Reading a bid request like a trader: the five signals that move price
Put two bid requests for the same audio placement side by side and one can clear at double the other's price. Buy-side algorithms aren't mysterious about why: they price what they can see. After processing billions of requests a day, we can tell you that most of the price variance concentrates in five places.
1. Content context — app.content
A request that says "an audio app" earns a floor bid. A request that says news_talk, English, live stream earns a considered one. Genre, language and livestream flags let a buyer match campaigns to moments — finance advertisers pay up for business talk; entertainment brands want music and comedy. Empty content objects are the single most common self-inflicted revenue wound we see from publishers.
2. Identifiers and their consent state
Not the identifier itself so much as its quality: is there a stable (pseudonymous) ID, is consent properly signaled under TCF, and does the ID connect to any audience data? A request with a consented ID and two or three segments attached is worth multiples of an anonymous one, because the buyer can frequency-cap, retarget and attribute. Where there is no ID, rich context (signal #1) is the fallback that keeps price alive.
3. Geo precision
Country-level geo buys you country-level budgets. Metro and postal-level geo unlocks local advertisers, dealership campaigns, retail footfall goals — demand that simply cannot bid without it. Deriving clean region/metro from IP (with consent where required) is one of the cheapest enrichments with the highest price impact.
4. Placement honesty — position, duration, skippability
Buyers learn fast. Inventory that declares midroll, 30s max, non-skippable, stitched and delivers exactly that builds algorithmic trust — DSP bidding models literally learn to pay more for your domains and bundles. Vague or inflated declarations get discovered in post-campaign analysis and priced down globally. The market's memory is long and it is stored in machine learning models.
5. The supply path itself
Before a serious buyer prices your impression, their supply-path optimization has already scored the route it arrived by: is the seller in sellers.json, does ads.txt/app-ads.txt authorize the path, is the SupplyChain object complete? A clean one-hop path with full declarations doesn't just avoid being filtered — it earns a structural bid premium because every dollar demonstrably reaches the publisher.
Floors don't create price. Signals create price; floors just decline it. Publishers who obsess over floor tuning before fixing their signals are optimizing the wrong end of the auction.
The takeaway
Whether you're on the buy or sell side, audit five things on your traffic: content objects, ID/consent quality, geo precision, placement declarations, supply path. That's the whole game — and it's exactly what our Enrichment Engine automates on every request that crosses the exchange.