Following the money

At least a quarter of programmatic ad dollars spent on advertising across the open web are being wasted, circa $22bn globally. Almost half of that figure is being pushed to ‘made for advertising’ (or ‘made for arbitrage’) sites according to a major study on behalf of some of the world’s biggest advertisers.

The ANA study – which mapped ads from the likes of Dell, HP, Kimberly-Clark, Nissan, Mondelez and Shell – managed to get some of the major SSPs and DSPs to lift the hood. Others, including The Trade Desk, Amazon, and Yahoo, Google AdX and PubMatic either declined or ‘could not provide data in the required timeframe’.

“Everyone says they are supporters of transparency … until they’re the ones asked to be transparent,” noted the ANA’s report.

The ANA study mapped log-level analysis across $123 million in spending with 21 advertisers and 35 billion impressions. It found only 36 per cent, or 36 cents on the dollar, on average landed in front of audience eyeballs via the programmatic open web, i.e. digital programmatic dollars not spent via the big walled gardens.

Inclusion lists good, exclusion lists bad

The ANA said advertisers should probably ditch exclusions lists – keeping up with an AI-generated stream of sites designed specifically to look good to bidding algorithms is a “a herculean and futile task”. Instead, work on inclusion lists, i.e. pick the publishers you want to work with. Per the report, 63 per cent of impressions mapped came from the top 500 websites, and 86 per cent from the top 3,000.  Yet on average, the 21 major advertisers in the study were pushing out their ads to 44,000 websites and apps.

Locally Mutinex boss Henry Innis made the same point at the recent IAB Measure Up conference, telling advertisers they are wasting money by pouring display dollars blindly into programmatic buys rather than making "intentional" investments with premium publishers.

“There is a case to be made to run 95 per cent of one’s programmatic activity through inclusion lists and 5 per cent via open marketplace inventory,” per the ANA report.

It suggested advertisers and agencies start with curating “75–100 high-quality, trusted publishers or sellers (versus individual domains)”. That means advertisers get thousands of individual websites in the buy, while largely avoiding made for advertising/arbitrage scams. But it might make smaller publishers reliant on programmatic advertising revenues highly apprehensive.

Cull tech supply chain, log-level or bust

Working with a smaller number of ‘trusted sellers’ is another recommendation, i.e. contract directly with the sell-side platforms with the shortest route to publishers to limit the number of resellers, murk and mark-ups. That also has a positive side effect: lower carbon emissions.  The likes of AppNexus founder turned Scope 3 CEO Brian O’Kelley reckon publishers could quickly halve emissions and save money by doing precisely that. As he told Mi3 earlier this year: "We can demand shorter supply paths for programmatic. We can stop buying crappy inventory. We can stop buying on ‘made for advertising’ [sites]. If we all stop buying crappy outstream video, I'm pretty sure that would have a massive impact."

According to the ANA, advertisers should work with 5-7 SSPs and apply pressure to them to cut out bloat: “While the ANA is not in the position to recommend specific SSPs, the participating SSPs in this study should at least be considered: Index Exchange, Magnite, Microsoft Advertising, OpenX, and SpotX (part of Magnite),” per the report. Plus, fewer SSPs means advertisers are less likely to be bidding against themselves.

Contracting directly with ad tech partners such as DSPs, SSPs and ad verification firms also allows brands to specify access to log-level data within their contracts. Which means they can properly audit it. “Log-level data analysis is essential to drive additional media investment productivity,” per the study.

Brands: Get a grip

Like the ACCC nearly three years ago, the report also politely told brands to get a grip and take ownership. It recommended firms appoint a Chief Media Officer instead of just handing over everything to agencies (whom procurement departments have often appointed at least partially on price).

At the least, marketers should ask themselves the following:

• How many websites are used for an average campaign?
• Do you implement inclusion lists (or exclusion lists) for your programmatic advertising?
• Does your company own direct contracts with any supply chain intermediaries? 
• If not, should you?
• Has your company done any work to optimise its SSP activity? 
• How much of your programmatic media activity is running on Made for Advertising websites?

In the end, advertisers must be prepared to accept higher prices for non-crap. Per the study, 50 per cent of impressions came in under US$3 per CPM.

Private marketplaces: Sold a lemon?

Private marketplaces – at least the larger PMPs – can be mutton dressed as lamb, per the ANA, with those including more than 500 publishers housing (hiding?) circa 20 per cent made for advertising sites (versus just over a quarter for the open web), yet charging twice as much on a CPM basis. The firm used a sub-prime mortgage analogy, whereby mixing the good stuff with the crap can create an illusion of premium inventory within PMPs, especially the larger ones.

“Our analysis suggests that not all categories of PMPs with matched DSP and ad verification live up to the promise of being ‘premium’ or ‘privileged’,” per the report.

Overall, the report suggested brands – at least larger spending advertisers – could within just a few months end up paying significantly lower ‘true CPMs’ by doing the work in-house, or paying consultants, to map log level data, set up direct contracts, create inclusion lists and then recycle wasted spend to working spend. Per the report, only 41 per cent of companies involved in the study are definitely using log-level data to analyse programmatic activity.

See the report here.