Debunking the tech tax and other programmatic myths
If there’s one thing that many marketers complain about, it’s that programmatic is a black box that salami slices their budgets. It steals away spend from its core task; reaching the consumer.
Most recently, at the ANA’s 2018 Masters of Marketing conference, the organisation’s chief executive Bob Liodice claimed that just 25% of CMOs’ digital media investment reaches target audiences, wasting (if this claim is true) more than $20bn of marketing spend.
Such arguments are also fuelled by 2014 research from the World Federation of Advertisers which says that 60% of spend disappears before it reaches the publisher.
It’s no surprise that advertisers ask “where does $100 of my programmatic spend go?” and it’s worth updating the programmatic waterfall with numbers that actually correspond to today’s reality.
The reality is that nearly $80 of the budget actually reaches the sell side, while $20 of the budget pays for the execution on the buy-side. The average breakdown for an ideal programmatic budget allocation of $100 looks like this:
The Advertiser: The buy-side is very much controlled by the Advertiser, as they chose to work with an agency (or not) and usually are involved in selecting the DSP and ad verification vendor.
The Agency fee: In a fully transparent programmatic service model, the average market fee is about 15%. Higher on small budgets and complex campaigns, lower on large budgets.
An agency fee of 15% does not mean, however, that the agency takes $15 from the advertiser’s budget. The fee is usually based on the media and data procured via the Demand Side Platform (DSP), that’s the $80 which reaches the sell-side. A 15% agency fee on media and data is actually $12 of the advertiser’s budget.
This may sound high compared to offline agency margins but running programmatic advertising is extremely complex and the agency resource would have to be replicated if an advertiser took this in-house.
The DSP licence costs: The same calculation applies to the DSP licence cost, which again is based on media and data throughput. A typical 10% licence cost represents $8, lower than the cut taken by ad networks in the past.
Ad Verification: No responsible marketer would disagree that controlling ad fraud, brand safety and viewability is not a key priority nowadays. In programmatic we can also use pre-bid filters to increase viewability and prevent buying fraudulent or non-brand-safe ads. Those tools are usually charged on a CPM basis but generally represent no more than 2% of the budget.
Where advertises have less control is on the sell-side. Here media vendors chose both the route to market and their partners.
The Sell-Side Platform (SSP): Publishers use a Sell-Side Platform to better monetise their media inventory. SSPs operate on a revenue share and competition between SSPs in header bidding means that their take-rates are under pressure. Right now, 10% seems to be an industry average.
10% on the publisher’s revenue of $60 means $6 for the SSP. It’s worth noting that a publisher would only allow an SSP to take this cut if they were creating greater value.
Data: Some advertisers are worried that they spend too much money on third-party data for targeting. The average spend on data is around $12. That sounds high unless you understand the contribution that data makes.
Data is only bought if the uplift it provides is higher than the cost. Targeting data helps to deliver the ‘right person’ proposition with the media’s ‘right context’ promise. No one seriously believes that sending the same message to everyone at the same time is effective on digital platforms.
The misperception of “working media”
Often, the discussion about where money ends up in programmatic is built around the concept of “working media” – this describes the share of the advertiser’s budget used to buy reach vs. the share spent on stuff like technology, data and agency fees.
This approach fails to understand where value creation occurs. One of the great advantages of programmatic is that we can optimise the targeting data and the media inventory independently. Both remain a ‘working’ investment.
When an advertiser places a direct buy for ‘Auto Intenders’ on Yahoo, for instance, they count every dollar of that media buy as working media. When they buy the same campaign programmatically and now use Auto Trader’s in-market data on Yahoo’s media inventory (Auto Trader has stronger intent-signals), it seems ridiculous to count only the media portion as ‘working’.
Many publishers also benefit from selling data. Sometimes they use data marketplaces to sell their data separately from media but they still get the money (and can even earn additional revenue outside of their own media).
I would be foolish not to acknowledge that my waterfall is based on the ideal. It describes a setup where there are no unknown intermediaries between advertiser and publisher. That is not always the case – particularly when resellers squeeze into programmatic marketplaces, although the value chain is becoming much more transparent with ads.txt and the coming ads.cert.
But, please, when we talk about the programmatic costs can we at least use more realistic numbers and not those that misunderstand the process and the value that each element creates?
The bottom line is that advertisers who opt for a fully disclosed programmatic model have full control over their costs. And they will deploy each element of the waterfall only if it brings with it added value.
Oliver Gertz runs MediaCom’s programmatic efforts across EMEA