Best practices: Bidding KPIs
Bidding KPIs give you more transparency into the auction, and are available for the networks that are bidders.
These are our recommendations on how to best evaluate your Bidding KPIs:
- Each bidding ad source receives one Bid Request per impression opportunity.
- Bid Requests are expected to be nearly equal for each bidding source when breaking by App x Ad Unit x Geo.
- A difference in Bid Requests between each ad source when the App x Ad Unit x Geo breakdown could indicate a tech or setup issue.
Bid Responses and Bid Rate:
- During the auction, each ad source can choose whether or not to bid. A Bid Response is counted when an ad source chooses to respond with a valid bid.
- Bid Rate is calculated as the number of Bid Responses / number of Bid Requests
- Bid Responses and Bid Rate can differ greatly depending on the ad source, due to differences in the ad source’s bidding strategy, tech, and demand. For example, a network may choose not to bid in a certain country or on certain devices.
- A low Bid Rate does not necessarily indicate low performance.
- Bid Rate can be evaluated similarly to “Availability Rate” of traditional instances, as it explains the percentage of time that an ad source has demand available for a specific impression opportunity.
- Rather than comparing Bid Responses and Bid Rates between ad sources, it is better to evaluate this KPI for each ad source over time, or to compare the ad source’s Bid Rate between different apps. Talk to the ad source if you see significant fluctuations in Bid Rate.
- A Bid Rate of 0% means there may be a setup issue.
- Win Rate is calculated as the number of times the Bidding ad source wins the auction / the number of Bid Responses.
- The auction is ended once the first ad source reports an ad fill.
- Win Rate is best used by evaluating an ad source’s Win Rate over time. If there is a decline in Win Rate over time
- Remember that Win Rate can be skewed by an ad source’s Bid Rate. An ad source that chooses not to Bid on all Bid Requests may have a higher Win Rate than a network that chooses to Bid on all Bid Requests. Therefore it is recommended not to compare Win Rates between networks.
- Render Rate is calculated as the number of times the Bidding ad source shows an impression / the number of Bid Responses.
- Render Rate differs from Win Rate because it shows the auction behavior only for auctions which led to an actual impression
- Render Rate can be compared with Win Rate. Large discrepancies between Render Rate and Win Rate may suggest:
- Network is bidding on less engaged impression opportunities (more relevant for Rewarded Video ads)
- Poor Traffic driver (not engaging or poorly placed)
- Publisher didn’t call to show the ad
- Try to isolate if the differences are stemming from a specific placement