You were cited. Now you're not.

Before we change anything, we find out whether the drop is real. Most of the time, nobody asks.

Book a callCheck your AI visibility

The diagnostic

Order matters

Step one, always

Is the decline real, or is it two unlucky samples?

  • Your site changed

    A deploy, migration or robots.txt edit lines up with the drop

    Technical repair, days to weeks

  • Retrieval changed

    Nothing changed on your side; the engines cite different sources now

    Earn the new source, a quarter

  • A competitor won the source

    The page you were cited from now names someone else

    Placement or comparison content

  • Nothing changed

    The decline doesn’t survive repeated runs

    No remediation. We say so.

The order we work in

01 Why this exists

First: is it even real?

Four causes. One question that comes before all of them, and skipping it is how a quarter gets spent fixing a decline that was never there.

Answers vary between runs. A model can name you on Monday and not on Thursday with nothing having changed on your site, your competitor's site, or anywhere else. Retrieval is stochastic and the phrasing of the question moves the result.

So the first job is separating a genuine decline from noise, and it's the job most panic responses skip. Rewriting your homepage because of two bad screenshots is expensive and, more often than not, addresses nothing.

Once a decline is established, the causes fall into a short list: something changed on your site, something changed in what the engines can fetch, a competitor earned the source you were being cited from, or the source itself changed. Each has a different fix, and guessing between them wastes a quarter.

02 Definition

What is AI Visibility Recovery?

AI Visibility Recovery
AI visibility recovery is the diagnosis and repair of a decline in how often AI assistants cite a brand. It begins by establishing whether the decline is real, model answers vary between runs, so a drop observed once is often noise, and then traces the cause to one of four things: a change on the site, a change in what engines can fetch, a competitor winning the source, or a shift in retrieval. Each has a different repair, and a different realistic timeline.

  • From panic to baseline

    The first question is whether anything changed at all. Repeated runs across engines and phrasings answer it before a pound is spent on remediation.

  • From one cause to four

    A decline is a site change, a retrieval change, a competitor win, or nothing. Guessing between them wastes the quarter.

  • From asserted to demonstrated

    Recovery is measured against the baseline that established the decline, the same prompts, the same engines, the same schedule.

03 What’s included

Four things land. All of them are built, not filed.

  • 01

    Signal or noise

    Repeated runs across engines and phrasings to establish whether the decline is persistent, and how large it actually is. Sometimes the finding is that nothing is wrong, and we'll tell you that.

  • 02

    Change reconstruction

    What changed on your side and when, deploys, template changes, robots.txt edits, migrations, schema that stopped validating, lined up against the timeline of the decline.

  • 03

    Source-side diagnosis

    Which sources the engines now cite for the answers you used to appear in, and whether you lost the placement or the placement lost its authority.

  • 04

    The repair plan, and the repair

    A prioritised remediation, technical, content, entity or authority, with the work shipped rather than filed, in whichever order will move the needle first.

04 How it runs

In order, and in the open.

  1. Step 01

    Establish the decline

    Baseline runs across engines and phrasings. No remediation is scoped until we know the drop is real and how deep it goes.

  2. Step 02

    Trace the cause

    Your change history against the decline's timeline, plus a crawler-level read of the current site and the sources now winning the answers.

  3. Step 03

    Repair and re-measure

    Fixes shipped in priority order, then tracked against the baseline so recovery is demonstrated rather than asserted.

05 The difference

The usual response vs This engagement

 The usual responseThis engagement
First moveRewrite the pagesEstablish whether the decline is real
DiagnosisAssumed content problemTraced across site, retrieval and sources
ScopeEverything, at onceThe one cause, in priority order
ProofAsserted improvementRe-measured against the baseline

06 Who it’s for

Four markets, four different questions.

  • B2B software

    Buyers research through assistants long before a demo. The questions are comparative, alternatives, integrations, fit for team size, and the answer names two or three vendors.

  • DTC and ecommerce

    Category questions now start with an assistant. “Best X for Y” returns a shortlist, and being absent from it costs the consideration set, not just the click.

  • Healthcare

    Trust and provenance dominate. Assistants lean hard on sources they can attribute, which makes structured, citable content and clean entity signals disproportionately valuable.

  • NDIS and disability services

    Participants and coordinators ask assistants for local providers. The answers lean on directories, reviews and clearly structured service pages, which most providers have never marked up.

07 Questions

What buyers ask about this one.

Something not covered? Ask us directly and we’ll answer straight.

Why did my AI visibility drop?
One of four things: something changed on your site, something changed in what the engines can fetch, a competitor earned the source you were cited from, or retrieval shifted underneath you. Each has a different repair, which is why the diagnosis comes before the remediation rather than alongside it.
How do I know the drop is real?
Repeated runs across engines and phrasings. Model answers are non-deterministic, so a brand can be named on Monday and not Thursday with nothing having changed anywhere. A decline that survives repeated sampling is a decline; one that doesn't was noise, and remediating it costs a quarter for nothing.
Can lost citations come back?
Often, yes, the speed depends on the cause. A blocked crawler or invalid schema resolves within weeks of being fixed. A source a competitor now owns takes a quarter or more, because the repair involves earning a place on something that isn't yours.
Do you work with our existing agency?
Yes. Recovery is usually a diagnostic engagement rather than a replacement one. We establish the cause and hand the remediation to whoever is best placed to ship it, which is sometimes them and sometimes us.
What does the diagnostic cost?
It is scoped as a fixed piece of work rather than a retainer, because it has a defined end: you get the finding. If the finding is that nothing is wrong, that is the deliverable and the engagement stops there.
How fast can visibility come back?
Technical causes can resolve quickly. Once a blocked crawler is unblocked or broken schema validates, pages become eligible again within weeks. Authority and source causes take a quarter or more, because the fix involves earning a place on something that isn't yours. We'll tell you which kind you have before you commit to a timeline.
What if the cause turns out to be a competitor?
That's a common finding and it changes the work rather than ending it: the target becomes the source they earned. Sometimes that's a publication placement, sometimes it's a comparison page they wrote and you didn't.
What if you find nothing wrong?
Then the report says so and the engagement ends there. We'd rather return that finding than invent a remediation plan for a decline that was two unlucky samples.

Next step

Start with what your site already says.

The free checker fetches your site the way the assistants do and scores what comes back. It takes a minute, and it makes the first call a conversation about your findings rather than about our process.

Book a callRun the free check

08 The rest of the work