Why historical data segment is history, and three ways forward for ad tech
"What the old mantra of ‘right audience, right time, right message’ describes is an approach that intrinsically cannot scale."
Ever since advertising and tech became joined at the hip, the promise of real-time media buying has dazzled us all. Especially when twinned with the incredible potential of data-driven targeting. Out went demographic estimation, in came empirical truth about your audience - or so we thought. And who hasn’t heard the now grating claim about delivering the holy grail of advertising, reaching the right audience at the right time, with the right message. Backslaps all round.
The fact is that truly real-time data-driven media buying is still surprisingly rare. No one is in denial about the primacy of data. But curiously, ad tech’s dirty secret remains that so much of that data is still historical. Which seems a pretty obvious anomaly, in a market where transactions take place in the blink of an eye.
Instead, 3rd party data segments are still propping up the real-time buying economy, based on a rickety mix of assumed or estimated behaviour, and signals recorded in the past. And that lag between data collection and deployment on its own is a serious flaw.
One for All, All for One
But as if that wasn’t enough of a failing already, there is yet another critical issue.
And that is data exclusivity, or rather the lack of it: if I can buy segment X, it’s also at the fingertips of any of my competitors (and any number of other businesses.) This immediately raises questions around impact on the end user, not to mention the quality of the data itself.
Let’s say our segment assumed incorrectly you’re an adolescent male. In actual fact, you’re a woman in your fifties. The first men’s razor brand to advertise to you would seem odd, maybe even raise a laugh. The second or third however, becomes a major annoyance.
This isn’t an entirely made up scenario either. We can’t know for sure, but when Gilette sent free ‘welcome to manhood’ razors, among others to a 50-year-old woman, you have to wonder about their data sources.
And the drawbacks of 3rd party data don’t stop there. In fact, one piece of research found a provider identifying fully 84% of users as both male and female. Overall, 4 out of the 11 vendors tested weren’t “much better than chance in targeting age and gender.”
In February, Lotame, one of the largest 3rd party data exchanges, said it had removed four billion profiles found to be botsor otherwise fake. And lest we forget, the furore around Facebook and Cambridge Analytica all stems from over-liberal sharing of personal data with third parties. That's putting it mildly: and now the public is starting to tune in to these practices, many aren't fans either. All of which doesn’t exactly bode well for GDPR, and the upcoming ePrivacy law. Little wonder some are asking how 3rd party data can even survive in its current form, with user consent required?
“From fraud, to bots, to the limitations of historical data in a medium that is constantly changing – it’s there for all to see. But that doesn’t mean there’s no hope of progress.”
All of this would give any advertiser pause for thought. But beyond even the most glitchy data sets, using historical insights quickly presents other, even more glaring problems. Especially in a live, ever-changing digital context.
Take retargeting for instance. Who hasn’t been chased around the internet based on a site visit to their bank, TV provider they already use, or other service they never intend to sign up for? Even though this may technically resemble an accurate, ‘data-driven’ approach, retargeting in the main can still be woefully annoying to users. Surely if the technology underpinning advertising was anywhere near as ‘real-time’ (not to mention artificially intelligent) as it makes out, it would try to measure that too? In practice, blanket user retargeting is just another sign of some of the conflicting motives at the heart of programmable media.
The End of Wastage
Much has been made of the promise of programmatic – from the potential to buy across all media and different screen types, to the end of wastage. But the truth is, while we have seen plenty of progress on the former, the latter is effectively still stuck in a rut. From fraud, to bots, to the limitations of historical data in a medium that is constantly changing – it’s there for all to see. But that doesn’t mean there’s no hope of progress.
There are three things I think we need to bear in mind, so we can finally move forward:
- The first is that however much we talk about AI, many ad tech products are still fundamentally pretty basic. Now more than ever is the time to develop genuinely smart, sophisticated technology.
- Second, approaches to programmatic need to become much more varied for the sector to evolve – after all, if everyone uses the same tool, it quickly gets worn out.
- Third, what we need is a less uniform focus on micro-targeting – a tactic that has led to many of the ills described above, and by its very nature an approach that is limited in scale. More R&D and investment is required into other - no less viable - ways of reaching consumers who show intent, in a variety of different ways.
Historical 3rd party data is by its nature a good that everyone and anyone can buy. Especially now, with GDPR upon us, segments have had their day. Expect them to be confined to history, along with other techniques whose success rate is ‘not much better than chance’.
What the old mantra of ‘right audience, right time, right message’ describes is an approach that intrinsically cannot scale. It’s also stuck in the past - much like the historical data segment itself. Time to move forward and leave it there, where it belongs.
If personal targeting is the past, the future is in context. illuma uncovers relevant audiences and new prospects, without 3rd party data – get in touch for more info, or to set up a product trial.