02 Nov 2023 | Steve Taylor

From hot mess to hot prospect: 5 ways to change digital advertising

Opinion: Strategy Leaders

We must focus on people, partnering wisely, and let our communication goals guide our channel choice, argues VCCP Media’s strategy chief.

Online media needs to be dragged into better advertising, according to this publication’s editor-in-chief.

It seems that outside of adtech’s inner circle, you’d struggle to find an opposing view. Digital advertising is a truly hot mess.

Almost everyone I talk to has one of two questions about it. “How can I carry on exactly as I am, regardless of cookie deprecation, privacy legislation, technical tracking challenges, the whole farrago of fraud, made-for-advertising sites, brand safety challenges and supply chain murkiness?”

Or, less commonly: “Digital advertising is totally FUBAR, how do we fix it?”

Uncomfortable questions

The first question has a short and usually unwelcome answer. You can’t. And even if you could, you shouldn’t. It’s ever more obviously not what customers want and not what advertisers want.

But, in some quarters, there’s a lot of money still being made from the way we approach and trade programmatic media, so not surprisingly it’s a top-of-mind topic for many media leaders.

Turning to the second question, if we’re in the business of fixing things, what could we do? Can we turn digital advertising from hot mess to hot prospect?

I’d best say what I mean by ‘digital’. As Nick Manning points out, the term is almost meaningless and covers many things.

Here I mean online display and video, on desktop and mobile web; because that’s where the biggest problems are, where there’s most doubt about effectiveness and value and where there’s an easy to-do list of five things that could fix it.

1. Forget targets, think people

Bereft of third-party cookie pools, we’ll have to find a new way to reach the people for whom our message is most relevant. There’s lots of talk of federated learning — see Google’s current proposals — but I think the answer might be simpler and more privacy centric.

Firstly, there’s a tech bleeding-edge solution using multiple data sources, sometimes using thousands of publicly available anonymised data points as well as first party data, to build a profile of relevant people and where they are most likely to be found, combined with data on when they are most likely to be there. This is done at a micro-geographic and time based level and bids are optimised towards a stated goal by an algorithm accordingly.

Secondly, contextual targeting is an attractive option. Using data about the content on the page is not a new means of targeting, but it is starting to look an attractive alternative to cookie based targeting.

And finally, let’s not forget that one obvious way to target people is through that good old targeting workhorse, audience profiles.

Working with publishers to place ads in places you know large volumes of your audience hang out doesn’t sound very technically complicated and there are far fewer opportunities here to make a quick ad tech buck. Yet, it’s wholly privacy centric, and sees media owners better remunerated for the valuable asset they have: an audience, quality content, and the contribution of their own brands.

A mix of all three represents the optimum approach and fixes a lot the problems programmatic currently faces, rather than just replacing them with new ones, which I think any tracking based solution, however federated, is likely to do.

2. Choose your partners like you’d choose, well, your partners

In media, you are the company you keep. Where ads appear says something about who a brand is for, and therefore what it is. Long tail programmatic has made media very cheap (back to that trading elephant in the room) but at what cost to its effectiveness?

We all know, instinctively (and the research backs it up) that advertising works harder when it’s in a quality environment, and that bigger more prominent formats work the hardest of all.

Buying better ads, with quality publishers means higher CPMs, of course. But if that drives effectiveness, and cost per meaningful action or outcome actually falls, then it makes solid business sense.

3. Your focus should be on business impact, not media metrics

And so our focus should be on business outcomes. Goodhart’s Law, roughly paraphrased, is that when a measure becomes a target, it ceases to be a good measure. The ‘system’ gets gamed and you find yourself tasked with driving click-thru rates instead of selling insurance, or cars, or whatever.

The answer is proper measurement frameworks that account for the mid and long-term, not just the short, and which use all the tools at our disposal to understand what’s working and how.

The revolution in computing and AI means we can easily and quickly run econometrics and predictive models using open source code from the likes of Google, Meta and Uber (nb: human expertise still very much required!). We can measure attention and use that as an insight point. And we can do all this alongside quantitative brand uplift studies. All helping us understand how our media, digital included, is affecting business outcomes.

It’s much harder work than reporting a click-through rate, but the reward for digital might well be a renewed interest in what it can do for brands, branding, and long-term business success.

4. Actually test, and actually learn

While budget and data science resource may be limited, all are able to do test and learn experiments. The trick is keeping the tests clean by testing only one thing at a time.

Too many advertisers buy into the test and learn approach without co-ordinating the tests across different agencies or departments and/or testing too many things at once.

Doing it properly is slow, but astonishingly powerful, and very convincing in the boardroom.

5. Your communication goals should guide your channel choice, not the other way around

In days gone ‘digital’ couldn’t get on the plan, because the money was already allocated to non-digital channels.

We now often have the opposite problem where money is allocated first to performance-driving PPC and direct-response paid social and there’s nothing left for any brand activity. Display and video suffers doubly here as when thoughts do turn to mid- or upper-funnel activities display and video for branding is last to be considered because “there’s already digital on the plan” and “you can’t build brand in digital display”.

We mustn’t consider ‘digital’ as an amorphous lump, echoing Manning’s point again. All channels work differently and each should have its own contributory role on the plan. Digital display and video might then sit more commonly in the upper funnel brand building top half of the plan.

As to whether you can build brand with digital? With scrappy small formats mostly not seen by humans, at the far end of the internet with no co-benefiting context and publisher credibility; well no, you can’t.

But what about big, attention-grabbing formats? With good quality publishers? That are well targeted using clever approaches to data and audience understanding to drive efficiency as well as effectiveness? And all as part of a thought-through approach to testing, learning and measuring what works?

That’s a long way from the current hot mess. It would make digital advertising a hot prospect.

Steve Taylor is joint chief strategy officer at VCCP Media.