One of the bestselling items in Amanda Nielsen’s e-commerce store is a ceramic coffee mug inscribed with a caption whose sentiment is near and dear to her heart: “Marketing attribution is fake,” it says. “We literally made it up.”
Nielsen, who now works in sales at a tech company, started her career in marketing. While she had a “pretty mediocre” base salary, part of her compensation was also tied to an attribution model that assigned scores to various customer touchpoints (like paid ads, organic blog posts, webinar signups). If customers visited any of the touchpoints Nielsen worked on before they converted, she’d get paid for those. But if they didn’t — she didn’t.
Common Marketing Attribution Models
- First-touch attribution: Gives credit for the conversion to the source where the customer first encountered you (within the period of time you’re looking back).
- Last-touch attribution: Gives credit to the source the customer last visited before converting.
- Linear attribution: Distributes credit equally across each touchpoint that was visited before the conversion.
- Position-based attribution: Gives 40 percent credit to the first touch, 40 percent to the last touch and spreads the rest across the other touchpoints.
- Time-decay attribution: Gives progressively more credit to touchpoints as they lead up to the conversion, with most of the credit given to the final touch. Based on an algorithm.
- Last non-direct-click attribution: Gives all the credit to the last source visited before the conversion, excluding direct traffic.
- Custom attribution: Relies on machine learning and historical data to distribute credit across touchpoints.
“Marketing attribution was what decided whether I could pay my rent,” Nielsen said.
What frustrated her about the arrangement was how arbitrary and at times inaccurate it could be (she once noticed a colleague incorrectly transfer the scores to a spreadsheet) in addition to how variable it was. And yet it dictated where marketers like Nielsen focused their efforts.
“It made us lose sight of qualitative aspects of our strategy, like how people perceived our brand,” she said. “It just left this horrible taste in my mouth.”
Nielsen isn’t alone in her frustration. The practice of determining what marketing touchpoints or channels contribute to a user’s conversion, however well intentioned, has gotten out of hand, according to several marketing professionals.
‘It’s a Bit of a Fantasy’
The goal of attribution is to bring accountability to marketing measurement and connect the dots, according to Samuel Brealey, a U.K.-based marketing consultant.
“But on the flip side, it’s also a bit of a fantasy,” Brealey said.
The inherent assumption of marketing attribution is that everything is causal — A leads to B leads to C. Problem is, people don’t really buy that way, especially when it comes to B2B. Sales cycles can be notoriously long and complex, often involving various stakeholders and dozens of encounters — both online and offline — which are impossible to measure accurately with a high degree of confidence.
To stare deeply into the minutiae of attribution data is to risk ignoring the bigger, more complete picture of what marketing is doing, Brealey said. Attribution data fluctuates rapidly; locking onto it with tunnel vision is myopic.
There’s also the problem of subjectiveness. Given the range of multi-touch attribution models available — not to mention custom ones that some companies have built for them — deciding which one is correct often comes down to the point of view of the person pushing for it.
The agency Nielsen worked for decided that bottom-of-the-funnel touchpoints ought to be given the most credit. That decision led her to peel her focus away from top-of-funnel touchpoints, even though she expected to find more traction there.
‘It’s Not Necessarily Nefarious, but It’s Naive’
If an overemphasis on attribution data is a problem, why did it come about in the first place?
When digital marketing took off, there were suddenly tons more data points to measure and “you had digital natives trying to prove their worth [by carving] out a piece of the budget for emerging channels,” said David Akermanis, strategy lead at Quietly Media, a content marketing agency.
And the people who stand to benefit most from stressing the importance of “perfect” attribution data are often either the software vendors (who want to sell sophisticated attribution software) and marketing agencies (who often want to take credit for that for which they’re paid), Brealey said.
“It’s not necessarily nefarious,” he added, “but it’s naive.”
Brealey is a fan of marketing mix modeling (or econometrics) for measuring the impact of marketing activities. Unlike attribution models, this measurement method takes external variables into account and doesn’t narrowly focus on the individual user level.
For years, though, measuring like this has been too costly and complex for most smaller in-house marketing teams to use. Brealey said he’s noticed that slowly changing, however, with better tools cropping up.
‘Consider Most Web KPIs to Be Directional’
Spending too much time debating the merits of specific attribution models or vying to get pristine web data is a fool’s errand, according to Adam Singer, vice president of marketing at LEX.
Instead, marketers should do what they can — set up tags correctly, apply the right filters, clean the bot traffic out from the analytics — and from there, treat attribution data as binary; it’s either getting better or getting worse.
“The marketing attribution problem is one that will never be solved,” Singer said. “You can make yourself crazy or you can consider most web KPIs to be directional.”
Directionality is about “moving the conversation away from, ‘Well, how do we know that X led to X?’ and more toward, ‘Well, it doesn’t matter, as long as everything is in the right direction,’” Brealey said. “You shouldn’t obsess over the day-to-day changes of things.”
Brealey also advocates for client-side marketers to resist the push of agencies to do more digital (measurable) marketing and instead stick to their guns when they know what is working for them — be it digital, television or out-of-home advertising.
Marketers may want to focus on measuring what matters to their business, as opposed to everything, Akermanis said. And to be scientific about it, too, making and testing hypotheses about marketing’s impact and using that to “drive the conversation about how your marketing should work overall.”
It remains to be seen if the tide will turn or if we’ll have to wait for the cookiepocalypse to obliterate attribution as we know it.
In the meantime, marketers will surely be watching — and sipping, perhaps, from their “Marketing attribution is fake” mugs.
