Real-time competitive data for repricing engines
Repricing quality is bounded by how fast you learn a competitor moved and how sure you are the move is real. Polling gives you the first badly and the second not at all.
Repricing
Push beats poll
Subscribe to price.changed on your competitive set. You stop paying to re-read a catalogue that has not moved, and you learn about the moves that matter in minutes.
Repricing
Refuse to act on weak data
Every value carries a confidence weight. A repricer can require direct_api confidence before it moves a price and treat anything lower as advisory — which is the difference between reacting and overreacting.
Endpoints that do this
Where it lands in the API.
Events
Six event types, signed on delivery and replayable by cursor. Every consumer eventually needs to backfill a gap.
Live stream
Server-sent events for anything that cannot wait for a webhook round trip.
Compare
One product across every seller that carries it, ranked by landed cost, with the spread. Server-side work is worth more than four raw reads the caller has to join.
Questions
Common questions.
What happens if our endpoint goes down?
Delivery retries on a 1 min / 5 min / 30 min / 2 h ladder, then dead-letters. The event log is replayable by cursor, so you backfill the gap yourself without contacting anyone.