Recommendation Rate (Top 5): How to Win More Shortlist Links in AI Answers
Jump to
Recommendation Rate (Top 5) shows how often your brand appears in the Top 5 recommendations with a clickable link to your site in AI answers. Use it to measure shortlist wins that can drive visits, sign-ups, and revenue.
Definition (short)
Recommendation Rate (Top 5) is the percentage of buyer questions where your brand appears in the Top 5 recommendations and includes a clickable link to your site.
This page is a practical playbook. For the canonical definition and rules, use: Glossary → Recommendation Rate.
What counts (linked-only standard)
What we count
Your brand appears in the Top 5 recommendations in the answer.
The answer shows a clickable link to your site (any relevant page).
The link is visible as shown in the provider interface.
What we do not count
Brand mentions without a clickable link to your site.
Mentions outside the Top 5 recommendation format.
Runs where links are not visible (linked outcomes are not observable for that run).
Measurement rules that keep results valid
Locale is set at the topic level (country + language) and stays fixed. If you need two countries, create two topics.
Prompts are sent in the topic language, and the selected country/location is applied in provider requests.
We rely on repeated runs and aggregated results because AI answers can vary.
Read more:Locale controls · Link measurement rules
How it’s calculated
Recommendation Rate (Top 5) is calculated as:
Recommendation Rate (Top 5) = (Buyer questions where you appear in Top 5 with a link) / (Total buyer questions measured) × 100
Important: We evaluate this on aggregated runs, not a single answer snapshot. This reduces one-off noise.
How to read Recommendation Rate (Top 5)
This metric answers one question: “How often do we win the shortlist in AI answers, with a link that can send traffic?”
Interpretation table
If Recommendation Rate is… | It usually means… | Do this next (one move) |
Rising | You’re being shortlisted more often with links | Identify winning pages and replicate the format |
Flat | Visibility is stable but not improving | Ship one “shortlist asset” for the highest-value topic |
Falling | Competitors or sources take shortlist spots | Open the dropped buyer questions and audit linked pages |
High, but Average Rank is low | You show up often, but not near the top | Add clearer decision criteria and trade-offs |
Low, but Citation Coverage is high | You get cited, but not shortlisted | Add “best for” clarity and direct comparisons |
Diagnose in 10 minutes
Use this quick workflow to find the main reason your Recommendation Rate is low. Keep the same topic locale (country + language) while diagnosing.
Pick 5 buyer questions where you are not in the Top 5 with a link.
Open the AI answers and list the Top 5 recommendations.
Copy every visible link (your pages, competitor pages, third-party pages).
Classify what is winning: comparison pages, proof pages, topic hubs, directories, media, forums.
Choose the main gap: clarity gap, proof gap, or match gap (see Glossary for full definitions).
Decide one fix (one page or one section update) for the highest-impact question.
Fast “gap” guide
Gap type | What you see in answers | Best fix |
Clarity gap | AI can’t tell who you’re best for | Add “Best for / Not for” + 3–7 criteria bullets |
Proof gap | Competitors have evidence; you don’t | Add proof page: case, numbers, limits, methodology |
Match gap | Competitor pages match intent better | Publish a comparison (“vs”) or alternatives page |
How to improve Recommendation Rate (Top 5) (7-step playbook)
This playbook is designed to improve “shortlist wins” without bloating your site. Ship one change, then re-measure in the same topic locale.
Step 1: Choose one topic and lock one locale
Pick a revenue-critical topic. Use one fixed country + language. If you need another market, create a second topic.
Step 2: Find “lost shortlist” buyer questions
Identify questions where competitors appear in Top 5 with links and you do not. These are the highest-leverage targets.
Step 3: Audit what AI is linking to (page-level)
List competitor pages that win links.
List third-party pages that appear as sources.
Note which page formats dominate (comparison, proof, hub, directory, media).
Step 4: Ship one “shortlist asset” (choose one)
Pick the smallest asset that matches the intent:
Proof page when trust is missing.
Comparison page when selection criteria is missing
Topic hub when coverage is missing.
FAQ block when objections block the decision.
Step 5: Add shortlist signals (quick on-page checklist)
Best for: who you serve best (plain language).
Not for: who you are not a fit for.
Criteria: 3–7 bullets buyers use to choose.
Proof: one case, one metric, or one clear constraint.
Next step: pricing, booking, trial, or consultation.
Step 6: Build “vs” and “alternatives” coverage
Recommendation Rate often improves when you publish:
One “X vs Y” page with criteria and trade-offs.
One “Best alternatives to X” page that explains when to choose each option.
Step 7: Re-measure using the same buyer questions
Use the same prompt set, the same topic locale, and the same output format. Track the delta in Recommendation Rate over time.
Action mapping (if X, do Y)
If you see… | Likely reason | One action to ship | Metric to watch |
You are never in Top 5 | Low intent match or missing coverage | Create one topic hub page for the highest-value intent | Recommendation Rate |
You appear, but low in lists | Weak decision structure | Add criteria + trade-offs section to key page | Average Rank |
Competitors win with proof pages | Proof gap | Publish one proof page with claims + limits | Recommendation Rate |
Third-party sites dominate sources | External reinforcement favors others | Pick 2–3 target source pages and distribute content there | Citation Coverage |
Example: dentistry (simple and practical)
Let’s say a dental clinic wants more bookings from AI answers in one market (one country + one language). The goal is not “mentions.” The goal is appearing in Top 5 with a link users can click.
Buyer questions (examples)
“Best dental clinic for Invisalign in Kyiv”
“Top dentists for emergency tooth pain near me”
“Dental implant clinic vs dental surgery center”
“Alternatives to braces for adults”
What to publish to improve Recommendation Rate (Top 5)
Buyer intent | What AI tends to shortlist | What to publish |
“Best for Invisalign” | Clear “best for” + proof | Invisalign page with outcomes, pricing, fit, and constraints |
“Emergency near me” | Location + service clarity | Emergency dentistry page with steps, availability, and booking |
“Implants vs surgery” | Comparison criteria | “Implants vs oral surgery” page with trade-offs |
“Alternatives to braces” | Option overview | “Clear aligners vs braces” guide with pros/cons and who each fits |
Quick win
Add a Best for section to the Invisalign page. Add one proof element (case summary, outcome metric, or clear constraint). Then re-measure in the same topic locale.
Common mistakes
Chasing generic visibility instead of shortlist wins with links.
Mixing countries or languages in one dataset (use one topic per locale).
Publishing long content without decision structure (criteria, trade-offs, proof).
Making “best” claims without evidence or constraints.
Measuring once and assuming the result is stable (use repeated runs and aggregation).
FAQ
Does Recommendation Rate count brand mentions without links?
No. This metric is linked-only. A clickable link to your site is required.
Why Top 5 and not Top 3?
Top 5 is a common shortlist format in many answers. It keeps measurement consistent while still showing competition.
Can we improve Recommendation Rate without publishing new pages?
Sometimes. If you already have strong pages, adding “best for”, proof, and clear criteria can lift shortlist inclusion.
Why does the same question give different Top 5 results?
AI outputs can vary. That is why we measure with repeated runs and aggregated results.
What should we do if links are not visible in the provider?
Treat that run as not observable for linked outcomes. Use views or providers where links are visible for link-based measurement.
How is Recommendation Rate different from Average Rank?
Recommendation Rate is how often you appear in the Top 5 with a link. Average Rank is your average placement when you do appear in Top 5 lists.
Related metrics
Average Rank
AI Share of Voice (SOV)
AI Visibility Score (linked-only)
Citation / Source Coverage
Linked Page Ranking
Next steps
Pick one topic and lock one country + language.
Track 20 buyer questions for one week (multiple runs per day).
Ship one shortlist asset and re-measure in the same topic locale.
For the canonical definition and rules, see: Glossary → Recommendation Rate.
