How to Track Competitors in ChatGPT and Claude
Most US marketing teams have no idea which competitors are winning in AI search. Here is the exact framework to track, benchmark, and outpace them.
Your competitors are being recommended by ChatGPT, Claude, and Perplexity to buyers who will never see your brand — and most marketing teams have no idea it's happening. Tracking AI search competitors is not optional in 2025; it is the only way to know whether you are winning or losing the channel that is reshaping B2B and B2C purchase decisions across the US market.
Why Competitor Intelligence Looks Different in AI Search
In traditional SEO, competitor tracking is well-understood. You monitor keyword rankings, estimate organic traffic, audit backlink profiles. The data is granular, tool-supported, and updates daily.
In AI search, the rules are different. There are no keyword positions. There is no rank tracker that shows you "ChatGPT position #2." Instead, you are measuring share of voice — how often your brand appears relative to competitors across a representative set of buyer queries. The model either mentions you or it doesn't. And it does or doesn't for structural reasons that you can diagnose and act on.
AI competitor intelligence is not about monitoring what your competitors publish. It is about understanding what AI models have learned to believe about them — and why. That belief is built from third-party reviews, editorial coverage, community presence, and entity consistency. Track those signals and you understand the gap.
Step 1 — Build Your Competitor Query Set
Before you can track competitors, you need a representative query set. This is the foundation of everything. A bad query set produces misleading data; a good one tells you exactly where the competitive battles are happening.
- 1Define Your Query Categories
Group buyer queries into four types: awareness queries ("what is [category]"), consideration queries ("best [category] tools for US companies"), comparison queries ("X vs Y"), and intent queries ("which [category] platform do experts recommend"). You need coverage across all four — different competitors dominate at different stages of the funnel.
- 2Write 50 Non-Branded Prompts
Write queries exactly as a buyer would type them into ChatGPT or Claude — conversational, specific, without using any brand names. Include US-market context where relevant ("for a mid-size US company," "recommended by US enterprise teams"). Avoid industry jargon your buyers wouldn't use. These 50 prompts become your permanent benchmark set.
- 3Validate the Query Set Against Real Buyers
Before locking in your query set, validate it against three to five real customers. Ask them: "What did you type into ChatGPT when you were evaluating solutions like ours?" Replace synthetic prompts with verbatim buyer language wherever possible. Validated queries produce benchmark data that reflects actual purchase intent.
what is AI search visibility and why does it matter for B2B brands
ChatGPT, awareness query
which AI visibility monitoring platforms do enterprise marketing teams use in the US
Claude, consideration query
Step 2 — Run the Benchmark Across All Platforms
Run your 50 prompts across ChatGPT (GPT-4o), Claude (Sonnet), Perplexity, and Google Gemini. Do this manually for the first benchmark — tool-assisted scraping comes later once you've validated the process.
For each prompt on each platform, record:
- ✓Every brand mentioned by name in the answer
- ✓The order brands appear (first mention carries the highest weight)
- ✓How each brand is described — positioning, tone, use case attributed
- ✓Whether your brand appears and how it is characterized
- ✓Sources cited (especially critical for Perplexity)
- ✓Any caveats or negative framings applied to specific brands
This produces a raw dataset of brand mentions across 200 data points (50 queries × 4 platforms). It takes three to four hours the first time. After that, weekly tracking of a subset takes under an hour.
Run each query in a fresh incognito session with no prior conversation history. ChatGPT and Claude personalize responses based on prior context — a contaminated session will skew your competitive data. Use separate browser profiles for each platform.
Step 3 — Calculate AI Share of Voice
Once you have raw mention data, convert it to share of voice metrics. This is the core competitive KPI for AI search.
AI Mention Rate for each brand:
Brand mentions ÷ Total queries run = Mention rate %
AI Share of Voice for each brand:
Brand mentions ÷ Total brand mentions across all competitors = Share of voice %
First-Mention Rate (premium metric):
Queries where brand appears first ÷ Total queries = First-mention rate %
The share of voice view immediately shows you the gap and who is winning. More importantly, it shows you who you are actually competing against in AI answers — which is often different from who you compete with on Google.
Step 4 — Diagnose Why Competitors Are Winning
A share of voice gap is a symptom. The cause is always one or more of four traceable signal advantages. Diagnosing the root cause tells you exactly where to invest.
For each competitor outperforming you, run this diagnosis:
- ✓Check their G2 and Capterra review count and average rating vs. yours
- ✓Search their brand name in Perplexity — note which publications cite them that don't cite you
- ✓Search "[competitor] reddit" and "[competitor] site:news.ycombinator.com" — measure community presence
- ✓Compare their LinkedIn, Crunchbase, and Wikipedia entity data against yours for consistency
- ✓Search "[competitor] vs [category]" — count how many comparison pages exist for them vs. you
- ✓Check if they are listed in G2 Grid reports or Gartner Magic Quadrant equivalents for your category
Do not copy competitor content strategies. AI models penalize thin, derivative content. Instead, identify which trust signals your competitors have built that you haven't — then build those signals more authentically and at greater depth.
Step 5 — Build a Competitive Response Roadmap
Once you know the gap and its root cause, you can build a prioritized roadmap. The goal is not to match competitors on every signal simultaneously — it is to close the highest-leverage gap first.
Prioritization framework:
| Gap Type | Time to Close | Effort | Action |
|---|---|---|---|
| Review volume deficit | 30–60 days | Medium | Customer review outreach campaign |
| Entity data inconsistency | 1–5 days | Low | Profile audit and canonical description deployment |
| Missing editorial citations | 60–120 days | High | Media outreach, contributed content, data studies |
| Community presence gap | 60–90 days | Medium | Sustained Reddit/HN participation, no pitching |
| Comparison content gap | 14–30 days | Low-Medium | Build honest competitor comparison pages |
| Analyst report absence | 90–180 days | High | G2 Grid nomination, analyst relationship building |
Step 6 — Establish a Weekly Tracking Cadence
Competitive AI visibility changes week to week. Competitors launch review campaigns. New editorial coverage shifts AI model citations. Your query set needs to be run on a fixed weekly schedule to detect signal changes before they compound into larger gaps.
- ✓Run 20-query subset of your benchmark set every Monday across all four platforms
- ✓Run the full 50-query set at the end of each month
- ✓Log all mention rate and share of voice data in a shared spreadsheet or BI tool
- ✓Flag any new competitor appearing in answers that wasn't in your original baseline
- ✓Track sentiment shifts — note if any competitor's description turns negative or gains a caveat
- ✓Review Perplexity citation sources monthly for new publications entering your category
- ✓Run a full competitor signal audit (reviews, coverage, entity data) quarterly
Weekly tracking is what separates brands that react to AI visibility changes from brands that cause them. The first time a competitor launches a review campaign or earns a major editorial feature, you will see it in the mention rate data within 30–60 days — early enough to respond before the gap becomes structural.
Frequently Asked Questions
How many queries do I need to get statistically reliable competitor data?
For most US B2B categories, 40–50 queries across four platforms (160–200 data points) produces reliable share of voice estimates. Fewer than 30 queries per benchmark creates sampling error that can misrepresent competitive position by 10–15 percentage points. For highly fragmented categories with many competitors, expand to 60–70 queries to ensure adequate coverage of consideration-stage and comparison queries.
Should I track competitors on all AI platforms or focus on one?
Track all major platforms — ChatGPT, Perplexity, Claude, and Gemini — but weight your attention by where your buyers actually spend time. For US enterprise B2B, ChatGPT and Perplexity account for the majority of AI-assisted research. For consumer categories, add Copilot. Different competitors often lead on different platforms, so single-platform tracking will give you an incomplete picture and may cause you to misattribute share of voice movements.
What if a competitor suddenly appears in AI answers who wasn't there before?
A new entrant in AI answers almost always signals one of three things: a significant increase in review volume in the past 30–60 days, a major editorial feature in a publication AI models cite heavily, or a viral community discussion (Reddit, Hacker News) that generated concentrated training signal. Run the diagnostic checklist immediately to identify which signal changed, then assess whether and how to respond.
Can I use automated tools to run AI competitor tracking queries?
Yes, but with important caveats. Automated querying via API (ChatGPT API, Claude API) produces responses drawn from the same models but in a different context than consumer interfaces. For baseline tracking and trend detection, API-based automation is efficient and reliable. For nuanced sentiment and source citation analysis, manual review of consumer-facing responses is still necessary because the interface affects how sources are presented and how responses are formatted.
How do I handle a competitor with a much stronger AI presence built over years?
Focus on the signals where the playing field resets most quickly: review recency (a fresh review campaign can neutralize an older competitor's volume advantage within 60 days), entity consistency (many established brands have inconsistent profiles across platforms despite their age), and comparison content (you can create honest comparison pages for any competitor regardless of their existing AI presence). Long-term training signal advantages erode as models retrain — recent, high-quality signals increasingly outweigh legacy presence.
Aeotics tracks AI brand visibility across 12 AI models, updated weekly. See how your brand compares →
