“Data-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Nico Neumann, assistant professor and fellow, Centre for Business Analytics at Melbourne Business School.
There were many explosive findings in the recently released ISBA report about the United Kingdom’s ad tech supply chain. For example, 15% of spending was untraceable. Equally concerning was that only 12% of the 267 million impressions analyzed could be matched and properly investigated. We can only imagine what the cost structure for the remaining 88% must be.
Furthermore, the ISBA study exposes again the high proportion of funds taken by middlemen: Publishers get about half of the money advertisers spend on digital media.
The question that many posed afterward was: Why is this such a problem?
Questionable comparisons of cost structures
No surprise, many industry commentators have quickly downplayed the issue that only half of the media spend ends up with publishers. Programmatic costs were compared with banana delivery logistics and even home loan interest that buyers pay off over many years.
But are these really accurate comparisons?
We cannot consume the banana unless it is delivered physically to us. There are logistical costs we cannot avoid. However, an impression for a media placement can be transferred digitally, at basically no costs. And the key purpose of a media placement – providing an opportunity to be seen – is also independent of programmatic services.
Likewise, loan interest costs only occur when one needs to borrow money and are independent of everything else. If I borrow $1 million to buy media inventory (no matter if it involves programmatic or not) and then pay this sum back over 30 years, I will add an incredible amount of expenses to my media placements too.
Ad tech connects buyers and sellers of media
Let’s keep in mind that the main goal of ad technology is to connect buyers and sellers. Hence, a better comparison for cost-structure discussions would be to look at the commissions of other matchmakers, such as travel agents or ride-sharing services.
Now let me ask you, how would you feel if you booked a vacation via Expedia or Kayak.com and only 50% of the money was sent to the hotel you booked?
Few people would probably use such a marketplace. This is why the commission of travel agents is typically around 15%.
Ride-sharing services, such as Uber or Lyft, make up a particularly interesting case study for cost-structure debates: While official ride-sharing commissions range between 5-25%, many providers seem to have (semi) hidden fees too. Therefore, the effective commission of ride-sharing providers turns out to be between 30-40%. If drivers bring home about 65% of the fare people pay, they would have to spend another 20% of that income for costs they must bear for things such as gas, tolls or maintenance (which would be the equivalent of ad-serving costs). The result is the driver ends up with about 52% of the cost of the ride. This number is very similar to the publisher income estimates (51%) from the ISBA study.
Back to our question: Is this a problem?
While those pocketing the high margins tend to defend the status quo, many ride-share drivers have protested about unfair pay and the lack of transparency. One of the riders’ goals has been to persuade legislators to investigate companies’ commissions and get clarity on costs. As a result, the first legislation has been passed in California.
The ad-tech flaws costing publishers and advertisers unnecessary dollars
The story is not vastly different for ad tech. Legislators in the United Kingdom and Australia have also begun looking into digital advertising markets. What could they discover? Is something really wrong with ad tech?
Let’s think this through. Historically, advertising agencies used to demand a 15% commission for “manual” media buys, contextual targeting included. This should be our benchmark. Because advertisers often use some form of new digital targeting (note: Audience targeting does often not pay off), we add 5-10% to our calculation (in line with the 10% demand-side tech fees for data, verification and ad serving identified in the ISBA study). These buy-side costs add up to 20-25% of ad spend, leaving about 75-80% to publishers and content owners. There is a large gap between that 75-80% and the roughly 50% found by the ISBA study. Why does ad tech seem to suck up another 25-30% of ad spend? Technology should make buying and selling easier and help reduce the price of media buying by automating manual processes. It should not make everything more expensive.
So what are ad tech inefficiencies and flaws that cost buyers too much money?
Many issues come to mind: ad fraud, questionable business practices (e.g., bid shading) and possible kickbacks hidden in the “unexplainable delta” of the ISBA study findings.
However, the likely greatest concern seems to be the long and murky supply chain of programmatic transactions – the ISBA study found that the 15 participating advertisers had nearly 300 distinct supply paths to reach only 12 different publishers.
In addition to disguising some of the problematic practices more easily, two further risks emerge when too many middlemen are involved. First, there seem to be too many parties (aka resellers) passing on the same impression without adding value while charging for their “services.” Second, having several resellers bidding on the same impression artificially inflates prices in auctions.
The remedy: transparency, independent analytics and more direct buys
Similar to the ride-sharing case, legislators may enforce some positive change in the future. But this could take time.
What can savvy advertisers do until then?
First, they should enforce transparency through their contracts and demand log-level access rights to their ad data. Second, advertisers need to take control of measurement and analytics. Media optimization should be done in house or by an independent party to avoid having media buyers grade their own homework. And the analytics should be based on scientific methods – such as experiments – and not attribution rules and pseudoscience that inflate ad effectiveness of the digital media placements.
Third, supply-path optimization and buying directly from publishers or through fewer exchanges should be attempted wherever possible. With shorter supply chains, it’s also easier to use first-party data and protect data privacy.
This approach should provide a clear win-win(-win) – at least for consumers, advertisers and publishers, which are the most essential parts of the ecosystem.
