Tracking & share-of-voice7 min read
Run the test across engines
Manual testing across ChatGPT, Perplexity, Gemini, and Claude — what to log and how to keep it under 30 minutes.
What to log per prompt
Five fields: engine, mentioned (yes/no), position in the answer (first, mid, last), source URL cited (if any), and competitors named. That's it — five columns is the sweet spot between useful and unsustainable.
Keep it manual (for now)
Tools exist, but for the first month do it by hand. You'll develop a feel for how each engine phrases answers and what kinds of pages they cite. That intuition is worth more than any dashboard.
Screenshot the painful ones
Whenever a competitor is named and you aren't, screenshot it. These screenshots are your prioritization tool — they show you exactly which gaps cost you the most visibility.
Hands-on
Try this before the next lesson
0 / 3 done