Ads.txt Errors Plague 16% of Publishers

Bad formatting and misspellings cost publishers money

The widespread adoption of ads.txt can be a double-edged sword for publishers.

On one hand, ads.txt—an Interactive Advertising Bureau (IAB)-backed text file that lets publishers publicly list all the vendors authorized to sell their inventory—helps publishers take control of their inventory by giving ad buyers a check against domain spoofing and arbitrage. On the other, making simple errors in the text file can lead publishers to miss out on some cash.

Among the top 5,000 websites worldwide selling programmatic ads, adoption of ads.txt increased from 8.5% in September 2017 to 51% by the end of February 2018, according to Pixalate. As publishers adopted ads.txt, many made basic mistakes in their text files. These errors can be problematic, since ad buyers and their demand-side platforms (DSPs) are using ads.txt to filer unauthorized inventory across their programmatic campaigns.

FirstImpression estimates that among the top 30,000 publishers according to Alexa ranks, 16% of those using ads.txt have errors in their text files. Similarly, ads.txt validator AdAuth, which scanned 300,000 publishers, found that over 10% of websites using ads.txt have mistakes in their files.

Syntax errors represent some of the most common mistakes in ads.txt files. The ads.txt spec gives guidelines on how the required information—which includes the exchange’s domain, as well as the publisher’s account identification number and relationship to the exchange—should be ordered and laid out. Syntax errors can occur when publishers forget to include a space between each of these variables, leave out commas to separate the variables or place the information in the wrong the order.

“The data is unparseable because it fails to adhere to the basic syntax of the spec,” said Ian Trider, director of real-time bidding (RTB) operations at Centro, a DSP. “So DSPs will ignore such records, which would result in not bidding on the corresponding bid requests.”

Publishers also misspell the names of the vendors they work with and list invalid domains in their text files. Another common type of error is missing records, which happens when the required pieces of information that DSPs rely on when buying inventory are missing. All of these errors can lead a DSP to not bid on a publisher’s impressions that come through the particular vendor where the errors in the text file lie.

The good news for those with errors on their files is that there are free open-source ads.txt normalization maps that list the common alternative names publishers have for vendors in their files. DSPs can use these maps to fix ads.txt data as it comes in, so that if a publisher used a commonly observed incorrect label they wouldn't be affected negatively, according to Trider.

Another thing worth pointing out is that an ads.txt error doesn’t invalidate an entire file. AdAuth and FirstImpression simply note how many files contain errors—their research does not yet show how many errors each file has. Many publishers have dozens of line items in their ads.txt files. If only one of these items has an error, it might not have a noticeable impact to the publisher.

“I see this as more of a minor hitch, rather than a critical problem,” Trider said.

"Behind the Numbers" Podcast