Developing cross-channel marketing capabilities was only the beginning. Making the most effective, profitable, and productive use of those channels means understanding how each and every touch and impression affect the customer journey and, ultimately, purchase decisions. Technologists are building new platforms to crunch the numbers from ever-growing pools of first- and third-party data, and data scientists are refining models that give credit where credit is due to the offers, branding, and messaging every customer is exposed to.
Welcome to the world of cross-channel attribution.
At its heart, cross-channel attribution helps marketers identify the strength and relevance of each of their marketing initiatives down to its component parts, from keywords and search engines, to creative and offer, to channel, day-part, and placement. There is no single master attribution formula for any brand. Rather, attribution helps marketers understand which of those factors are relevant to particular segments and campaigns. Database and app platform developer MongoDB is a perfect case in point, as the company markets to two distinct audiences.
Clients in its small to midsize business audience are reached primarily through paid search and email. Enterprise-level clients receive a full account-based marketing approach, which adds a variety of new channels, including webinar and white-paper content, as well as social media outreach. Each audience needs its own attribution model.
Adopting an attribution discipline uncovered some surprising positives for MongoDB, including a much heavier-than-expected contribution from a Facebook retargeting campaign aimed at small businesses.
“It impressed us how well Facebook did, so we’re now working to figure out how to apply Facebook on the enterprise side,” says Meagen Eisenberg, MongoDB’s CMO.
Less than a year after engaging several partners, including Oracle Marketing Cloud, to assist with an attribution overhaul, MongoDB is now shifting its channel spending on a quarterly basis based on performance. Total lead volume per quarter has already tripled.
“And the percentage of marketing influence on our pipeline is increasing, because we know more of what we’re doing to communicate to leads in the pipeline,” Eisenberg says.
Brands with sophisticated cross-channel attribution models are still in the minority. According to Webmarketing123’s 2015 State of Digital Marketing survey, 38% of marketers had no attribution model in place whatsoever. Another 34% relied on just a single touchpoint. The reason is simple: Attribution isn’t easy.
The size and scope of attribution modeling can be daunting because consumers are increasingly mobile and cross-channel in their behavior. Forrester Research has identified the segment of the “always addressable” consumer, defined as using at least three smart devices from different locations throughout a given day. As of 2015, more than 60% of U.S. adults met its definition.
This is a case of good news–bad news for marketers. Marketers can reach always-addressable consumers at any time of the day with messaging that an attribution model recommends as ideal to advance them to the next stage in the buying cycle. Using multiple devices and IP addresses, however, makes the majority of consumers harder to pin down at any one time.
That’s why the first step in an attribution-readiness program is to create and then align around a common identifier. Loyalty programs or other individual sign-ons are a good place to start, but are unavailable at early stages of the customer journey. Furthermore, not all brands have a compelling reason to ask an individual to present a loyalty ID or other sign-on across all of their most commonly used devices.
Attribution above all
Data vendors can reinforce tracking with a host of solutions to link IP and home addresses with device and household associations. Cookies and tracking pixels can help, but for long buyer cycles, particularly on the B2B side, they may not be enough. Cookie deletion and device replacement can interfere with these IDs.
“If your conversion window is longer than the stability of the identifier, then meaning is lost,” says Ric Elert, president of Conversant Media.
The more intricate the device and data matching, the smaller the resulting data sample size. This means attribution-driven marketers must make a crucial choice: refine the model until they can build larger sample sizes or commit to campaign directions and spending based on a smaller audience than usual. Regardless of the selection, rigid attention to data cleanliness and ongoing testing are vital.
“If your ID match rates are inaccurate, your attribution metrics will be, as well,” explains Alex Hooshmand, VP of product management for Oracle Marketing Cloud.
Elert isn’t bothered by small sample sizes.
“Attribution comes down to one-to-one marketing because every person is going to have a different reaction across channels,” he says. “You want to be able to understand personalized offers across channels. Otherwise, you end up back to doing things in big, broad swaths.”
Tying those identifiers and other data sources together requires organizational collaboration and alignment.
“When a [company’s] media, marketing, CRM, and email teams all sit on different floors, it’s difficult to wrap your head around what’s needed to build true attribution and the data that needs to be fed into it,” says Marisa Skolnick, digital director at Mindshare.
Above all, marketers need to be ready to make a clean break with the simplistic attribution that may have guided their marketing strategy for years. First-touch attribution seemed like a revelation because it identified the seed from which all future customer impressions and decisions grew. But first-touch doesn’t capture the reasons why two seemingly equal prospects would end up at different outcomes. Cross-channel attribution captures the activities and exposures that inform the decision beyond an initial awareness campaign.
Last-touch attribution has been a perennial favorite, as it’s comparatively easy to capture and reinforces the old “right message, right time, right channel” mentality. But it fails to recognize the development of a buyer’s interest over time and through various messages and touchpoints and does little to understand the impact of offline, less-addressable media. Sticking with last-click mostly satisfies digital media partners.
“The search guys are petrified by attribution because they’re really making out when [marketers] use a last-click methodology,” says Visual IQ CMO Bill Muller. “But attribution isn’t doom and gloom for search: It’s another way of looking at success,
including finding other keywords and engines.”
Cross-channel attribution presents marketers with statistical bases for doubling down on high-performing channels and messages and reducing or eliminating spending on the laggards. It should be more than that, however. It should be seen as a series of spotlights and alerts that demonstrate where marketing results aren’t in line with expectations, identifying opportunities to improve where results are missing the mark rather than simply shifting the majority of spending to a few high-performance media.
It also helps marketers already working with recency and frequency modeling to better understand the effects of repeated exposures and overexposures.
“You can understand the incremental value of recency and frequency, as well as the impact that fatigue builds in,” says Jon Casciari, VP of analytics at Epsilon.
Attribution also refines the understanding of how much the magnitude of an offer, as opposed to the timing of or the creative in a message, impacts conversions.
“You can’t do discount optimization unless you can see what someone’s buying and at what price points,” says Persio CEO Marc Grabowski. “When you give a one-size-fits-all discount, you’re needlessly giving margin to some customers.”
IT reseller DSS Corporation adopted cross-channel attribution when it evolved past siloed, focused campaigns and adopted a content-driven inbound marketing model. Attribution insights drive better understanding of those results and have fostered tighter links between marketing and the rest of the organization.
“It used to be that we knew we wanted to sell more of a particular product, so we would do a campaign just on that,” says Kristy Slimmer, senior marketing specialist at DSS. “Attribution helps us align our marketing with the buyer’s journey; 2015 was our benchmarking year, and we were able to provide cross-channel insights at our 2016 sales kickoff, which wouldn’t have been possible had we not made that shift.”
Also expect a bigger role in planning and strategy for in-house data modelers and scientists. They’re vital contributors to attribution efforts, both to ensure that clean data is being properly matched for attribution and that models and formulas proposed by vendors and partners are valid.
“Ten or 20 years ago, the role of the data scientist was much like an old mainframe developer — doing things nobody understood that, hopefully, resulted in some business value,” says Emad Georgy, CTO of Experian Marketing Services. “The use of attribution makes that role sexier and more relevant.”
Avoiding attribution roadblocks
Addressing the technological, data-matching hurdles of cross-channel attribution is necessary but not sufficient. Eliminating cultural roadblocks is even more important, particularly because attribution threatens the credibility of siloed practitioners who have been able to define their own metrics for success.
“You have an attribution platform say that search is being measured incorrectly,” says Scott Denne, research analyst at 451 Research. “The usual response to that is to say that the attribution vendor and its numbers are wrong. It takes culture within a marketing organization to get past that.”
Understanding the constraints of an organization’s ability to effectively shift budget, particularly in the short term, is also important to managing expectations. Long-term media commitments, as well as a limited buying power or inventory in certain channels, mean that attribution must be grounded in real-world possibilities.
“You need the ability to set constraints, to find the optimal mix of how to spend money when you can’t move radio budget to TV, or TV money to digital,” Visual IQ’s Muller says.
Offline exposures are traditionally held up as major blockers to cross-channel attribution. Make no mistake: It’s more difficult. Forrester Research data found that 90% or more of marketers include online display and paid search in their attribution model, but only 42% include mass media and a mere 26% incorporate direct mail or catalogs. Near-field communication and beacons make certain physical and outdoor media more likely to be trackable in the near future.
“Proximity-detecting technology has the most promise to be woven into an attribution mix, but print is still the longest shot,” says AdTheorent CEO Anthony Iacovone.
Aggressive brands aren’t waiting for others to solve the offline attribution challenge. MongoDB uses manual tracking to allocate the buzz from disruptive but non-trackable media, such as product launch briefings and third-party mentions, to account for traffic spikes outside a marketing campaign. Cabela’s uses marketing mix modeling to allocate the influence of mass media to conversions and sales. And receipt-scanning technologies are easy to deploy in customer-friendly apps, making it easier to close the loop of marketing activity with a confirmed sale.
The path to attribution is not cheap. Annual subscriptions to point solutions in this space can run tens of thousands of dollars, and some of the third-party data reports that feed them cost tens of thousands more. Notably, when Forrester polled cross-channel-attribution customers for its report The State of Cross-Channel Attribution Technologies 2015, 69% of respondents were from companies with more than 1,000 employees.
Marketers for small businesses, such as regional quick-serve chain Hwy 55, have still made strides by committing to detailed measurement of individual campaigns.
By isolating other strategic factors, the chain learned that a concerted push to quadruple the size of its customer loyalty program over 18 months produced a 9.5% same-store sales increase.
Working with loyalty platform provider FiveStars, Hwy 55 is now eyeing specific cross-channel attribution between paid search and the mobile app, email, and SMS components of its loyalty program.
“There’s still a lot of ground-level work to be done making sure our staff is all on the same page,” says Andy Moore, Hwy 55 director of communications. “But we want to make sure that we’re accounting for the quality of our offers, as well as how many impressions they get.”
The volume of data, campaign characteristics, and potential outcomes of cross-channel attribution analysis can make the problem seem unwieldy even to the largest marketing department. Letting go of the need to map every model against every valid prospect nationwide frees marketers to start challenging long-held expectations and biases about the strongest indicators of success and take the first steps toward comprehensive cross-channel attribution.
“You don’t have to start attribution with a huge campaign. You can start with a small test to segments of just 25 or 50 people and see what kind of response you get,” says Malinda Wilkinson, CMO of marketing automation vendor Salesfusion. “Don’t let the fear of trying new things out stop you.”