a black box.

04

Re-scan

After installation we re-measure with exactly the same question set. The change becomes visible prompt by prompt - honest, because the test is identical. No flattering math, no shifted yardstick.

02The score in detail

Four signals, four concrete questions

A single percentage says little. We break AI visibility into four components - each answers a clear question and is tied to a lever we work on.

Are you mentioned?
0–100

Visibility

Across all buyer prompts: how often does your brand appear in the answer? We count mentions, position and whether you appear first or as a footnote.

Is the picture right?
0–100

Accuracy

Does the AI description match your positioning, audience and services? We detect when the AI confuses you, describes you out of date, or drifts toward a competitor.

Can machines read you?
0–100

Structure

Schema.org markup, clean headings, entity markup, llms.txt and extractability - the technical foundation AI crawlers can actually parse and quote.

Do they trust you?
0–100

Authority

Third-party sources, industry directories, diversity of mentions and link signals - the evidence an AI looks for before putting a name into its answer.

The bars shown here illustrate the scale - your real values only emerge when we scan your domain.

03The levers

Four levers - each tied to one signal

Optimization here is not a vague list of measures. Each lever targets one of the four signals directly and becomes verifiable at the re-scan.

// Lever 01 → Structure

Make it machine-readable

llms.txt, Schema.org (Organization, Service, FAQPage), a clear heading hierarchy and entity markup. So the AI crawler finds unambiguous, quotable facts instead of prose to guess from.

// Lever 02 → Visibility

Quotable answer blocks

We write the short, precise Q&A blocks AI assistants prefer to pick up verbatim - matched exactly to the questions your buyers ask.

// Lever 03 → Accuracy

Sharpen positioning

Unambiguous statements about service, audience and region, so the AI classifies you correctly instead of confusing you with similar providers.

// Lever 04 → Authority

Build verifiability

We show which third-party sources, directories and mentions are missing, so the AI finds your claims confirmed - the basis for being named at all.

04The measurement loop

Why we re-measure after 30 days

AI systems don't adopt changes instantly. They have to re-read your website, re-weight new sources and update their index. That is why the re-scan is a fixed part of the methodology - not the end, but the proof.

Day 0

Probe

Buyer-grade prompts hit up to five engines. Baseline value per system - your honest starting point.

Day 7

Roadmap

Prioritized actions per lever, each with a target signal and expected effect.

Day 14

Delivery

Finished building blocks installed, website made machine-readable, technical base set.

Day 30

Re-scan

The same probe, the same yardstick. Change evidenced prompt by prompt.

↑ measurable

In the monthly subscription this loop repeats continuously - so you stay visible even as the models change.

05Reach & honesty

Up to five web-searching engines - and clear limits

We test the assistants your buyers really use, and we are open about what measurement can and cannot do.

Tested across up to 5 enginesChatGPTPerplexityGeminiClaudeGrok

What we do

What we don't promise

Improve technical AI discoverability

Structure, readability and verifiability of your content - the things that are in your hands.

No guaranteed mention

Whether and when an assistant names a company is decided by its provider - as with any reputable SEO or PR service.

Make change measurable

Same question set before/after, result documented prompt by prompt.

No manipulation of the models

We shape the public context, not the AI itself. No tricks, no back door.

Free · 60 seconds

See your baseline

The free scan shows phase 1 of the pipeline for your domain - your honest starting point, no obligation.

Start free scan →