AI Discovery Intelligence
Most AI visibility tools tell you whether you show up, and stop there. We measure what ChatGPT, Claude, Gemini and Perplexity say about you, tie it to your traffic and cost, and do the work that moves it.
$9,000 for six months, one market. Written scope back within one business day.
Engines now name Northvale first in 45 of 160 open category answers, up from 35 last wave. Branded questions are already near ceiling at 62 of 64 - the gap is on questions where the buyer names nobody. The #1 move: answers that cite one of Northvale's own pages name Northvale; answers citing none mostly do not.
| Query | Sep 1 · Latest | ||
|---|---|---|---|
| 1 | best sustainable furniture shops in the uk? | ||
| 2 | i'm furnishing a flat in london, where should i buy? | ||
| 3 | where to buy a solid oak dining table in the uk? | ||
| 4 | is handmade furniture worth the extra cost? |
This is the intelligence, not a slide. Your three headline rates and the signals behind them, who is winning the queries you are losing, and what to do about it - live, in the tools your team already uses. Illustrative data for Northvale, our sample brand.
QUERY UNIVERSE
A category-sized query set across 4 engines, every response captured. Mapped to your intent classes - Direct Brand, Comparison, Discovery, Local, Purchase Intent, Competitive Displacement.
RETRIEVAL PROBES
Whether the engines can reach, render, and cite your content - the posture that decides if you are even eligible to be recommended. Most teams are blocked before the answer is written.
CITATION GRAPH
Which of your pages get cited, which get skipped, and which competitors sit ahead of you on the queries that matter. The deliverable your team builds against.
One method, run the same way every month - a locked baseline, and a clear read on what moved.
What the audit captures
Are you recommended?
Whether you come back as the answer, or anywhere at all.
Across which intents?
Discovery, comparison, local, purchase - where coverage holds or drops.
Who is cited above you?
The competitors the engines position ahead of you, per query.
Can engines reach you?
Whether your retrieval posture helps or blocks the engines.
What becomes part of your infra
Each layer builds on the one below and runs against the systems you already operate - AI-channel signal, CDP, paid feeds, analytics. Once live, it flags material moves in real time and refreshes monthly. Every claim is confidence-tagged and auditable back to its method.
Four engines in. Four layers. Three team-ready outputs.
Every engine response captured across your category - the measurement base every other layer feeds from. Reproducible from raw run data, so any number traces back to the response that produced it.
A per-URL citation graph across all four engines: which pages get cited, which get skipped, and where competitors sit ahead of you. The concrete targets engineering builds against.
The real AI-channel demand signal your JavaScript analytics can't see - genuine human-driven visits, separated out and joined into your existing stack. The feed is yours.
A quarterly, dollar-terms view of where AI demand can offset paid spend, by intent class and category - the line finance underwrites against. Treated as incremental, never one-for-one.
Priced per market. Every market gets its own panel.
For one brand that needs to win one market at a time. Panel build included.
Written scope back within one business day.
Several brands or markets as one programme, with the layers your data science team and CFO read.
Enterprise is scoped on a call. We send the calendar link.
Snapshot - $1,000, one brand, one market
Not ready for a programme? A preliminary baseline: 20 of the programme's 60 questions, 4 engines, 80 answers each read and scored, in your hands within 48 hours - and credited in full if you start a programme within 30 days.
Case study
“When people ask AI where to buy near them, Scoop went from 9% of answers to 32% in three months. On ChatGPT we now show up in more than half.”
Iryna Nestsiarovich CEO, Scoop Wholefoods UAE
Three months, identical locked panelOn ChatGPT: now over half of answers
Architecture
Between your first-party data feeds and your analytics. An outbound feed: no SDK, no dashboard, nothing in your rendering path.
We measure what AI says about you, then we do the work to improve it. Every monthly wave re-measures against the locked baseline.
Your existing infra
AI-channel signal · CDP · paid spend feeds
↓
28 Labs layer
AI Discovery Intelligence
Recommendation share · Citation graph · Demand read · Spend read
↓
Your existing surfaces
Web analytics · Exec dashboards · Budget instruments · Build targets
What it looks like operationally
Each team gets what it needs, in the tool it already uses.
URL-level page specs, schema and linking patterns, mapped to where citation share is most winnable. We deploy where you give us access; otherwise your team ships from the spec.
The real AI-channel demand, isolated and joined to your analytics, plus the raw run data to re-run or model. The feed is yours.
Per-engine share trajectory, the competitors named above you, the moves shipping next, and the delta since last wave. No new dashboards.
A quarterly, dollar-terms view of where AI demand can offset paid spend, by category and intent - a defensible underwriting line. Incremental, never one-for-one.
Outbound enrichment feed. No SDK, no JS overlay, no vendor in the production codepath - the same surface as ingesting any data feed. Most security reviews already approve the pattern.
Training-data exposure. Verbatim user prompts. Non-clicked citations. Revenue attribution from logs alone (needs a CDP join). We name our limits before your data science team asks.
FAQ
AI Discovery Intelligence is the measurement and engineering of how AI answer engines (ChatGPT, Claude, Gemini, Perplexity) recommend companies in a given category. You start with an AI Visibility Audit: a category-level read of where you are recommended, who is cited ahead of you, and what it costs you - delivered as an output your teams can act on.
SEO optimizes for ranked search results. AI Discovery Intelligence optimizes for the citation behavior of AI answer engines - what they retrieve, cite, and recommend when a buyer asks the engine directly. Different surface, different signal, different engineering.
Standard scope: a category-sized buyer-query set across 4 AI engines, every response captured, plus a read on whether the engines can reach and cite your content, with reliability-tiered findings (HIGH directly observed / MED statistically inferred / LOW industry intel). Enterprise audits extend to first-party AI-channel signal and a quarterly, dollar-terms read on where AI demand can offset paid spend.
We measure citation behavior - what AI engines show users when they answer category queries. We do not measure training-data exposure directly. Anyone claiming to measure how often your company appears in an engine's training corpus is overpromising; that signal is not reliably available externally.
Yes, and we measure it first. We measure your share of AI recommendations across ChatGPT, Claude, Gemini and Perplexity on a locked panel of buyer questions, then execute around a dozen ranked moves per market - deployed by us where you give us site access, drafted for your team to publish where you don't. Every monthly wave re-measures against the baseline, so you see whether the work moved the number.
Tell us your brand and market. Written scope and draft questions back within one business day. Want to see the output first? See a sample audit.
Get a proposal