HomeOur method, in the open
The glass box

Every other findability score is a black box. Ours is published.

Most tools hand you a number and ask you to trust it. We don't. This page shows you exactly how every score is built, every check we run, how your day-one baseline is sealed so it can't quietly change, and the honesty rules we never break. Read it, check our work, and hold us to it.

The three scores

Three scores, and exactly how each one is built.

We roll everything we measure into three plain 0–100 scores. Here is the recipe for each — the parts that go in, and how much each part counts. These are the real weights we use, not a marketing simplification.

GFI — Google Findability Index

0–100

how easily people find you on Google today (0–100).

Index coverage45% of the score

how much of your site Google has actually indexed and can show.

SERP position45% of the score

where you land on the Google results page for the questions your buyers ask.

AI Overview presence10% of the score

whether Google shows its AI-written summary for those buyer questions.

SEO — Search Engine Optimization score

0–100

how well your website is built to be found and trusted (0–100).

Technical25% of the score

the on-page basics: titles, headings, structured data, clean links.

Content & E-E-A-T15% of the score

whether your pages fully answer the question and show real expertise and trust.

Internationalization10% of the score

whether each language version is set up correctly for search.

Core Web Vitals20% of the score

how fast and stable your pages feel to a real visitor.

Security & health10% of the score

a valid certificate, one clean address, no penalties or warnings.

Authority20% of the score

how many trusted sites link to you — a vote of confidence Google counts.

AFI — AI Findability Index

0–100

how often AI assistants like ChatGPT mention you when people ask about what you do (0–100).

Mention rate35% of the score

how often you're named at all across the AI answers we check.

Citation rate30% of the score

how often your own site is cited as the source.

Sentiment15% of the score

whether what's said about you is accurate and positive.

Competitive share20% of the score

your share of the conversation versus the competitors named alongside you.

AFI carries a confidence band (±). It's an estimate that gets sharper as we run more checks each week — never a guaranteed ranking.

Weights shown are the live values used to compute your scores. When a part can't be measured yet, it's left out and the rest is re-balanced — never guessed.

The full check set

Behind the three scores: every check we run, named.

Your scores aren't a vibe. They roll up from a fixed set of 62 checks across your Google and AI findability — the same set, every client, every audit. Here is the whole list, grouped, so you can see exactly what we look at. We measure all of it; we only ever charge to fix what's worth fixing.

Google findability checks

Can Google find and index your pages

7
  • noindex directives
    good only intentional pages noindexedleak money pages noindexed
  • robots.txt Disallow
    good no buyer-relevant path blockedleak key routes disallowed
  • Canonical conflicts
    good self/correct canonicalleak points elsewhere
  • Sitemap health
    good present, count = real pagesleak 0 / far below
  • Coverage / crawl errors
    good high indexed-valid ratioleak discovered-not-indexed, soft 404s
  • Orphan pages & internal link graph
    good money pages linked, shallow depthleak orphans
  • XML sitemap validity & lastmod
    good valid XML, honest lastmodleak malformed/false

Where you land on the results page

7
  • Position on buyer keywords
    good pos 1–10leak absent / >20
  • SERP features (AIO/PAA/snippet)
    good client cited / owns snippetleak competitor cited
  • Click curve / CTR vs expected
    good CTR ≥ benchmarkleak high impressions, low CTR
  • AI Overview citation footprint
    good cited in AIOs for money termsleak zero
  • Share-of-voice / full footprint
    good broad ranked footprintleak a few terms only
  • Keyword cannibalization
    good one URL per intentleak competing URLs
  • Striking-distance
    good few near-page-1 gapsleak many pos 11–20

The on-page technical basics

7
  • Title / meta / H1
    good one H1, title ~50–60, meta ~120–160leak H1=0/2, bad meta
  • Structured data / schema
    good valid Org/sameAs/Product/FAQleak missing/invalid
  • Redirects & broken links
    good clean 200sleak chains, 404 internal links
  • Alt-text coverage
    good descriptive altleak many missing
  • JS-render visibility
    good content server-renderedleak JS-only
  • Mobile-first parity
    good mobile = desktop content/linksleak hidden on mobile
  • Deep schema validation & eligibility
    good no Rich-Results errors, eligibleleak invalid

Content depth, expertise, and trust

4
  • Content depth vs intent
    good fully answers the queryleak thin vs intent
  • E-E-A-T signals
    good author/expertise/trust + citationsleak anonymous
  • Thin / AI-spam content
    good original, usefulleak duplicative/auto-gen
  • Content freshness / dateModified
    good updated, honest dateModifiedleak stale/false

Multi-language setup

3
  • hreflang reciprocity & x-default
    good bidirectional, valid codes, x-defaultleak one-way/broken
  • Locale targeting & parity
    good each locale indexed, translatedleak duplicate/untranslated
  • i18n routing & geo-redirects
    good crawler-safe locale redirectsleak cloaks content from bots

Speed and page experience

3
  • Lab LCP / INP / CLS
    good LCP ≤2.5s, INP ≤200ms, CLS ≤0.1leak slow/heavy
  • Field CWV p75 (CrUX)
    good all Good at p75leak poor / no field data → field|lab
  • Page weight & resource budget
    good lean payloadleak render-blocking, heavy 3rd-party

Security and site health

3
  • HTTPS / HSTS / mixed content
    good valid cert, HSTS, no mixedleak http resources
  • Host canonicalization
    good one host, single 301leak duplicate hosts/chains
  • Manual Actions & Security Issues
    good noneleak penalty / hacked / spam

Authority and links from other sites

7
  • Referring domains + backlinks
    good healthy/relevantleak near-zero
  • Domain rank (0–1000)
    good comparable/above competitorsleak far below
  • Backlink gap vs competitors
    good parity/advantageleak competitors hold domains you lack
  • Backlink toxicity / spam
    good clean profileleak spam/PBN/toxic
  • Anchor-text distribution
    good natural mixleak over-optimized exact-match
  • Unlinked brand mentions & PR
    good mentions + reclaimable linksleak no presence
  • Disavow status
    good current/cleanleak unmanaged toxic links

Local search presence

5
  • Google Business Profile completeness
    good complete, verifiedleak missing/sparse
  • NAP consistency
    good identical everywhereleak conflicting
  • Reviews / ratings
    good volume/recency/responsesleak few/old
  • Local-pack presence
    good ranks in map packleak absent
  • LocalBusiness schema
    good valid + geoleak missing/invalid

AI findability checks

Whether AI answers mention and cite you

4
  • Mention rate
    good surfaces across neutral Qsleak never named
  • Citation rate
    good domain citedleak zero
  • Sentiment / accuracy
    good accurate + positiveleak wrong (−1)
  • Competitor share-of-voice
    good meaningful shareleak competitors dominate

Whether AI engines can read your pages

5
  • AI-crawler access
    good explicit allows for major AI botsleak blocked
  • Server-render of answer passages
    good quotable text in server HTMLleak JS-only
  • Expanded AI-crawler list & WAF
    good full bot list allowed at robots + WAFleak blocked at edge
  • llms.txt presence & validity
    good valid /llms.txtleak absent/malformed
  • Answer-passage render depth
    good SSR passages on all answer pagesleak only home

Your footprint across AI surfaces

3
  • Multi-engine footprint
    good present across the probed AI answer enginesleak absent
  • Google AIO citation footprint
    good cited in AIOsleak zero (AI-axis view of A9)
  • AI-referral traffic baseline
    good AI-engine referrals capturedleak untracked

Whether you're a recognized entity

4
  • Entity / sameAs consistency (on-site)
    good consistent name/identity/sameAsleak fragmented
  • Knowledge Panel / knowledge-graph
    good recognized entity / panelleak none
  • Wikidata / Wikipedia entity
    good accurate entityleak none/outdated
  • Cross-web NAP / entity consistency
    good identical entity across webleak off-site drift
The sealed baseline

Your day-one baseline is sealed, so no number can quietly change.

When your first full audit is done, we freeze it. Every part of it gets a unique fingerprint — a long code calculated from the exact numbers we recorded. Change a single digit later and the fingerprint no longer matches, so any edit is obvious. That's what lets us measure honest movement: your starting line can't drift to make us look better.

01

Written once

Your baseline is recorded once and never overwritten. If we ever re-baseline, it's a new, dated version that records what it replaced — never a silent edit of the original.

02

Tamper-evident fingerprint

A unique code is calculated from the exact numbers. If anything changes, the code changes too — so you can prove the baseline wasn't touched.

03

Every number traces to its source

Each part of the score is stamped against the actual measurement it came from, so any figure can be traced back to where it was read.

How we check AI answers

We ask the real AI answer engines the questions your buyers actually ask.

To measure your AI findability, we put your buyers' real questions to live AI answer engines and read what comes back — are you named, are you cited, what's said about you. We do this only through official, sanctioned access to those engines, never by scraping a consumer app or breaking anyone's terms.

Official access only

Every AI check runs through official, sanctioned channels. We never use scraped consumer screens, borrowed personal accounts, or back doors. If an engine can't be checked properly, it's left off your report and marked not measured — never faked.

Today we check these answer engines: ChatGPT, Claude. The list is data-driven — it grows as we connect more engines, and we will never name an engine on your report that we didn't actually check.

ChatGPTClaude
The rules we don't break

What we will never do with your numbers.

A published method is only worth something if the rules behind it are firm. These are ours, and they hold for every client, every report, forever.

We never promise rankings, traffic, or leads

We show you proof charted against your own baseline. We don't guarantee a position on Google, a number of visitors, or a flow of leads — anyone who does is guessing.

Every number is measured or marked not measurable

If we can't measure something yet, we say so plainly — it shows as not-yet-measurable, never as a fabricated figure to fill a gap.

Estimates carry their confidence

An estimate always travels with its confidence band, so you know how sure we are. We never present a banded estimate as a hard fact.

We follow GDPR and only measure what we're allowed to

We work within European data rules. We measure your public findability and the data you grant us access to — nothing more, and you can ask what we hold at any time.

Standards we measure against

The public standards behind our checks.

Our method isn't a private invention. Every check on this page maps to guidance the search and AI engines publish openly. These are the primary sources behind the scores — read them yourself and check our work against them.

Request your baseline.

Send your site, market, and goal. We'll set your sealed day-one baseline using exactly the method on this page — then chart every move against it. No ranking guarantees, ever.