exchanges replying with "estimates" or "averages"

[djlosch]

As described in the docs for prebid, and on appnexus, some exchanges don't actually respond with their true bid value. Instead, they reply with estimates, ranges, or averages.

This defeats the entire purpose of header bidding. Mathematically, there's little difference in this from the days of waterfalls and line item averages... but now it's implemented manually instead of letting DFP do the work.

_If an exchange does not respond with their true bid value for that impression, they should not be supported_. Alternatively, is there a chart that shows who responds with true bid values, and who doesn't?

[prebid]

+1. You're right - header bidding without price can be done through a waterfall in DFP. The latter is actually better without having to make the page wait. We're trying to keep track of this as well, as it indeed affects our implementation for different bidders.

For a bidder that responds with exact price and creative, the implementation can be standard. For a bidder that obfuscates the price or not returning one, the line items have to be setup differently. More can be found here:

http://prebid.org/bidders.html

[djlosch]

I get that a standard can exist that says bidders must reply with actual bids, but it's already gotten out that some bidders aren't following that standard. Is there yet a chart or list of who is / isn't replying with true bids? There are a bunch of bidders with adapters that are already coded... do they all reply with true bids? A public list would be awesome because...

• it'd make it a lot easier for everyone to know not to waste time implementing exchanges that don't respond with actual bids, and
• it'd hopefully pressure the non-compliant bidders to get with the program.

[prebid]

Great suggestion and we hope this page would somewhat help:
http://prebid.org/blog/header-bidding-bidder-analysis/

It's possible that we don't have the most up to date info from bidders. Some may be developing new APIs that allow real price back. This above analysis is what we've seen publishers are using today.

[prebid]

Closing this for now. Hopefully the analysis page helps.