5 steps. ~10 minutes to first signal. Hourly tracking starts immediately. Prompts are weighted by your ICP — winning on queries your actual buyer types counts more than winning on generic ones.
1 · Domain
2 · Buyer ICP
3 · Competitors
4 · Prompts
5 · First run
No https://, no path. Just the apex domain.
The category buyers ask about.
Why this step matters: Generic prompt sets miss queries your actual buyers type. We use this profile to generate prompts that match their language, weighted toward your funnel stage. Per Mert D. (Populon AI Labs) — public LinkedIn comment, May 2026.
The persona who'll actually search ChatGPT.
Country + tier-1/tier-2 city if local intent matters.
These show up in the generated prompts. The more specific, the better.
Effect: prompts will be weighted by their relevance to this ICP. A prompt your buyer would never type gets weight ~0.3; a prompt they explicitly would type gets weight ~1.5. Blended SOV uses these weights.
We auto-suggest from your domain + keyword. Add or remove. Up to 6.
Tagged by intent + ICP relevance. Bar = weight. These run hourly across ChatGPT, Perplexity, and our local Qwen.
0 selected
Confirm and run the first scan now (3 prompts × 2 engines = ~30 seconds). The hourly cron then takes over.
Asking ChatGPT and Perplexity… ~30 seconds.
✓ Tracking is live with ICP-weighted prompts. The hourly cron will fire at the next :07.