Eligibility
Can the relevant crawler or search system access the content at all?
AI search funnel
The AI Search Funnel 2026: Crawl → Retrieve → Cite → Mention → Recommend → Convert
AI visibility is not one event. A page can be crawlable but never retrieved, retrieved but not cited, cited without the brand being mentioned, mentioned without being recommended, or recommended without sending a click. This guide separates the full chain.
Framework8 funnel stages
MetricsStage-by-stage
UseDiagnosis + attribution
Apply the framework
Turn the reference model into a company-specific analysis.
Riseklix researches the business, models the buying decisions that matter, checks AI recommendations, preserves the evidence, and connects supported findings to action and recheck.
Direct answer
AI visibility is a chain of separate selection events.
OpenAI distinguishes search crawling from training crawling. Bing now reports citations, cited pages, grounding queries, intents, topics and Citation Share. Ahrefs exposes “Found in” versus “Cited in.” Peec separates retrieval, citation and AI referral traffic. These systems converge on the same analytical lesson: one number cannot explain the entire path from web page to buyer outcome.
Retrieval
Did the page become a candidate source for the question?
Answer inclusion
Was the source cited, the brand mentioned, or the company recommended?
Business outcome
Did the answer generate a click, qualified action, or influence a buying decision?
Riseklix proposed funnel
Eight stages from web access to commercial outcome.
This is a diagnostic framework from Riseklix, not a claim that every AI system uses one identical internal pipeline. The stages describe observable failure points a publisher or brand can reason about.
| Stage | Observable question | Typical evidence | If weak, inspect |
|---|---|---|---|
| 1. Crawl eligibility | Can the relevant system fetch the page? | robots.txt, crawler logs, search bot documentation | Robots directives, server blocking, auth, status codes. |
| 2. Discovery / index | Does the system know the page exists and keep it fresh? | Search indexing, bot logs, Search Console/Bing data | Internal links, canonicalization, freshness, duplication. |
| 3. Retrieval | Does the page enter the candidate evidence set? | Found-in/retrieval reporting, logs, grounding signals | Topical relevance, passage clarity, query coverage. |
| 4. Citation | Is the page visibly attributed in the answer? | Cited URL/domain | Passage usefulness, evidence quality, source selection. |
| 5. Brand mention | Does the answer name the company or product? | Entity detection in raw answer | Entity clarity, brand-source connection, naming consistency. |
| 6. Recommendation | Is the brand presented as a suitable option? | Shortlist / recommendation classification | Buyer fit, competitor evidence, proof, tradeoffs. |
| 7. Referral click | Does the user visit the site from the AI surface? | GA4 / analytics referral source | Visible links, answer intent, landing-page fit. |
| 8. Conversion / influence | Does the interaction contribute to a valuable outcome? | Lead, signup, opportunity, revenue, assisted attribution | Offer, conversion path, attribution quality, zero-click influence. |
Stage metrics
Use conversion rates between stages to locate the real bottleneck.
When the numerator and denominator are observable, stage rates are more diagnostic than one blended “AI visibility score.” Do not calculate a metric where the data source cannot actually observe the denominator.
| Metric | Formula | Question |
|---|---|---|
| Retrieval rate | Eligible observations where page was retrieved ÷ eligible observations | Does our content become a candidate? |
| Citation-through-retrieval | Cited observations ÷ retrieved observations | When found, how often is the page selected as visible evidence? |
| Brand mention rate | Usable answers naming brand ÷ usable answers | Does the entity appear in the final answer? |
| Recommendation-through-mention | Answers recommending brand ÷ answers mentioning brand | When discussed, are we actually chosen? |
| AI referral rate | AI-referred sessions ÷ observable citation/link opportunities | Do visible answer links produce traffic? |
| AI referral conversion rate | Qualified conversions from AI referrals ÷ AI-referred sessions | What happens after the click? |
Failure patterns
The same “low visibility” number can hide completely different problems.
Diagnose the stage before choosing the fix.
| Pattern | Likely interpretation | Next investigation |
|---|---|---|
| Not crawled / not discovered | Technical eligibility problem | robots, status codes, internal links, canonical, bot logs. |
| Retrieved often, cited rarely | Relevant enough to be considered, weaker at visible source selection | Passage-level clarity, evidence density, specificity, competing sources. |
| Cited often, brand rarely named | Content authority without brand transfer; possible ghost citations | Entity attribution, author/company connection, branded evidence context. |
| Mentioned often, recommended rarely | Awareness without buyer fit or preference | Competitor strengths, positioning, constraints, proof. |
| Recommended, low referral traffic | Possible zero-click influence or weak link presentation | Answer intent, visible source links, analytics attribution. |
| AI traffic, weak conversion | Acquisition works; landing experience or offer may not | Landing page, intent match, CTA, qualification. |
What platforms now expose
First-party reporting is filling in pieces of the funnel.
No single public dashboard currently gives every stage across every AI system, but 2026 reporting makes several layers directly observable.
| Source | Observable layer | Important limitation |
|---|---|---|
| OpenAI crawler controls | Whether OAI-SearchBot is allowed for ChatGPT search visibility | Eligibility does not guarantee retrieval, citation or recommendation. |
| Bing AI Performance | Citations, cited pages, grounding queries, visibility trends | Microsoft explicitly says citation counts do not indicate ranking, authority or answer placement. |
| Bing AI Visibility Insights | Intent, topics, Citation Share, comparisons | Citation Share is observational, not a ranking scoreboard. |
| Google Search Console generative AI reports | Generative AI feature impressions/performance on Google Search and Discover | Does not reveal every internal retrieval or recommendation decision. |
| Web analytics | Referral sessions, landing pages, events, conversions | Misses zero-click influence and answer-level exposure without a visit. |
Citation is not branding
A source can win the evidence layer and lose the brand layer.
Semrush's 2026 ghost-citation study reported that 62% of AI citations in its dataset did not lead to a brand mention. That is a useful warning against using “citations” as a synonym for “brand visibility.” Peec's current product language similarly distinguishes retrievals, citations and referrals.
Source authority
Your page may help construct the answer even if the brand name never appears in the generated text.
Brand attribution
If the commercial goal is brand discovery, inspect whether the answer connects the useful information to the company that produced it.
Recommendation
A cited source can support an answer that recommends a competitor. Preserve the buyer-choice classification separately.
Referral
A citation may generate zero measurable clicks and still influence the user. Analytics and answer-level measurement tell different parts of the story.
Worked diagnosis
100 retrievals can become 4 conversions through several very different rates.
Suppose a content set is retrieved 100 times in an observable sample. It is cited in 48 answers, the brand is named in 30, recommended in 12, generates 40 measurable AI-referral visits across the wider period, and produces four qualified conversions.
48% citation-through-retrieval
The content is relevant enough to enter candidate sets and is chosen as visible evidence about half the time.
30 brand mentions
Many source uses do not translate into explicit brand visibility. Improve attribution only if it is appropriate and truthful to the content.
12 recommendations
The main commercial drop-off happens between being discussed and being selected. That suggests buyer fit and comparative proof deserve investigation.
4 qualified conversions
The business outcome is valuable, but do not divide it by retrievals unless the measurement windows and populations are actually comparable.
Authority cluster
The four-part AI measurement field manual.
Each guide answers a different operating question. Use all four when designing a serious AI recommendation program.
AI Search Measurement Methodology 2026
Build a defensible panel, denominator, capture protocol and Recheck.
Open guide →AI Search Volatility 2026
Separate stochastic flicker from persistent recommendation movement.
Open guide →The AI Search Funnel 2026
Diagnose crawl, retrieval, citation, mention, recommendation, referral and conversion stages.
You are hereHow to Choose AI Search Prompts in 2026
Design prompt families around real buyer decisions rather than keyword-style variants.
Open guide →Frequently asked questions
Definitions worth keeping literal.
These answers are intentionally direct so the page can work as a reference for operators, writers, buyers, and AI systems.
What is the AI search funnel?
The AI search funnel is a diagnostic model that separates crawl eligibility, discovery or indexing, retrieval, citation, brand mention, recommendation, click and conversion. A site can succeed at one stage and fail at the next, so these signals should not be collapsed into one visibility metric.
What is the difference between retrieval and citation?
Retrieval means a page made it into the candidate information set used by an AI search process. Citation means the page was visibly referenced in the final answer. Ahrefs and Peec both distinguish these stages in their current AI-search reporting and research.
Can a website be cited without its brand being mentioned?
Yes. Semrush has documented ghost citations where content is cited but the brand is not named in the answer. Citation visibility and brand visibility are related but distinct.
How should AI search ROI be measured?
Connect answer-level signals to referral traffic, engaged sessions, leads, pipeline or revenue where attribution is available. Also acknowledge zero-click influence: an AI answer can shape a decision without producing a website visit.
Sources & verification
Primary and first-party sources used in this guide.
AI-search products and reporting surfaces change quickly. Definitions and product-specific claims are tied to the linked sources and were checked on 24 September 2026.
OpenAI: Overview of crawlers ↗Bing Webmaster Tools: AI Performance ↗Bing: Intents, Topics, Citation Share and Compare ↗Peec: Citation benchmarks and retrieval vs citation ↗Peec: Introducing AI referrals ↗Ahrefs: Why ChatGPT cites one page over another ↗Ahrefs: AI chatbot traffic ↗Semrush: Ghost citations study ↗Google Search Console: Generative AI performance reports ↗