How Artificial Intelligence Could Change Attribution Tools
Source – emarketer.com
The marketing technology landscape is busy, but there isn’t a solution out there for everything. Multitouch attribution solutions that help marketers map their customers’ path to purchase, for example, are still scarce. Julie Lyle, chief revenue officer at acquisition marketing platform DemandJump, spoke with eMarketer’s Maria Minsker about how artificial intelligence (AI) could improve attribution solutions and enable marketers to paint a fuller picture of the customer journey.
eMarketer: What’s missing from the marketing technology landscape today?
Julie Lyle: There are over 5,000 different marketing technology tools in the marketplace today, but more than 90% of them are pointed at retention, cross-selling, upselling and loyalty—extracting the greatest value from a customer or a prospect once they’ve been identified as such. What’s missing are solutions that are pointed at acquisition, namely solutions that present data that helps marketers make better investment decisions about their marketing spend.
eMarketer: What do these acquisition solutions look like?
Lyle: Companies need acquisition solutions that can look at their brand alongside their competitors and map out an ecosystem that shows what drives interest and engagement. It’s important to map any source that drives traffic to the brand’s or their competitor’s site—that way, it’s possible to map the flow of traffic two or three steps out, and determine when and why customers opt for a competitor.
Think of air traffic control. If an airline wants to fill every seat on a flight from Orlando to New York, it’s important to not only look at potential passengers in Orlando, but also everyone in hubs like Atlanta or Dallas who might fly through Orlando to get to New York. To truly understand their customers, marketers always have to take a couple of steps back.
eMarketer: What role does AI play in customer acquisition?
Lyle: To acquire customers, marketers have to understand attribution, and artificial intelligence can fill a lot of the gaps in existing attribution models. AI can unwind the path to purchase, and pinpoint when a customer comes to a brand’s site and leaves without converting.
Maybe they looked at a particular blog post about a product, searched for that product, found it, but still didn’t convert. Why? Because they went on their mobile phone on the way home, found a coupon on an affiliate site and then converted. AI can sequence that path to purchase and explain all that.
eMarketer: What are some of the limitations of existing attribution tools?
Lyle: In most cases today, marketers are using last-click attribution to define how they allocate their spend. Some are using first-click attribution, and some are using linear regression models, which assign a weighted proportion of an arbitrary value across different touchpoints in the path to purchase. But a lot of those estimates are based on just that—estimates.
In reality, brands are only able to see about 20% of their entire ecosystem. Artificial intelligence can unwind the remaining 80%, including second- and third-party nodes, which marketers just can’t see any other way.
eMarketer: What other areas of marketing technology will artificial intelligence touch?
Lyle: That’s the million-dollar question. Artificial intelligence sounds sexy, but it’s not going to solve every problem. My advice for marketers is before you start researching tools, define success. What are you trying to achieve and what would solve your business challenges? Then go look for the AI that can solve that.
My second piece of advice is that you are not one algorithm away from greatness. There is no algorithm that will solve every business problem. Algorithms are only capable of learning based on the quality of data that they have access to, so make sure you’re putting the right data in, or you’re going to get garbage out. That mantra has not changed.