ai visibility for developer tools
In this audit the answers named the client in 80% of category questions and 27% of use-case questions. We measure both, and type the sources deciding each.
Use-case questions sit outside the high-intent group, so they are in neither the headline share nor the score — 27% of 11 answers is three. Treat it as a fit signal to act on, not a scored metric.
use-case visibility and source typing
One anonymized audit of a WordPress filtering plugin, run the way every audit is run. The sources behind these answers are documentation, listicles and comparison articles — not the review platforms most visibility work assumes — and some of them are pages the vendor already owns.
The questions a buyer asks before they know what they need.
The questions a developer asks once they know exactly what they need. Same product, different answer.
33%cite your own domain
35%cite a page that mentions you
91%rank a competitor above you
The share of cited answers that link to a page on your own domain — named in the answer is not the same as anchored to your site. The ranking figure is measured across the 11 answers that presented an ordered list.
| page | cited in |
|---|---|
| home page | 13 answers |
| documentation FAQ — free vs pro | 7 answers |
| pricing page | 4 answers |
| documentation index | 3 answers |
| reviews page | 3 answers |
| documentation FAQ index | 1 answer |
| support-forum thread | 1 answer |
Documentation accounts for 11 of them. These pages exist and you own them, which is what makes them the tractable half of the source problem.
| source type | citations |
|---|---|
| listicle | 473 |
| comparison article | 259 |
| unknown | 214 |
| competitor site | 159 |
| official site | 134 |
| video | 51 |
| review platform | 49 |
| forum thread | 13 |
Review platforms are second from last — 49 citations against 473 from listicles, while official sites account for 134. A category whose answers are written from documentation and comparison articles needs a different play from one written from review profiles. “Unknown” is the classifier’s own bucket and is left in.
| competitor | mentions | share |
|---|---|---|
| FacetWP | 57 | 81% |
| WP Grid Builder | 44 | 63% |
| Filter Everything Pro | 35 | 50% |
| SearchWP | 33 | 47% |
| JetSmartFilters | 28 | 40% |
No client row: this is raw mention share across every valid answer, and the client's headline figure is qualified high-intent — a different basis, and one chart holding both would be a measurement error. On the one same-basis comparison the data supports — raw share of high-intent answers — the client holds 80% against the leader’s 81%.
A loop that fans queries across four providers is an afternoon. The parts that took us the longest are the ones that decide whether the number is true: excluding failed and off-topic answers before they count against you, judging whether a mention is a recommendation or a passing reference, typing 349 cited URLs by who owns them, and fetching pages to confirm a claim is on the page rather than in the answer. Without those, an in-house pipeline produces a number that moves for reasons nobody can name — and the person who built it now owns it every week.
methodology
This is the part most tools skip. An unfiltered mention count is easy to produce and easy to mislead with.
Without this gate a recommendation can name a page you are already listed on — because the claim came from the answer, not from the page.
Gate widths above are schematic — they show the sequence, not measured attrition. The only published figure is the final count of 70 valid answers in this sample. We won’t invent the rest.
where this sits
the foundation
Goal: be findable in a search index.
the new layer
Goal: show you a number that moved.
the implementation layer
Goal: change what the answer says.
the human layer
The pipeline measures. It does not decide what the measurement means.
The pipeline surfaces what moved; it can’t tell you whether the move matters. A four-point shift sits inside our measured run-to-run variance, and a person is the one who says that plainly instead of drawing a trend line through noise.
The same judgment applies to opportunities: the pipeline ranks a listicle by citation weight, but it takes someone reading the page to notice the client is already on it and the real opening is elsewhere.
And when a provider degrades mid-cycle, a person decides whether those answers count or are excluded — rather than quietly averaging them in.
score this cycle
48of 100
8 points Two runs of the same sample differ by up to this much.
A move inside the repeat-run range is reported as unchanged, not as a gain. Reported flat this cycle.
what the automation can and cannot do
The log shows what ran, when, and under which setting.
pricing
Every tier measures the same way. What changes is how much of the plan you carry out yourself.
30-day money-back guarantee. Start with the free mini-audit — no credit card, and your sample comes back within 1 business day.
If there isn’t a meaningful gap, we’ll tell you — no upsell. And the one-time full audit below is a single engagement: no subscription, nothing to renew.
one-time · no subscription
Full AI visibility audit, competitor gap analysis, and a 30-day action plan. The lowest-commitment way to find out whether there is a gap worth closing.
$750–$1,500
Start hereOngoing monitoring and execution
starter
$1,199/month
A done-for-you service — every cycle read by a person, not software you are left to mine yourself.
Also included
Every action in the plan arrives written and ready to use. If you would rather it were carried out for you, that is the Execution tier.
execution
$1,799/month
The plan does not just arrive — it gets done. You choose how much of it runs without you.
Also included
You set the autonomy. Fully automatic, or every action held for your approval — switchable per action, at any time.
custom
From $2,500+
Everything in Execution, plus the campaign work around the actions — and a scope built to your category.
Also included
answers
No. SEO measures whether you rank. We measure whether an assistant names you, cites you, or leaves you out of the answer entirely. The two are complementary, and the second is currently unmeasured at most companies.
The four audited surfaces are ChatGPT, Perplexity, Claude and Gemini, with more added over time. Provider health is checked every cycle and reported openly — if a provider failed during a run, you see that rather than a silently lower score.
What you buy from us is the monthly full report — the judgment about what changed materially and what to do next — plus a weekly pulse telling you what moved and whether it mattered. Dashboard access comes with every recurring plan, so the honest answer is not that we have one and they do not. The difference is what sits on top. The dashboard shows you what the numbers are. The weekly pulse tells you what moved and whether it matters. The monthly full report tells you what to do about it. And once the plan exists, the Execution tier can carry it out for you — action by action, or automatically. What a dashboard cannot do on its own is tell you whether a weekly move is real or just measurement spread; by our own variance work, that is the call that is hardest to make unaided.
Your company URL, your category, and two or three competitors. That is enough for the free mini-audit. For a full engagement we also want to know which segments matter most, so the query map reflects the deals you actually want.
No, and anyone who does is guessing. Visibility depends on sources we do not control. What we can do is show you exactly where you are absent, which sources are shaping that answer, and the moves most likely to change it — then measure whether they did.
A question asked by someone close to a decision: comparing vendors, asking for alternatives, or building a shortlist. “Greenhouse alternatives” and “best ATS for startups” are bottom-funnel. These are the answers worth being in.
Because assistants lean on them. Review platforms, comparison articles, listicles and community threads shape what gets recommended. Your own site is rarely the deciding input, which is why we map the sources rather than only auditing your pages.
Starter deliberately stops at knowing where to act — every action in the 30-day plan arrives drafted and ready, and you ship it. Execution ($1,799/month) carries those actions out for you, autonomously or holding each one for your approval, and integrates with your site and email so they can actually ship. The campaign work around the actions — outreach sequences, a content calendar, owning the publisher relationship, creator partnerships — is Custom.
You get a short visibility snapshot. If there is a meaningful gap we will show you what it would take to close it. If there is not, we will tell you that — there is no upsell attached.
Being cited and being cited for the right question are different things. In this audit the client's own pages appear across 32 distinct answers — documentation in 11 of them — while only 33% of cited answers link to the client's domain at all, and the use-case questions are where the client is thinnest. Owned pages that are already cited are the most tractable thing on the list precisely because they exist: the work is which questions they answer, not whether anyone will publish them.
All of them, typed by what they are. This audit recorded 1,352 citations across 174 domains and classified every one — listicle, comparison article, competitor site, official site, video, review platform, forum thread, and an “unknown” bucket we leave in rather than quietly reassign. Review platforms came to 49 of them. In a developer category that is the finding: the play is not a review-collection campaign, and a tool that only watches review sites would have reported almost nothing.
That is why the sampling is weekly rather than the report being one-off. Answers and cited sources shift, and the plan is measured against what actually moved rather than assumed to have worked. The honest limit is the other direction: this run has a measured repeat-run range of 8 points, so a move smaller than that is measurement spread, not a result, and a person says so rather than drawing a line through it.
free mini-audit
Pre-scoped as a use-case audit, not a category one: we run the implementation questions a developer actually types, against the alternatives that come back, on you and two of them. The form arrives filled in — change anything that is wrong for you.
If there isn’t a meaningful gap, we’ll tell you — no upsell.
We’ll come back with a short visibility snapshot. If there isn’t a meaningful gap, we’ll say so.