Online Ad Spending Estimates Are Bogus, Some Say

Both Nielsen Online and TNS Media Intelligence unveiled display ad spending estimates recently. While there is significant disparity between their numbers, some argue both firms’ methods for gauging online ad spending result in faulty data.

Advertisers and their media buying and planning agencies rely on ad revenue information from firms such as Nielsen and TNS to determine how much they should spend on various media to run successful campaigns. While spending reports for other media such as television and print are considered reasonably reliable, advertisers and agencies have expressed concern about the validity of online ad spending data for around a decade or more.

Earlier this month, Nielsen Online reported that online display ad spending dropped 1 percent in the first half of this year compared to the first half of 2008. TNS, on the other hand, said this week that online display spending rose 6.5 percent in the first six months of ’09 compared to the same period last year.

TNS combines CPM-based rate card data with audience impression data from its Compete service to calculate its display spending estimates. Nielsen considers a variety of factors along with CPM rates. According to a company spokesperson, “The estimated dollar value for a given campaign, site, industry or advertiser is calculated by using statistical models analyzing estimated CPM values against a site’s genre, ad dimension, technology and delivery format.” Neither TNS nor Nielsen provided sources outside their PR teams for this article.

The use of rate card data is just one reason critics say online ad spending estimates are unreliable. While many Web sites publish rate card prices, they often negotiate with ad buyers and in the end sell ads at lower costs. Also, display ads are often served through ad networks, whose display prices have been dropping. “From my best understanding, published rate card is typically not going to represent the average CPM,” said Forrester Research Senior Analyst Emily Riley. “If you use the rate card, you’re going to get the higher-end estimate.”

Nielsen contends its method corrects for that. “Our predictive modeling approach is designed to mimic factors taken into account when media buyers are negotiating ‘true’ CPM rates by weighting against four key variables into our estimated CPM equation,” the firm’s spokesperson told ClickZ News.

According to TNS, the firm “assigns a CPM value to every display ad for purposes of estimating expenditures.” However, the company “does not distinguish between display ads that are paid for on CPM vs. [those that are performance-based].”

Still, the fact is measuring CPM rates, whether they’re based on rate card prices or industry averages, could result in a rose-colored view of reality. In recent years, media buyers have increasingly favored performance-based display buys — often placed through networks — over CPM-based buys.

Riley believes the reliance on CPM prices to measure display ad spending “absolutely” distorts ad spending estimates. Forrester Research bases its spending estimates in part on public revenue data from publishers, and isn’t considered a direct competitor to TNS or Nielsen Online.

Mediasmith CEO David L. Smith agrees that using rate card data “is a concern,” but believes the bigger problem with display spending estimates lies in the use of impression data. Both TNS and Nielsen estimate the number of times particular ads were served on a particular Web site. The research firms calculate impressions based on their site audience traffic measurements, which are also estimated based on online audience samples. Not only are publishers highly skeptical of the audience estimates, advertisers and agencies involved in purchasing display advertising often dispute the number of ad impressions the research firms report.

“They’re using a sampling system of bots and they can’t do enough sampling to pick up a representative measure of the actual spending,” said Smith. In addition, he suggested increased use of ad targeting makes it even more difficult to provide an accurate measure of ad impressions. For instance, ads could be targeted across several Web sites to particular users based on behavioral data, making a site’s audience traffic a moot factor in calculating impressions of a certain ad or campaign.

“When a client tries to find out how is Wal-Mart’s spending against Target, there’s no way they can tell,” he said, adding that research firms like TNS and Nielsen “charge tens of thousands of dollars a year for this inaccurate service.” Smith believes, “The only possible solution is for the industry to voluntarily contribute the numbers,” meaning site publishers should provide information on true ad costs and ad revenue.

That’s a tough proposition. The Interactive Advertising Bureau already tried to create a service to track true ad costs, but to no avail. In December 2006, the publisher-centric organization opened its Advertising Expenditure Service, or AdEx, initiative to participants to provide ad pricing information. The goal was a database for advertisers to gauge ad costs for easier media planning. “AdEx will be the most reliable and credible source of online activity available,” stated the IAB at the time. But, because many small and large publishers grew leery of giving away proprietary information for fear of increased competition, the project never got off the ground.

UPDATE: This story was updated to include additional information about TNS’s model for estimating ad prices.

Related reading