Win Rate

Win Rate answers a simple question: of all the auctions you participated in, how many did you win? It is the most basic pulse check for bid competitiveness in programmatic advertising.

Platforms define the denominator differently. Some count every bid request the DSP received; others count only bids actually sent to the exchange. The difference matters — a DSP that throttles bids aggressively may report a high win rate on sent bids while missing many eligible auctions.

Why it matters

• A win rate that is too low means your bids are losing consistently. You are spending on bid processing without winning impressions.
• A win rate that is too high (above 80-90% on open exchange) may mean you are overpaying or missing valuable low-price inventory.
• The sweet spot depends on your vertical, audience, and auction type (first-price vs. second-price).

So what? Win rate is a directional signal, not a standalone KPI. Always pair it with loss reason data from OpenRTB loss notices to understand why you lost.

Win Rate = Auctions Won / Auctions Participated

Most platforms express this as a percentage. The denominator is the key variable.

Denominator caveats

1. Bids sent — only auctions where the DSP actually submitted a bid. This is the most common definition in DSP analytics.
2. Bid requests received — includes auctions the DSP chose not to bid on. This definition yields a lower rate and reflects targeting efficiency.
3. Bid responses with a valid price — some platforms exclude bids that were filtered by the exchange before the auction.

Example

• A DSP receives 10,000 bid requests.
• It sends bids on 4,000 of them.
• It wins 200 auctions.
• Win rate on bids sent: 200 / 4,000 = 5%
• Win rate on bid requests: 200 / 10,000 = 2%

Check your vendor's definition. The same campaign can show 2% or 5% depending on the denominator.

A single win rate number tells you almost nothing. You need context.

What to pair it with

• Bid volume — high bid volume + low win rate = uncompetitive bids or floors set too high.
• Average CPM — if win rate drops as CPM rises, you may be hitting a price ceiling.
• Loss reason breakdown — OpenRTB loss notices tell you if you lost on price, floor, or brand safety filters.

Common dashboard patterns

• Win rate < 2% on bids sent — your bids are rarely competitive. Review bid price strategy and audience targeting.
• Win rate > 50% on open exchange — you may be bidding too high. Check if you are winning only expensive inventory.
• Win rate stable but CPM climbing — the market is getting more expensive. Your bids are keeping pace, but efficiency is dropping.

So what? Use win rate as a diagnostic, not a target. A "good" win rate for a retargeting campaign (often 10-20%) looks very different from a prospecting campaign (often 1-5%).

Bid price

Raising your bid CPM directly increases win rate — but at the cost of efficiency. The relationship is not linear: doubling bid price rarely doubles win rate.

Floor price

If a publisher raises the floor price, your existing bids may suddenly lose. Check if win rate drops correlate with specific publishers or deals.

Audience targeting

Narrow audiences reduce the pool of eligible auctions. You may win a higher percentage of the few auctions you enter, but total wins may drop.

Creative and format

Rich media, video, or interactive creatives sometimes have higher win rates because fewer bidders support them. Standard display banners face more competition.

Tradeoffs

• Chasing win rate in isolation can lead to overbidding. A 50% win rate with a 10% margin is worse than a 5% win rate with a 50% margin.
• Improving win rate by lowering floors (as a publisher) may increase fill but reduce revenue per impression.
• Improving win rate by narrowing targeting may reduce scale. Balance win rate with reach goals.

Win Rate is a vital diagnostic for auction competitiveness, but it is meaningless without context — bid volume, CPM, and loss reasons.

• Always verify your platform's denominator definition.
• Pair win rate with loss reason data to diagnose why you are losing.
• Do not optimize for win rate in isolation; it is a means to an end, not a KPI.