LLMs like ChatGPT, Gemini, Claude or Perplexity are becoming default discovery layers for many users. People now ask AI assistants:
- “What are the best places to buy affordable sports gear in Spain?”
- “Which software should I use for AI search optimization?”
Behind those natural-language questions there is a simple risk:
If your brand doesn’t appear – or appears in the wrong way – LLMs are quietly rewriting your market.
This article is about:
- How do you measure where you stand today?
- How do you run a practical, repeatable LLM visibility audit?
We’ll define the main KPIs, explain topics vs prompts, give you a 10-step checklist, and walk through an example audit using Decathlon in Spain, a brand with strong retail presence and high trust scores among Spanish consumers.
What is LLM brand visibility?
LLM brand visibility is the degree to which large language models:
- Mention your brand when users ask about your category or problems you solve
- Describe you accurately (products, positioning, values)
- Prefer your content as a source when generating answers
This visibility shows up across multiple surfaces:
- Chat interfaces (general-purpose LLMs, AI search tools)
- AI summaries layered on top of search results
- In-product copilots and assistants
An LLM visibility audit answers questions like:
- “Do we show up when people ask about our category?”
- “Are we positioned as a leader, a challenger, or not at all?”
- “Which sources do LLMs use to talk about us?”
LLM visibility signals & KPIs
We’ll work with six core KPIs. Together they form a visibility scorecard for your brand.
Share of Voice (SoV)
Definition
The percentage of mentions of your brand vs a defined competitor set in LLM answers for a given topic and prompt set.
Example
If you test 50 prompts about “sports retailers in Spain” and LLMs mention:
- Decathlon in 40 answers
- Three other retailers in 20, 15 and 10 answers respectively
Then Decathlon’s Share of Voice for that topic is:
- 40 / (40 + 20 + 15 + 10) ≈ 44%.
Visibility Score
Definition
A composite 0–100 score that combines:
- Share of Voice
- Average Rank (position of your mention)
- Coverage across topics
- Sentiment and quality of description
Think of it as the headline number you can track over time.
Average Rank
Definition
When LLMs list multiple brands or options, Average Rank is the average position where your brand appears (1 = first, 2 = second, etc.).
You calculate it only on answers where you are actually mentioned. See an example from Decathlon in Spain.
Brand Sentiment
Definition
The overall tone of how LLMs describe you:
- Positive (praise, strong recommendations)
- Neutral (matter-of-fact descriptions)
- Negative (criticisms, warnings, controversies)
You can code this manually or with additional models. Tacmind calculates automatically for you.
Citations
Definition
How often LLMs cite or link to your properties (website, docs, blog) when answering questions about topics where you appear.
- Total number of citations
- Citations per prompts
- Diversity of pages cited (home vs deep content)
Owned Presence
Definition
Within all citations and references to your brand, Owned Presence is the share that points to owned assets vs third-party sites (media, marketplaces, aggregators).
High Owned Presence means LLMs see your site and content as primary references, not just reseller pages or news coverage.
Topics vs prompts: how to scope your audit
Before you start testing, you need a clear scope.
Topics
Topics are the conceptual territories where you want visibility—usually aligned with your SEO and product strategy.
For a sports retailer in Spain, topics might include:
- “Sports equipment retailer Spain”
- “Affordable running gear”
- “Camping and outdoor gear for beginners”
- “Click & collect sports stores”
Each topic reflects what users care about, not your internal product catalog.
Prompts
Prompts are the exact questions or instructions you send to LLMs to probe those topics.
For the topic “Affordable running gear”, example prompts:
- “Where can I buy affordable running shoes in Spain?”
- “What are the most popular sports retailers in Madrid for budget-friendly running gear?”
- “Recommend three places to buy entry-level running equipment in Spain.”
Each topic should have a prompt set that:
- Covers different phrasings and levels of detail
- Includes both generic and location-specific questions
- Mixes informational and transactional intent
Rule of thumb
Start with 5–10 topics and 5–10 prompts per topic for your first audit. You can scale up later.
10-step LLM visibility audit checklist
Here is your practical framework: 10 steps from zero to a complete audit.
1. Define your objectives and markets
- Are you auditing one country or multiple?
- Are you focusing on brand-level presence or a specific product line?
- What decisions will this audit influence (content roadmap, PR, product positioning)?
2. Select priority topics
- Use your existing SEO keyword sets, content pillars and sales conversations to identify 5–10 topics.
- Prioritize those with high business impact (top-of-funnel awareness, key revenue categories).
3. Build your prompt library
For each topic:
- Write 5–10 prompts.
- Include variations: short vs long, casual vs formal, different personas (“as a beginner runner”, “for a family with kids”), location mentions.
- Store them in a structured format (spreadsheet or Tacmind workspace).
4. Decide which LLMs and configurations to test
- Choose 2–4 major LLMs or AI search tools your audience is likely to use.
- Decide on language and region for each test (e.g., Spanish / Spain).
- Keep configurations (web search on/off, temperature) as consistent as possible.
5. Run baseline tests and capture outputs
For each prompt in each LLM:
- Record the full answer, including any citations or link cards.
- Take screenshots where useful (for later QA and context).
- Note date, model version, and any special settings.
6. Tag mentions and attributes
Go through the answers and tag:
- Which brands are mentioned (including yours)
- The position (1st, 2nd, “also consider…”)
- Any descriptive sentences about your brand (for sentiment)
- Which domains are cited (owned vs third-party)
This is where a tool like Tacmind can automate a lot of the heavy lifting, but you can start manually.
7. Calculate your KPIs
For each topic and across the whole audit:
- Share of Voice
- Visibility Score (using your chosen formula)
- Average Rank
- Brand Sentiment split
- Citations and Owned Presence
We’ll discuss benchmark ranges in the next section.
8. Compare against competitors (and against expectations)
- Are you a market leader being treated like one—or as just another option?
- Are local challengers over-represented?
- Do LLMs recommend marketplaces and general retailers more than category specialists?
This contextualizes your numbers.
9. Identify gaps and risks
Examples:
- Strong SEO but low Owned Presence (LLMs mostly link to third-party sites).
- Decent Share of Voice but weak sentiment (mentions framed around discounts, not quality).
- Good branded coverage but low visibility on generic category prompts.
Each gap should map to specific AEO, GEO, SEO or PR actions.
10. Turn the audit into a regular scorecard
- Choose a cadence (quarterly or bi-annually).
- Re-run the same prompts and LLMs to track trends.
- Update topics and prompts as your strategy evolves.
LLM ecosystems change quickly—visibility audits must become an ongoing process, not a one-off project.
KPI benchmarks: what “good” looks like
These ranges are indicative and should be adapted to your category and ambition.
Share of Voice (SoV)
- Core home market leader: aim for 40–60% SoV on key topics.
- Challenger brand: 15–30% may be acceptable short-term; below that means LLMs barely acknowledge you.
Visibility Score (0–100)
Using a simple weighted model (e.g., 40% SoV, 30% Avg Rank, 20% Topic coverage, 10% Sentiment), you can interpret:
- 80+ – clear category leadership in LLMs
- 60–79 – strong presence with room to refine positioning/coverage
- 40–59 – visible but inconsistent; priority improvement zone
- <40 – effectively invisible or misrepresented
Average Rank
- For branded and product-specific prompts, you want to be #1 or #2.
- For generic category prompts, consistently appearing within the top 3–4 options is a strong signal.
Brand Sentiment
- Aim for >70% neutral/positive mentions.
- Negative share >10% deserves investigation: what issues are being surfaced (pricing, quality, ethics)?
Citations & Owned Presence
For prompts that include your brand name:
- You should see citations to your official domain in at least 60–70% of answers.
- If third-party sites dominate, that’s a signal to improve your AEO and GEO work: more authoritative, structured, answer-friendly content on your own domain.
Example: LLM visibility audit for Decathlon Spain
Let’s walk through a simplified, illustrative example using Decathlon in Spain—one of the most trusted retail brands among Spanish consumers and a leader in sports retail.
1. Context: why Decathlon?
- Over 2.1 billion euros in revenue in Spain in recent years, with online sales growing double-digit and representing more than 13% of total sales.
- Around 170+ stores and several logistics centres across Spain, confirming national physical presence.
- Ranked as one of the most trusted retail brands in the country, with a trust score above 80 in recent consumer studies.
Given this offline and online strength, we would expect Decathlon to have high brand visibility in LLMs for sports retail topics.
2. Topics and prompts
Example topics for the audit:
- “Sports retailer Spain”
- “Affordable sports equipment”
- “Where to buy running shoes in Spain”
- “Best places to buy camping and hiking gear”
- “Click & collect sports stores in Spain”
- Branded: “Decathlon Spain online shopping”, “Decathlon returns policy”
For each topic, we create 5–10 Spanish-language prompts targeting Spain.
3. Hypothetical audit results (summary)
Imagine we test 60 prompts across several major LLMs.
Share of Voice (SoV)
- For generic “sports retailer Spain” prompts, Decathlon is mentioned in 48 out of 60 answers where any retailer is listed.
- Competitors appear in 30 and 22 answers respectively.
SoV ≈ 48 / (48 + 30 + 22) ≈ 44%
Average Rank
- Decathlon appears on average in position 1.6 across those answers—frequently first, occasionally second or third.
Brand Sentiment
- 75% neutral (“Decathlon is a popular sports retailer with stores across Spain”),
- 22% positive (emphasis on affordability and variety),
- 3% negative (comments around store queues or stock issues).
Citations & Owned Presence
- In branded prompts, LLMs cite decathlon.es or official Decathlon Spain pages in roughly 70% of answers; the rest point to news articles or general ecommerce directories.
Visibility Score
Using our example weighting, Decathlon might score around 78/100 for Spain:
- Strong SoV,
- Good average rank,
- Solid but improvable Owned Presence.
4. Interpretation
- Strengths
- LLMs recognize Decathlon as a default answer for sports retail in Spain.
- Descriptions are mostly neutral-to-positive, aligning with its strong offline trust indicators.
- Risks / Opportunities
- For specialist prompts (“high-performance trail running gear”, “premium ski equipment”), SoV drops and other brands take the lead.
- Owned Presence could be higher—LLMs sometimes lean on generalist comparison sites instead of Decathlon’s own guides.
This would lead to two main action tracks:
- Strengthen topical authority and AEO around advanced sports categories (trail running, technical outdoor gear).
- Create more answer-friendly, structured content on decathlon.es so LLMs naturally cite Decathlon’s own resources instead of third parties.
Turning audit insights into action
Once you have your KPIs, you need to convert them into a roadmap.
If Share of Voice is low
- Create or improve pillar content for underperforming topics.
- Run GEO-style initiatives (thought leadership, frameworks, educational content) so LLMs have more high-quality material about you.
If Average Rank is poor
- Analyze answers where you appear:
- Are you buried in long lists?
- Do LLMs describe you as “one of many options” rather than a go-to choice?
- Adjust your positioning and messaging so your differentiators are clearer in your owned content.
If Sentiment is weak
- Identify recurring issues surfaced by LLMs (pricing, quality, ethics).
- Coordinate with CX, product and PR to address real problems, not just the narrative.
- Publish transparent, helpful content about how you handle those issues.
If Owned Presence is low
- Improve AEO basics on your own site: clear answers, FAQs, structured data.
- Ensure key information (policies, guides, category explanations) lives on crawlable, optimized pages.
- Consider digital PR to consolidate references to those owned assets.
Tools like Tacmind can track these KPIs over time and connect them back to specific content and actions—so your LLM visibility work doesn’t stay theoretical.
FAQs
Can I run an LLM visibility audit manually?
Yes. For a first pass, a spreadsheet and screenshots are enough. It’s time-consuming but valuable. As your topics and markets grow, you’ll want a dedicated workflow (where Tacmind can help) to automate prompt runs, tagging and KPI calculations.
How often should I audit my brand on LLMs?
For most brands, twice a year is a good starting point. If you operate in a fast-moving category or rely heavily on inbound education, consider quarterly audits.
Should I test every LLM on the market?
No. Focus on:
- 2–4 major LLMs your audience is likely to use
- AI search products active in your key markets
Depth is more valuable than breadth—you want consistent, comparable data.
How do I measure sentiment objectively?
You can:
- Use a simple three-label system (positive / neutral / negative) and have two reviewers tag answers; resolve disagreements.
- Or use a separate sentiment model as a first pass and manually QA edge cases.
The important part is consistency over time, not perfect precision.
Are these KPIs standard in the industry?
LLM visibility is a new discipline, so there is no universal standard yet. The KPIs we’ve defined—Share of Voice, Visibility Score, Average Rank, Sentiment, Citations and Owned Presence—are designed to be practical, explainable and comparable across audits. You can adapt weights and thresholds to your own context.
How do AEO and GEO relate to this audit?
- The audit tells you where you are now.
- AEO gives you tactics to improve how often your content is selected as an answer.
- GEO helps you shape how generative engines talk about your brand and frameworks.
Together they form a full loop: measure → optimize → re-measure.
Conclusion: Make LLM audits a regular habit
LLMs are already influencing how users discover, compare and trust brands. If you don’t know how your brand appears in those answers, you’re effectively flying blind in a new search channel.
By:
- understanding the key KPIs,
- structuring topics and prompts thoughtfully, and
- following the 10-step audit checklist (with concrete, interpretable metrics),
you turn LLM visibility from a vague concern into a measurable, improvable part of your marketing strategy.
From here, a practical next move is to:
- Choose 5–10 priority topics for your brand.
- Build a first prompt set and run a pilot audit.
- Use Tacmind to turn that pilot into a repeatable GEO/AEO measurement system that keeps you visible, accurate and competitive in the age of AI search.
Was this helpful?
