[Editor’s note: Aaron Bell, CEO at AdRoll, was a speaker at #RampUp17. We invited him to share his thoughts on data and marketing trends.]
Marketing attribution has been a headache for marketers because it’s a seemingly intractable problem. Over the years, marketers have struggled with too many options and too much data, leaving many stuck in ‘analysis paralysis’.
Now there’s more urgency to solve the attribution puzzle. At AdRoll, our annual industry survey confirms that, in 2017, almost 75 percent of marketers find attribution critical or very important to their programs’ success. In addition, a recent eMarketer survey showed for the first time more than 50% of marketers plan to use a multi-touch attribution model in 2017. Marketers realize in a multi-device world that’s gearing up for artificial intelligence-based solutions, accurate attribution is key.
RIP Last-Click Attribution
Until recently, last-click attribution has served as a “good enough” solution. This method is a holdover from the early days of search advertising. Since advertisers only paid Google when someone clicked an ad, the click became all-important.
When Google launched Google Analytics a few years later to track visitors’ activities on websites, Google applied the same thinking. Since no one could be sure why a consumer made a purchase, Google assumed that the click right before the consumer’s purchase was responsible for the sale.
This offers an incomplete picture. What about all the online and offline ads the consumer saw before that click? What about the vast majority of people who don’t click on ads? That’s why so many industry leaders are moving away from last-click.
[Editor’s note: learn more about moving past first- and last-click attribution models with our Half-Blind Attribution Modeling Guide.]
The Missing Link: CTR and Brand Effectiveness
Clickthrough rates are great for measuring direct response, but not for branding ads. There’s little correlation between CTRs and ad ROI, according to Nielsen.
Buying ads on a per-click basis is also pricey. “Clicky” users cost 5.5X more than less-clicky users. Why? Because so many people are buying clicks.
Buying less-clicky consumers saves money and delivers the same ROI. Advertising works just as well on those consumers.
Changes in media consumption over the past few years have rendered CTRs more irrelevant. Consumers aren’t just on desktop anymore, but use up to 7.2 devices. The rise of nondirect response formats like passive scroll on Instagram and video isn’t reflected in a click-oriented world.
This environment demands a more nuanced approach.
What Marketers Can Do Now
To address the shortcomings of last-click, many marketers are building custom multitouch attribution models. AdRoll’s approach has been to first gauge the market. In 2016, we worked with eConsultancy to survey marketers in the UK, France, and Germany on this topic.
The survey found 72 percent of respondents said a perfect marketing attribution model is impossible. I agree.
The premise of a perfect attribution model is that there is a pristine end state and a perfect measurement. Actually, marketing attribution is an ongoing process of improving decision-making.
Rather than wait for a perfect solution, I advocate incrementalism. An incremental approach to measurement compares the success metrics for an audience exposed to an ad, campaign, or channel to a control group that hasn’t. The difference between the two groups shows whether one of those prompted an incremental lift.
The Future: AI and Beyond
Our annual State of Performance Marketing report shows 60 percent of marketers plan to change their attribution model in 2017. One reason we’re seeing increased urgency is marketers realize they need to solve this problem.
A weak attribution model will sabotage efforts to fully leverage the power of data. Cross-channel attribution is essential for real-time optimization of people-based behavioral data.
If marketers don’t square this away, they won’t be able to harness AI solutions, either.
The way forward isn’t easy. Marketing attribution will continue to be a headache, but marketers can ease the pain with a scientific approach. That means avoiding cherry-picking data and instead looking at the data as a whole.
Likewise, siloed units like search, display, and affiliate need to put their respective goals aside and look at what’s working for the entire company. Only then will marketers begin to make progress.
Aaron Bell was a speaker at RampUp 2017. Watch him in the panel, Hot Zone: The Marketing Industry’s Most Difficult Questions Answered, below.