1 · Otterly.ai (free tier)

What you get free: Track ~5 queries across ChatGPT, Perplexity, and Gemini. Weekly automated checks. Email alerts on citation changes.

What's paid: More queries (up to ~50 on the $99 plan), Claude tracking, competitor benchmarking, API access.

Use it if: You want a "set and forget" check on 5 highest-value queries without thinking about it.

2 · PromptWatch (free tier)

What you get free: Track ~10 queries across all four major engines (including Claude). Daily checks.

What's paid: More queries, prompt-level analytics, team workspace.

Use it if: You want broader engine coverage on free — particularly if Claude citations matter to your audience.

3 · The manual check (free forever)

The path that costs nothing: open ChatGPT Search, Perplexity, Claude, and Gemini in four browser tabs. Run your top 5 commercial queries through each. Note which engines cite you and where.

Why it still works in 2026: AI engines are designed for human use. Manual checks return the same answers a paid tool would query through the API. The only thing you lose is automation.

Use it if: You're starting out, have fewer than 10 critical queries, or want zero vendor lock-in.

Operator tip. Make this a Sunday evening routine. 10 minutes a week of manual checking will tell you more about your GEO standing than most paid dashboards. Once your query list grows past ~25, automate via Otterly free tier first, then PromptWatch, then build your own agent.

4 · Bing Webmaster Tools (the hidden free win)

ChatGPT Search uses Bing as its underlying index. If you're not in Bing, you're invisible to ChatGPT Search — full stop. Setting up Bing Webmaster Tools (free, takes ~10 minutes) is the single highest-ROI GEO move most operators skip.

What to do: Verify your domain → submit sitemap → check index coverage. That's it.

5 · Building your own agent (the scale path)

If you track 100+ queries weekly OR want citation alerts piped into Slack / your existing dashboard, the economics flip toward building your own. Claude API costs roughly $0.003 per query at the queries-per-day volume most operators need; the agent itself is ~200 lines of Node.js.

What it costs in practice: ~$5–20/month in API spend for serious tracking volume. ~1 day to build if you've used the Claude API before; ~3 days if not. I include this agent in the $9,500 audit as part of the deliverable.