Artifexa Visibility Framework · v1.0
The method has a name, a version and a receipt.
Most visibility advice is one person's opinion, delivered once and never re-checked.
Artifexa runs the Artifexa Visibility Framework: a fixed method with 40 named checks and a stated rule for what counts as evidence. It is versioned, so you can see when it changes. Two reviews of the same site, three months apart, are comparable because the checks did not move.
Four questions
Four questions, in plain English.
You should not have to learn a new vocabulary to know whether your marketing is working. AVF asks four plain questions, and keeps the technical detail behind them.
Can you be found?
Whether the right buyer, or the answer engine they asked, discovers your firm for the problem you actually solve.
The work it covers
- Technical search foundations: indexability, metadata, structured data, redirects, sitemaps.
- Local search presence, where the firm serves a region.
- Presence in generative answers: whether the firm is mentioned and cited when a buyer asks an answer engine.
Can you be understood?
Whether a person or a machine can extract a clear, accurate answer about what you do and what you know.
The work it covers
- Buyer-question answer pages, written so the answer is on the page rather than implied by it.
- Article briefs built around the questions buyers actually ask.
- Machine-readable identity, so the right company comes back for the firm's own name.
Can you be trusted?
Whether your claims and your experience can be corroborated somewhere other than your own sales copy.
The work it covers
- Case studies with a stated baseline, the work done, and the measured outcome.
- Proof that survives a sceptical reader: labelled findings, named limits, no invented numbers.
- Consistent identity across independent sources.
Can you be chosen?
Whether the page a buyer lands on helps them take a truthful, usable next step.
The work it covers
- Website builds for specialist firms, where the site has to carry technical credibility rather than decorate it.
- Interactive tools that qualify buyers before they make contact, including pricing and scoping calculators.
- Enquiry paths that work on a phone: forms, trust signals, calls to action.
Behind those four sit nine disciplines and 40 checks. They are in the report if you want them. The four are what most people need.
A finding
What a finding looks like.
Every check comes back with a condition and an evidence label. Both are on the page, next to the finding, not buried in an appendix.
Vocabulary one · condition
Five values, and they run in a severity order. Read left to right.
Least severe
Healthy
Checked, and nothing needs doing.
Needs attention
Working, but costing you something.
Most severe
Critical
Losing you enquiries now.
Outside the order
Not assessed
A real answer. It appears when it is the honest one.
Outside the order
Not applicable
The check does not apply to your firm.
Vocabulary two · evidence label
Seven values, and they have no order. This is provenance, not quality.
Measured
Came from data we took.
Observed
Seen directly, once, in a stated state.
Partly checked
Some of the scope, not all of it.
Not checked
We did not run it, and we say so.
Not knowable
No access or no data exists to run it.
Client stated
You told us, and we have not verified it.
Analyst note
A judgement, held to the same visibility.
That second list is the part worth reading twice. A finding labelled measured came from data. A finding labelled analyst note is a judgement. You are told which is which, so you can weigh them differently. Nothing is dressed up as a number it is not.
The discipline matters past the report itself. A site that states its own project count three different ways on three different pages is not just untidy, it is a site an AI engine has learned to hesitate over: the same cross-referencing that catches an inflated claim also catches a contradicted one, and either way the citation goes to a source the model trusts more. A fixed method with one set of numbers, stated once and reused, is not just honest. It is the version machines can afford to repeat.
The receipt
A finding we could not close.
A real recheck, with the client and their property identifiers removed. The verdict is not closed. It is on the page because a framework that publishes a finding it could not close is worth more than one that only shows wins.
The original finding
A tracking tag runs on 56 of your pages and records nothing.
An analytics tag was installed on 56 of 57 pages. A tag-manager container was on all 57. The analytics account existed, but the data stream, the recording end, had never been set up.
Measured
What changed
The client replaced the property.
Artifexa did not implement this change and does not hold tag-manager access for this site. Old tag absent, new tag loading, container inert. Homepage only, pre-consent, one page of 57.
Observed
The verdict
The condition changed. The finding cannot be marked resolved on this evidence.
Tag detection proves readiness rather than recording, a rendered-browser observation is not an accepted source for the check, and one page cannot speak for 57. What would close it: analytics evidence of received events from the new property, and a page-set recheck at the original scope.
Not closed
The client changed the property after the review raised the finding. That is a sequence. This record does not claim the review caused the change.
The recheck
One review is a snapshot.
A snapshot proves nothing on its own. The same checks are re-run against your original baseline after work is done, and the report shows what changed, what did not, and what is still open.
Baseline, June 2026
80
Current audit
94
Next Layer Consulting scored 80 when first audited in June 2026. Fixes were re-scored against that baseline across six areas, from technical health and SEO essentials to performance and accessibility. The current audit is 94. Every figure measured against the same checks, on the same site, under three weeks apart.
The limits
What it will not do.
01
No score out of ten with no working shown.
02
No promised ranking.
Nothing here controls what a search engine does.
03
No reported measurement that was not taken.
Where a check could not be run, it says so and says why.
Next step
See the method applied to your firm.
The Visibility Review is free, written, and yours to keep. Findings arrive labelled the way they are labelled above, including the ones we could not close.