"Data-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Andrew Shebbeare, founding partner and global innovation director at Essence.
Awesome. We have the tools to optimize in real time. Unfortunately, brand marketers can’t measure success in real time. Not even close. Even implanting Orwellian mind-reading devices into the world’s consumers won’t get past the problem that it takes time for people to form opinions.
So what should a brand marketer use to plug into their programmatic bidder, crack the whip on their agency or show off to their boss while they wait?
For many marketers, the default answer is “engagement.” It’s a term that has grown to encompass everything from not skipping a video to expanding a banner or from browsing site content to tweeting about it. These auspicious signals have become the currency of digital branding. These big, impressive numbers have come to symbolize the power of online advertising over other media.
Yet I will argue that chasing engagement can cause marketers to make exactly the wrong investment decisions. Too often, they end up moving dollars diametrically away from the places they are working best.
How It Happens
The audiences most likely to engage with your brand are those that like you already. Optimizing toward engagement at the expense of other aspects of your campaign’s impact typically means moving away from audiences that might need to hear about you so you can preach to the converted. I’ve consistently seen higher awareness baselines and lower brand lift per dollar from placements that skew toward engagement.
If you’re very lucky, perhaps one in five people won’t skip your video ad, one in 30 will expand your banner, one in 300 will click through and one in five of those clickers will spend some quality time with your site. Chasing engagements means basing your investment decisions on the behavior of a tiny, unrepresentative minority. It is ironic that many of those who rightly spurn click rates as a brand metric then rely on site engagement as a brand campaign diagnostic. But most people who engage with site content will have clicked to get there.
There isn’t any solid proof that engagement drives brand love. The two are correlated, but which way does causality flow? There are plenty of studies comparing engaged audiences to non-engaged audiences and/or a control. But this is bad science. Engagers are not like those in the control group; they are different people who probably like you more. The only fair comparison is between the overall control and exposed audiences – engaged and not. Since only a handful of people engage, you’re very unlikely to see the difference between a rich ad and a standard one at this level.
Engagements are expensive. When you cut a cost-per-engagement (CPE) deal, you pass on risk to publishers and they naturally price it in. If you work back what your CPE media costs in eCPM terms, you might think again. You also pay the price in terms of control, usually losing control of reach, frequency and even audience targeting, all of which are the critical fundamentals of brand marketing. Publishers need to earn a crust, so they will serve what makes them the most money, not what carries the message most effectively.
When we chase engagement we risk overlooking the fundamentals. Are we reaching the target? Are our ads getting seen or noticed? You are probably also spending most of your design thought on a tiny fraction of your audience. Your unexpanded ad is what 97% of the audience will see, probably for less than two seconds most of the time. If you put thought into making those two seconds matter, you will make a much bigger difference.
An Alternate Construct
I am not suggesting that engagement is worthless. All other things being equal, I would much prefer that people spend time with my brand than not. My point is only that fixating on engagement risks doing more harm than good.
It’s all about order: Optimize in the same sequence as the potential of your metrics to make a difference. This will ensure that you hone in on performance each time while never allowing one metric to lead you astray at the expense of anything more fundamental.
Here’s a suggested hierarchy:
- Brand safety: An ad served in a bad place can destroy value.
- Viewability/fraud: An ad that’s never seen by a human can’t add value.
- Audience composition: An ad that’s seen by the right human has a better chance of working.
- Reach/frequency: Two sides of the same coin; ads are more effective and efficient when seen the right number of times by the right number of humans.
- Attention: Signals, such as time in view or scroll speed, point to the likelihood of an ad being noticed and seen, which can help us find the “best” impressions.
- Engagement: Once all of these boxes are checked, go ahead and optimize toward engagement. Just don’t compromise the first five metrics in the pursuit of No. 6.
Avoid The Trap
This isn’t rocket science. You could probably accuse me of an overly elaborate restatement of old-school marketing basics. If you have good creative, you should be able to place your trust in it. Think about getting the right message in front of the right people in a manner they will notice and a frequency that is appropriate before you think about reacting to the behavior of an engaged minority.
None of the above is a substitute for proper brand measurement. These are stepping stones that help keep us on track or make more granular decisions than brand studies can support.
This story is neatest at the very top of the funnel. Brand awareness can only grow among audiences that haven’t heard of you. It gets fuzzier in the middle, where gains in persuasion, relevance, feature awareness and the like can happen even among brand advocates.
Clearly converting a brand-aware audience into intenders demands a different media plan and creative treatment than establishing a new brand. The impact of these filters will be different in each case. But I would argue that this “hierarchy of needs” still holds true, even if the power of each filter varies and engagement becomes a more important splitter.