As companies market across more channels, it only intensifies an age-old problem: Which ads or promotions worked to drive a sale? Marketers refer to it as "attribution." You might know, for instance, that a customer clicked on an email offer and bought something, but did a prior direct-mail offer, a search keyword buy, a Web or print ad, or a social media interaction make that customer more receptive to the email pitch?
Attribution determines where CMOs spend their marketing dollars. In a conventional, direct-attribution approach, companies look at each channel separately, measuring response within that channel without considering other efforts. Now companies are coming up with ways to study customer-interaction histories across channels and give credit where it's due. But this has mostly relied on crude rules applied manually or replicated within marketing management systems. For example, a system might assign a sale to the first or last marketing touch, or it might average the attribution across all marketing interactions with a given customer.
Advanced analytics hasn't cracked this nut yet, but vendors are trying. IBM, for one, last month released an Attribution Modeler application that uses advanced algorithms to assess the impact of efforts in each channel.
Macys.com uses SAS software for attribution analysis. "You need to understand the causation and correlations between digital and offline activities so you know what's triggering which behaviors and which activity drives the next," says Kerem Tomak, VP of marketing analytics for the online retailer.
Did display advertising drive traffic to search and then to a website, or did the interaction start with search? Is a visit to a website a failure if a customer puts an item in a shopping cart and then abandons it, or did the customer decide to go pick up the item at one of our stores? Can individual campaigns lose money while still moving customers a step closer to a valuable sale?
Macys.com is using its attribution modeling techniques to allocate marketing budgets by channel and to determine which products are best promoted through which channels -- "building a bridge," as Tomak describes it, between marketing and merchandising decisions.
"If you track interactions, attribute correctly and test your models so you know you can trust your analysis, then you have a very powerful tool that will help you orchestrate everything you do," Tomak says.
This Isn't A Threat
The complexity involved in knowing how best to market to your would-be customers is only growing. Forty-four percent of store shoppers surveyed during the last holiday season said their first step was to go to a store, but 20% said they went to that retailer's website first and 10% said they started with a general online research. That's according to a study of more than 24,000 consumers across 100 websites, 29 retail stores and 25 mobile sites, conducted by customer experience analytics vendor ForeSee.
Among mobile buyers, 43% said their first choice would be to buy in store, but they ended up buying through a mobile app, likely because the item was out of stock or they looked at the item in store and found a less expensive option online.
In this kind of multichannel environment, marketing is a whole lot more than a brand message. It doesn't matter if the CMO or CIO is signing the checks, companies must use technology to interact better with their customers. They need to better understand their customers so that they can respond more quickly to changing buying patterns and please customers in every interaction, be it online, on the phone or in person. As IBM's Kennedy puts it: "How we operate can be a bigger factor than what we communicate."
IT pros shouldn't see these marketing trends as a threat. The Red Cross's marketing tech initiative meant spending more on the IT side and hiring a director of data strategy. The lesson: Whether they're supporting customers or personalizing an email marketing message, great technologists remain at the center of making sure the customer experience lives up to the marketing promise.