When ChatGPT became an ad surface, we were already watching.

Measured share-of-voice on a brand-new ad surface.

For every audited merchant the team started recording: how often ChatGPT returned a sponsored card on a brand-relevant query, who was running it, and what creative they were running. Our answer reader was extended to capture ads; the SponsoredBrandPlacement and SponsoredCreativeCard tables started populating.

• Day 1platform-level monitoring online
• 4AI surfaces watched (ChatGPT + 3 others)
• Hourlycreative-change detection cadence

Simulated the displacement risk: when a competitor buys your branded query.

The team runs `detect_brand_term_conquesting`: the moment a competitor’s sponsored card surfaces against the merchant’s own brand name, the team opens an alert and queues a creative-side response.

Shipped the alert pipeline: sponsor changes raise an Inbox event.

Every change to who’s running, what they’re running, and how often they’re running it on a watched query becomes a sponsor_change alert in the merchant Inbox. The merchant team sees the displacement before their CMO does, and gets a draft response queued.

• `get_sponsored_placements`: full historical record per query
• `compute_share_of_sponsored_voice`: daily SOSV trend
• `detect_ad_cannibalization`: flags self-cannibalisation across ad surfaces

Compounded: every new AI ad surface gets the same treatment, on day 1.

The pattern that landed for ChatGPT (ad capture, schema for placements, cards table, alert pipeline) is the same pattern the team will run when Perplexity, Gemini, or Copilot ship paid placements. New surface, same monitor, same day.