Attribution has long been a headache for digital marketers.

When a customer clicks on an ad and ends up buying a product, then that ad is responsible for convincing the customer to buy the product. But what about all those other ads that led to the final click?

What is multi-touch attribution?
Last-click attribution gives credit for the sale to the last ad that the customer last clicked. But, as brands spend more on traditional platforms like audio, TV, and out-of-home, it creates two problems for the brand: one, because these aren’t digital platforms, a brand can’t simply trace the number of clicks to determine if a customer ends up buying something. Second, the more marketing channels a brand has, the harder it is to tell where a customer first heard about the brand.

Multi-touch attribution attempts to chart how a customer moves down the purchasing cycle, from the first ad they viewed to the last, and determine which of those marketing channels was most important in ultimately getting the customer to buy.

“Multi-touch attribution is really about trying to make sense of what in your marketing mix gets credit for what,” said Chris Toy, the CEO and co-founder of MarketerHire, a platform for freelance marketers.

Who cares about it?
Every marketer, unless they’re just dumping all their money into Google and Facebook ads. Figuring out what ads work and how much they contribute to growing sales is critical. What’s more, platforms and publishers that aren’t specialists in driving clicks want credit for the help they give to a customer taking a desired action.

What are the different ways to measure it?
There are a variety of multi-touch attribution models, with each assigning a different level of credit to different marketing channels. There’s linear, which gives all of the different places where the customer hears about the product an equal level of credit for the purchase. So if a customer first views a Pinterest ad and then clicks on a Facebook ad before buying a set of cookware, both the Pinterest ad and the Facebook ad would be equally considered responsible for the purchase.

There’s also time decay, which gives more credit to the ad the closer it is to the bottom of the funnel. U-shaped, meanwhile, gives more weight to the channels responsible for the first and last touchpoint.

Marketing platforms see an opening in helping brands figure out how to measure multi-touch attribution, since it requires having the resources to measure many different channels. Google Analytics, for example, has a multi-touch-attribution modeling system built-in. Vendors like Neustar and Nielsen Visual IQ also sell solutions that help retailers measure multi-touch attribution.

Is any one model better than the other?
No. Marketers say that there’s no one-size-fits all multi-touch-attribution model, because it depends on what type of product a company is selling and how long it typically takes to convince a customer to buy. For example, if you have a low-cost impulse purchase, like sunglasses, a customer may only view a couple of ads before buying.

“You have to start thinking really consciously of what each of these channels did,”said Will Flaherty, vp of growth at DTC health-care brand Ro. “You could have a model that puts more weight on the channel that drives an email sign-up, because by getting someone into an email flow, there’s a higher likelihood that that person will eventually convert. You really just have to think through the journey of the customer.”

Are there any models that DTC brands typically use?
Many DTC brands, particularly those who have an affinity for developing marketing in-house, are instead trying to build their own custom attribution models. To do so, they have to have multiple methods of collecting data to determine what ads a customer may have viewed — that could include collecting click data, or developing a post-purchase survey that asks the customer to tell the company where they first heard of the brand.

DTC brands also need to be sophisticated enough in their data-collection abilities that they can track a customer through the entire purchasing and consideration cycle.

Custom-suit manufacturer Indochino is one such brand that’s developed its own custom attribution model. In 2017, Indochino started investing in building its own model that attempts to take into account “traditionally unquantifiable channels,” including public relations and storefront presence, according to the company’s marketing director Lisa Craveiro.

“For us, lifetime value is really important,” Indochino CEO Drew Green told Modern Retail in June. “And, importantly we know the lifetime value within a 12-, 24-, 36-, 48-month window of each channel.” So, Indochino’s model focuses heavily on that.

However, there are drawbacks to developing a custom model.

“Multi-touch attribution takes a lot of work to do accurately on your own and is very expensive to do with a partner,” wrote Evan Woods, head of growth for DTC pet food brand Ollie, in an email. “To effectively do this in-house, it requires support from data science, tech and marketing teams, and plenty of time to tweak and get your attribution right.”

How important is multi-touch attribution going to be?
Toy believes that younger DTC brands can spend too much time obsessing over creating and refining their own multi-touch attribution system.

“It’s distracting from what is most likely the actual problem — which is your ads suck, or your brand’s not quite right, or your pricing’s wrong,” Toy said.

He said that it makes more sense for companies to start thinking about how to develop a more sophisticated attribution model when they’re “bumping up against $10, $15, $20 million in sales,” though they may be able to earlier if they’re a venture-backed company with plenty of cash to go around.

“When you are a large company, you can afford it, and it’s worth your while to invest heavily in finding a 2%  improvement [in marking spend], ” Toy said. “That’s a lot of money when you’re making $200 million a year in sales.”

A version of this post first appeared in Modern Retail, Digiday’s sister site.

  • LinkedIn Icon