Talking Programmatic: Is it Time to Retrain the Machines?

In the first of a regular series, programmatic expert Paul Gubbins asks if its time we retrained the machines to value context as well as audience.

Following the recent article on the front page of The Times about big brands funding terrorism via their advertising; fake news winning US elections; Methbot apparently sucking up between $3-5m per day from unsuspecting advertisers buying inventory from SSPs and exchanges; and P&G’s Marc Pritchard publicly suggesting the supply chain is murky at best, programmatic buying finds itself in the limelight again for all the wrong reasons.

One of the main benefits of data-driven buying via demand side platforms (DSPs) is that buyers can find audiences at scale. What does this mean? Well, it means a DSP can match cookies or device IDs to originating impressions regardless of the environment of their origin (so long as the site or app of origin are not on the advertiser’s blacklist).

So how is this any different from yesteryear when I could call my favourite newspaper and book ads to run in their travel channel via an insertion order (IO)? Well, the volume of impressions available to me as an advertiser was limited – and was sold to me on a fixed CPM basis so I could not cherry-pick each impression – and we know the audience of said travel channel does not just look at this content all day; they also look at other channels and sites and engage with many mobile apps.

Limited traffic
OK, but how can a DSP help me then if you are telling me there is only a limited amount of traffic I can target in my desired travel channel of the premium newspaper publisher? Well, by feeding your selected DSP a combination of first, second and third party data and a blacklist, you empower it to find travel users across the web and app ecosystem.

This migration from insertion order to programmatic buying has had both pros and cons for premium publishers. The pros are that it opens up their inventory to incremental demand that their direct sales team could never have secured via a feet-on-the-street, IO-driven approach. It also enables smart publishers to interrogate the incoming programmatic data to inform their rate card and act as a leads list that their direct sales teams can mine to inform their larger strategic conversations around formats than cannot be pushed through OpenRTB pipes, this being the protocol most programmatic vendors follow when they build their platforms in order to encourage interoperability within the ecosystem.

The cons? Well, many premium publishers early on had their yields eroded by programmatic and channel conflict – the fact that advertisers were buying their inventory/audience for a lower CPM via RTB than their direct sales team sells it for was an area of concern for many. However, fast forward to 2017 and channel conflict is less of a concern as major SSPs and exchanges such as The Rubicon Project, Index Exchange and OpenX are no longer in iteration 1 and provide very powerful tools to publishers that enable them to manage things like dynamic floor pricing, advertiser controls and bid landscape insights.

Data leakage
The concerns are now around things like data leakage and how technology is empowering advertisers to find the same or similar audiences on other environments that have lower price floors. This has changed the balance of power from seller to buyer as the algorithmically-powered bidders licensed by buyers autonomously direct demand based upon audience and price. This dynamic has meant some premium publishers have had an uphill struggle on their hands to protect and grow their digital revenues in this new dawn of data-driven buying.

Net result? The machines have been trained to find audiences and at times can be blind to a 100-year old premium brand and instead favour an app that launched six months ago that has little or no contextually relevance for the advertiser they are buying for.

Although many buyers now leverage third party contextual data to ensure their DSPs are not only finding audience (Cookie or Device ID) but the environment their ad is served in is also contextually relevant, I think we are starting to see the pendulum swing back in favour of premium publishers that can provide both audience and context within a controlled, open and transparent environment.

Emerging models
Issues such as those raised by The Times and the current focus on things like fraud (Methbot et al) are all going to be catalysts that force the buy side to better understand the make-up of their supply partners.

New, emerging programmatic buying models, like Programmatic Guaranteed, are empowering the buy side to still transact via OpenRTB but are now enabling them to secure the same pricing and volume commitments from publishers they could have secured yesterday via the IO. This means they are no longer exposed to the unpredictable dynamics of biddable open and private exchange environments and can guarantee both price and volumes to the advertisers they are buying for – a big win for the industry to encourage more brand spend to transition from IO to programmatic.

I think we are going to see the debate of audience versus context play out a lot this year on the AdTech panel circuit. I am constantly involved in debates about what is premium in mobile and how we should collectively be valuing gaming and utility apps if audience can be quantified within these environments. Do we empower the DSPs to find an advertiser’s audience regardless of the environments (News, Gaming, Utility etc.) they originate from, or do we force the DSP to work within the pre-defined boundaries of a publisher whitelist as this is perceived to be the only way we can truly mitigate against things like fraud & fake news.

The bottom line is that buyers will always be looking for cost efficiencies and scale. However, premium publishers are now in a stronger position than ever before as CMOs start to ask more informed questions of their programmatic activation and supply partners. Expect the debate to continue, as that narrative at MWC this year will be around audience versus context for the many businesses that are built around the App ecosystem.