AI Brand Audit: A Step-by-Step Checklist
Find out what AI says about you. Fix what's wrong. Here is the checklist.
AI Brand Audit
AI Brand Audit: A Step-by-Step Checklist
An AI brand audit tells you exactly what ChatGPT, Claude, Gemini, and Perplexity say about your brand right now. 47% of B2B buyers already use AI for purchase research. If you haven't checked what AI tells them, you're flying blind.
This checklist takes one working day. Five phases. No fluff. Print it, run it, and you'll know where you stand by end of day.
Why You Need This Audit
Gartner predicts that by 2028, 30% of web traffic from search engines will be replaced by AI-driven answers. That traffic has to come from somewhere. It's coming from you.
The problem: most brands have never once asked an AI about themselves. They optimize for Google and assume AI follows. It doesn't. AI pulls from a completely different set of signals, and what it says about you might be inaccurate, outdated, or simply nothing at all.
Run this audit before you build any AI visibility strategy. You can't fix what you can't see.
Phase 1: Manual AI Testing
Test what AI actually says. Not what you hope it says.
The Prompts to Run
Open each platform (ChatGPT, Claude, Gemini, Perplexity) and run these queries. Copy the responses into a spreadsheet.
Brand queries:"What is [your company]?"
"Tell me about [your company]"
"What does [your company] do?"
"Is [your company] any good?"
"What's the best [your category] tool?"
"Top [your category] platforms in 2026"
"Recommend a [your category] for [your ICP]"
"[Your company] vs [top competitor]"
"[Your company] vs [second competitor]"
"Alternatives to [your company]"
"Alternatives to [top competitor]"
"What do people say about [your company]?"
"[Your company] reviews"
"Problems with [your company]"
What to Record
For each response, document:
| Data Point | What to Record |
|---|
| Mentioned? | Yes / No / Partially |
|---|---|
| Position | 1st, 2nd, 3rd, or not listed |
| Accuracy | Correct / Outdated / Wrong |
| Sentiment | Positive / Neutral / Negative |
| Sources cited | What URLs or sources does the AI reference? |
| Competitors mentioned | Who else appears in the response? |
Run at least 15-20 queries per platform. That's 60-80 total data points. It sounds like a lot. It takes about 90 minutes.
Phase 1 Checklist
- ChatGPT tested (all 4 query types)
- Claude tested (all 4 query types)
- Gemini tested (all 4 query types)
- Perplexity tested (all 4 query types)
- All responses documented in spreadsheet
- Accuracy issues flagged
Phase 2: Source Audit
AI doesn't invent information. It reads sources. Check whether those sources help you or hurt you.
Sources to Audit
Wikipedia:- Do you have an article? Is it accurate?
- Are your competitors' articles better?
- Last edit date? (Stale articles get less weight)
- G2 profile complete and current?
- Capterra/GetApp listing accurate?
- TrustRadius profile claimed?
- Average rating across platforms?
- Number of reviews (more = more authority)
- Search your brand on Reddit. What comes up?
- Are mentions positive, negative, or nonexistent?
- Are you mentioned in recommendation threads?
- Does your homepage clearly state what you do?
- Is your pricing page current?
- Do you have FAQ pages with direct answers?
- Is your content structured with clear headings?
- Google News search for your brand. What appears?
- Any press coverage in the last 12 months?
- Industry publication mentions?
- GitHub presence?
- Stack Overflow mentions?
- Documentation quality?
A 2025 analysis found that 78% of AI citations trace back to just 5 source categories: Wikipedia, review sites, news outlets, Reddit, and the brand's own website. If you're weak on any of these, that's your gap.
Phase 3: Competitor Comparison
Run the same audit on your top 3 competitors. Same queries, same platforms, same spreadsheet.
What to Compare
| Metric | You | Competitor 1 | Competitor 2 | Competitor 3 |
|---|
| Mention frequency | _ / 20 | _ / 20 | _ / 20 | _ / 20 |
|---|---|---|---|---|
| Average position | ||||
| G2 reviews | ||||
| Reddit mentions | ||||
| Wikipedia article? | Y/N | Y/N | Y/N | Y/N |
| Schema markup? | Y/N | Y/N | Y/N | Y/N |
| FAQ pages? | Y/N | Y/N | Y/N | Y/N |
Phase 3 Checklist
- Competitor 1 audited across all platforms
- Competitor 2 audited across all platforms
- Competitor 3 audited across all platforms
- Comparison table completed
- Key gaps identified
The competitor data matters more than your own data. If Competitor 1 gets mentioned 18/20 times and you get mentioned 4/20, that's a 4.5x gap. You now know the target.
Phase 4: Content Gap Analysis
What topics does AI associate with your category that you haven't covered?
How to Find Content Gaps
| Content Type | Do You Have It? | Priority |
|---|
| "What is [your category]" explainer | Y/N | High |
|---|---|---|
| Comparison pages vs competitors | Y/N | High |
| How-to guides for common use cases | Y/N | High |
| FAQ page with 20+ questions | Y/N | High |
| Pricing page with clear tiers | Y/N | Medium |
| Case studies with specific results | Y/N | Medium |
| Integration/compatibility pages | Y/N | Medium |
| Industry-specific landing pages | Y/N | Low |
According to Princeton's GEO research, content with statistics gets 41% more AI citations. Content with direct quotations gets 28% more. If your content is opinion without data, AI has nothing concrete to cite. For more on structuring content AI actually extracts, see our content strategy guide.
Phase 4 Checklist
- Topic list from AI responses compiled
- Content inventory completed
- Gaps identified and prioritized
- Content calendar drafted for top 10 gaps
Phase 5: Technical Check
The plumbing matters. AI crawlers need access, and structured data helps them understand what they're reading.
Technical Checklist
Crawl access:robots.txt allows AI crawlers (GPTBot, ClaudeBot, Google-Extended, PerplexityBot)- No accidental blocks on important pages
- Sitemap.xml exists and is current
- Organization schema on homepage
- Article schema on blog posts
- FAQ schema on FAQ pages
- Product schema on product/pricing pages
- BreadcrumbList schema for navigation
llms.txt file exists (emerging standard for AI readability).well-known/ai-plugin.json configured- Content is in crawlable HTML (not locked in JavaScript that fails without rendering)
- Page load time under 3 seconds
- Mobile-friendly
- No broken links on key pages
- HTTPS everywhere
For a deep dive on schema implementation, see our schema markup guide.
What to Do With Your Results
You now have data. Here's how to use it.
Prioritize by Impact
Fix immediately (this week):- Inaccurate information in AI responses (trace to source, update)
- Missing or broken technical setup (robots.txt, schema)
- Empty profiles on review platforms
- Content gaps for high-priority topics
- Competitor gaps where you're significantly behind
- Missing FAQ sections on key pages
- Review generation campaigns
- Press and publication outreach
- Comprehensive content strategy
Set Your Baseline
Your audit results are your baseline. Record:
- Total mention rate across platforms: ___%
- Average position when mentioned: ___
- Accuracy rate: ___%
- Sentiment breakdown: ___% positive, ___% neutral, ___% negative
Come back in 30 days and run the same queries. That's how you measure progress.
Automate the Ongoing Work
Manual audits work for a baseline. They don't scale for ongoing monitoring. Renown tracks your AI visibility across platforms continuously, so you know the moment something changes.
FAQ
How often should I run an AI brand audit?
Full audit quarterly. Spot-check key queries weekly. AI models update constantly. What was true last month might not be true today.
How long does a complete audit take?
One person, one day. About 2 hours for Phase 1 (manual testing), 1 hour each for Phases 2-4, and 1 hour for Phase 5. Total: 6-7 hours.
Do I need special tools for this?
Not for the initial audit. A browser and a spreadsheet. For ongoing monitoring, tools like Renown automate the manual testing across all platforms.
What if AI says something wrong about my brand?
Trace it to the source. AI gets information from somewhere. Find the outdated review, the wrong Wikipedia entry, or the inaccurate website copy. Fix the source and the AI output eventually updates.
Should I audit all AI platforms or just ChatGPT?
All of them. ChatGPT has the most users, but Claude, Gemini, and Perplexity each pull from different sources with different weights. Your share of voice varies by platform. Being invisible on one while visible on another is still a problem.
What's the most common finding in AI brand audits?
Absence. Most brands simply aren't mentioned. The second most common: outdated information from 1-2 years ago that nobody bothered to update. Both are fixable.
Resources
Related Guides
AI Visibility for Agencies: The Client Playbook
94% of CMOs are increasing AI visibility investment. Your clients need this. Here's how to build and sell the service.
AI Visibility for Digital PR: From Press Releases to Machine Relations
Press coverage used to influence journalists. Now it influences machines. Every article about your brand becomes a citation source for AI. Here's how to make that work.
AI Visibility for E-Commerce Brands
ChatGPT has a shopping feature. Perplexity has a Buy button. AI is recommending products. Yours might not be one of them.
