Index Exchange Halts Bid Caching, Used Auction Game For 50% Of Impressions

Bid caching at scale

Jounce Media’s research offers insight into the scale of Index’s bid caching practice.

The programmatic consultancy began looking into the practice in Q1 after inquires from a marketer. In early August, it independently funded and conducted a test where it purchased 1.7 million impressions across comScore 1,000 websites and nine exchanges.

Between 44% and 50% of all impressions filled by Index Exchange showed signs of bid caching.

But if bid caching happens on Index’s servers, and not out in the open, how could Jounce Media find out it had happened? The researchers looked for impressions that both violated a once-in-a-lifetime frequency cap it put in place and raised two of three red flags.

The first red flag looked for time lags of more than 10 seconds after an impression was served.

The second two red flags looked for URL mismatches that indicated an ad was served on a different piece of content than appeared in the bid request.

If the bid request contained just the homepage URL of a publisher, or if the previous ad served was on a different article page than the one where Jounce Media bought its second impression, those discrepancies indicated a bid was being held from earlier in the user’s session.

Jounce saw all of Index’s bid caching cease between Aug. 14 and Aug. 16, when DSPs and media buyers were made aware of the practice and Index Exchange paused bid caching.

The use of bid caching is limited to Index Exchange, according to Jounce Media’s research.

Inventory purchased on AppNexus, BidSwitch, Google Ad Manager, Oath, OpenX, PubMatic, Rubicon Project and Sovrn did not raise red flags on more than 5% of their inventory, which Jounce Media considered a false positive.

Index Exchange dropped within this same range after it turned off bid caching, said Jounce Media founder Chris Kane.

“Index was alone in engaging in a bad practice, and they responded to pushback to become a cleaner marketplace,” Kane said. “Our industry is healthier today than it was a week ago, and it gets healthier when you shine a light.”

User experience?

If 50% of all impressions Index Exchange served used bid caching, it casts further doubt on its argument that the practice improved user experience. And it raises further concerns about the harm this practice caused to buyers.

The Jounce Media study included no video or mobile app impressions, where the exchange argued that its practice had justifiable uses.

When Index Exchange first admitted to using bid caching in a Wednesday blog post, it claimed it did so “dynamically” and did not articulate the rules around what types of content justified the practice.

In an interview last week, Index Exchange product SVP Drew Bradstock told AdExchanger he couldn’t offer an example of a scenario where the exchange did not use bid caching. It’s not clear why Index Exchange ran bid caching on the percentage of inventory that it did.

Casale said in his letter to partners that he regretted the company’s lack of transparency – but maintained that buyers would opt into the practice once they understood it better.

Index Exchange did not respond to requests for comment about its letter to partners or the Jounce Media research.

This post was syndicated from Ad Exchanger.