Lingua programmatica: The IAB is trying to standardize ad tech’s messy terminology

Smith co-chaired an Interactive Advertising Bureau Tech Lab task force that’s pushing an OpenData spec to standardize the language that vendors, advertisers and publishers use in their campaign reports. The idea is that having everyone use the same terms and definitions will make it easier for advertisers, publishers and vendors to transfer and use large data sets. The IAB opened the spec last week for public comment.

Vendors use their own lingo to describe their data to stand out among the clustered Lumascape. This creates confusion, but the intention of tech firms’ marketing isn’t to create awkward acronyms — it’s to appear unique to prospective customers, said Jason Krebs, chief business officer of GIF engine Tenor. This is why vendors pushed their own nomenclature for years around the concept that we now call header bidding.

“Vendors started to compete on the basis of ‘my metrics are better than yours,’ and with each new tech update, they had new things they could measure or call by a different name,” said Altimeter analyst Omar Akhtar.

As the number of middlemen in the supply chain ballooned, ad buyers ended up getting reports that all use different terms to describe the same thing, leading advertisers to spend money on analysts to manually restructure data sets, said Bidtellect CTO Mike Conway, who sits on the IAB Tech Lab’s OpenData task force.

Unlike other tech sectors like finance, ad tech has little oversight. Groups like the IAB can create a Rosetta Stone for data labeling, but there is no fine for failure to use these standards since the industry is self-regulated. The IAB’s specs will gain adoption if ad buyers withhold their dollars from vendors that refuse to abide by them, Smith said.

The IAB’s spec is open to public comment through March, said Dennis Buchheim, gm of the IAB Tech Lab. The task force is only creating standards for data mentioned in campaign reports. Later in the year, it may do the same for audience segments and data used for billing purposes, he said.

The biggest barrier to getting people to pay more attention to how they label data is that this work is wonky and unsexy, Smith said. “People aren’t going to say, ‘Wow, you normalized field headers? Let’s give you an award for that one.’”