You don't need a $500/month tool to test AI brand visibility. A systematic manual testing protocol — run consistently every 30 days — gives you the core data to track progress and identify optimization priorities. Here's the exact protocol Magna uses with new clients.
Before you can optimize for AI visibility, you need to know where you stand. Most businesses have no idea whether they appear in ChatGPT, Perplexity, Gemini, or Claude — or if they do, how they're described, what position they hold, and what competitors are mentioned alongside them.
Paid tools like Profound and Otterly.ai automate this tracking at scale. But for businesses starting out — or anyone who wants to understand the data firsthand — the manual protocol below provides the same core insight at zero cost. Magna uses this as the first step with every new client.
Step 1: Develop Your Query Set (20–25 Queries)
Your query set should represent how your target customers actually prompt AI assistants — not how you wish they would. Most people ask AI for recommendations in one of three ways:
- "What [your service type] do you recommend in [your city]?"
- "Best [your service type] for [your target customer]"
- "Top [your service type] agencies in 2026"
- "Which [your service type] company should I use for [specific use case]?"
- "[Your service type] vs. [alternative] — what do you recommend?"
- "[Competitor name] or [your niche] — which is better for [use case]?"
- "I need [specific outcome] — which [your service] should I use?"
- "How do I [solve specific problem] — any recommendations?"
- "What's the best way to find a [your service type] for [specific scenario]?"
Document your full query set in a spreadsheet before testing. This ensures consistency across platforms and months.
Step 2: The Testing Procedure
For each query in your set, run the test across all four major platforms. Use incognito/private browsing mode to avoid personalization bias from previous sessions.
| Platform | Version to Test | Data Type | Key Note |
|---|---|---|---|
| ChatGPT | Free (GPT-4o) + Plus (with search) | Training data + live web | Results can differ significantly between versions |
| Perplexity | Default (web search) | Live web retrieval | Check sources cited — they reveal what's being retrieved |
| Google Gemini | Gemini 2.0 Flash/Pro | Training data + Google index | Most correlated with Google KG presence |
| Claude | Claude.ai (free) | Training data | More conservative with specific brand recommendations |
For each query-platform combination, document these seven data points:
- Date of test
- Exact query used
- Platform and version
- Brand presence — Yes/No, and if yes, what position (first, second, mentioned in passing)
- Exact description used — screenshot the exact text
- Competitors mentioned — list every brand that appeared
- Overall sentiment — positive, neutral, or negative framing of your brand
Step 3: Competitor Gap Analysis
For every query where a competitor appears and you don't, this is your highest-priority optimization signal. The analysis question is: what do they have that you don't?
Are they cited in trade publications you aren't? Industry blogs, news sites, research reports?
Do they have significantly more Google, Trustpilot, or industry platform reviews?
Do they have 30+ articles on the topic vs. your 3? More original data, case studies, research?
Do they have Organization, FAQ, or Article schema you're missing? Use Google's Rich Results Test.
Each gap you identify becomes a prioritized action item in your AEO program. The most common gaps are review volume, third-party media mentions, and content depth on specific subtopics.
Step 4: Tracking Trends Over Time
A single snapshot tells you where you are. Monthly tracking tells you whether your AEO efforts are working. Build a simple spreadsheet with these columns and run the protocol every 30 days:
Date | Query | Platform | Brand Present (Y/N) | Position | Sentiment | Competitors | Notes
Month-over-month, you're looking for:
- Queries where you weren't appearing that now show your brand
- Position improvements (mentioned third → mentioned first)
- Description improvements (vague → specific and accurate)
- Platform gains (appearing in Perplexity now even if not ChatGPT yet)
- Competitor movements (they gained or lost positions)
Step 5: From Data to Action
Every monthly test produces a prioritized action list. Use this decision framework to triage:
| Finding | Root Cause | Priority Action |
|---|---|---|
| Not appearing anywhere | Entity not recognized by AI | Entity foundation: GBP, Wikidata, Organization schema |
| Appearing in Perplexity, not ChatGPT | Training data gap, not real-time gap | Build third-party mentions for next training cycle |
| Described inaccurately | Entity data is unclear or inconsistent | Audit and align brand descriptions across all platforms |
| Mentioned third, competitors first | Competitors have more corroborating signals | Increase review velocity, media mentions, content depth |
| Not appearing for local queries | Local entity signals weak | GBP optimization, local citation building, LocalBusiness schema |
- Set a recurring calendar block on the 1st of each month: 2–3 hours for full protocol
- Maintain the same core 15 queries each month; rotate 5–10 queries to explore new territory
- Always use incognito mode — every platform, every test
- Screenshot every response, not just the ones where you appear
- Track competitors as rigorously as you track yourself
- Connect findings to your AEO action list within 48 hours of testing