Ask an ad ops team how their floors got set and you usually get a history lesson rather than an answer. Someone raised the floor on a mobile app segment during a strong quarter. A partner complained about a country, so that country got blocked. A spreadsheet went around, and three of its rows became rules. All of it is still running, and nobody has measured any of it since the day it was written.
That is normal at an exchange, and it is not a competence problem. A hand-written rule is cheap to create and expensive to judge, because judging it means knowing what would have happened without it. So rules pile up while the market underneath them moves.
A platform can respond to that in two ways. It can take the pricing decisions off you and promise to make them better. Or it can leave pricing with you and work on the part that wastes money most quietly: which requests each buyer receives. Floxis does the second.
Where the waste is
Most buyers cap the requests per second they accept, and most of what an exchange sends never gets a bid. When a cap binds, the requests that get through are effectively picked at random, so a request the buyer will never answer takes the place of one it would have bid on.
Even when no cap binds, every unanswered request costs both sides parsing and connection time, and buyers notice. Our guide to RTB traffic shaping covers the economics in detail.
That is a routing problem rather than a pricing one, and you can work on it without touching a single floor.
The Optimizer: one goal per demand endpoint
Every Floxis client gets the Optimizer on every demand endpoint. It has three settings:
- Off sends traffic exactly as you configured it, a plain baseline.
- Bid rate shapes traffic toward what the buyer bids on, for more bids per request sent.
- Revenue shapes traffic toward the requests that earn, for more revenue cleared.
The Optimizer changes where traffic goes and nothing else. Your floors, your margin and your rules stay exactly as you set them, so a commercial commitment you made to a publisher or a buyer is never rewritten by the platform. You pick the setting per endpoint, and you can change it when a buyer changes.
AI Booster, for the buyers you care most about
Floxis AI Booster is an optional layer on top of the Optimizer. It is in early access and offered on request. It adds per-buyer scoring, which ranks each request by how likely that buyer is to bid on it and sends more of what the buyer bids on and wins. It also adds segment blocking, which stops sending a buyer the segments where its demand economics are poor.
You test it before you rely on it. Switching AI Booster on creates a copy of the endpoint that receives a share of traffic, 20 percent by default, and you can change that share while it runs. Your original endpoint keeps the rest. Results come with a significance read, and nothing is promoted automatically. You decide whether to keep it, and switching it off returns all traffic to your original endpoint.
AI Booster does not set bid prices for the buyer, does not detect fraud and does not guarantee more revenue. Where an endpoint is nowhere near its cap, the likely gain is fewer wasted calls rather than more money.
What about floors?
Floors stay yours. You set them per placement, format and geography, deal floors are honoured as the contractual terms they are, and they apply as written.
We did build adaptive floor logic into our own exchange and measured it against a holdout on live traffic. On the partners we tested, the result came back marginal to negative, so we did not make it a Floxis feature. Our guide to bid floors explains why, including the trap that catches most floor experiments: raising a floor lifts your average CPM whether or not it made you money, because it removes the cheap clears from the average.
Check it at log level
Whatever you set, every bid, win and drop lands in a log-level ledger about a minute after the auction, with the reason each request was dropped. You can see what an endpoint received and what came back, instead of trusting a summary line.
Four questions for any vendor selling you optimization
Whoever you evaluate, including us, ask these:
- What exactly does it change: traffic, floors, margin or bid prices?
- Can you switch it off per endpoint, and where does the traffic go when you do?
- Is a control running alongside it, or is the result a before and after chart?
- Can you audit what it did, request by request?
A vendor who answers all four plainly is selling you something you can check.
If your floors are older than your last demand partner and your buyers are capping your QPS, that is the conversation worth having. Book a demo and we will walk you through the Optimizer and AI Booster.