Overview
Competitor research is one of those tasks that never gets done consistently — a Custom GPT changes that. This build wires ChatGPT's web browsing + a knowledge file of your own positioning + a strict output template. Point it at a URL, get a one-page teardown in the exact format your team already uses.
How it works
- GPT builder with browsing enabled and file search on a Knowledge file (your positioning doc).
- System prompt enforces the output template: 5 sections, no filler, cite the source URL for each claim.
- Browsing pulls landing page + /pricing + /changelog if present.
- Output is markdown; paste directly into Notion or Slack.
Benefits
- Weekly teardowns take 3 minutes instead of 45.
- Consistent format makes month-over-month diffs easy.
- Grounded in your positioning, so ideas are relevant.
Use cases
- Startup CEOs tracking 5-10 competitors monthly.
- PMMs preparing quarterly battle cards.
- Founders scanning YC batch launches.
Step-by-step guide
Step 1: Draft your positioning doc
One page: audience, promise, differentiators, prices, weaknesses. This is the file the GPT searches.
Step 2: Create the Custom GPT
Name it 'Competitive Analyst'. Description: 'Given a competitor URL, produce a 1-page teardown.'
Step 3: Write the system prompt
Include the output template verbatim (Sections: Positioning, Pricing, Product changes, Gap vs us, Recommended response).
Step 4: Enable Browsing + File Search
Turn off Code Interpreter and Image Generation — they slow tool selection.
Step 5: Test with 3 competitors
Refine the template until sections stay under 4 bullets each.
Step 6: Share with your team
Publish to your workspace. Add a link in your PMM Notion homepage.
Example
Prompt: 'Teardown of https://example.com'
Output: 'Positioning: horizontal PM tool for engineering leaders. Pricing: $19 seat. Product changes this month: AI issue clustering (Jun 4). Gap vs us: they lack cross-repo view. Response: highlight cross-repo demo in Q3 launch.'