Keep it Simple: Why Last-Touch Attribution Is All [Most Of] You Really Need

Many smart, well-meaning marketers have spent a lot of time trying to figure out which specific touches across different channels drove consumers to make a purchase. For example, a buyer clicks on a banner ad and does nothing, then two weeks later searches on Google and follows a paid AdWords link to the site, and finally clicks on a link in a personalized email the next day and makes a purchase. All these marketing touches contribute to the conversion, but how important was each one? This is the problem of “multi-touch attribution” in campaign marketing.

Of course, marketers need to know which efforts and channels have performed best so they can gauge ROI and optimize their marketing spend. They invest in complicated attribution modeling systems that attempt to measure how much each touch actually contributes to the buyers’ decisions. But if you actually look at the data, these complicated modeling systems are unnecessary. The best approach to attribution is often the simplest: in most cases, this means simply crediting the last touch as the most important driver in consumer purchasing decisions.

What’s the Big Deal About Multi-Touch Attribution?

Multi-touch attribution can get very complicated very quickly, with measurements based on a combination of linear, time-decay or time-window approaches. These give different weight to different touches depending on which stage of the process they occurred in or how long it took before the buyer eventually made the purchase. This way, the banner ad that the buyer clicked two weeks ago gets credit along with the email campaign the buyer clicked on last before making the actual purchase.

This complex modeling may be important for purchases with long decision cycles, like buying a mortgage or car or health insurance. But for the vast majority of retailers, the purchase cycle is much shorter. Imagine you’re buying a shirt or a pair of headphones — most likely you do a little browsing and then make your decision pretty quickly (or you’ve decided to buy because something caught your eye or you got an enticing offer). With such short decision cycles, last-touch attribution provides more than enough accuracy. You aren’t visiting the website multiple times before you make your purchase — it’s just one AdWords link or product recommendation or email that pulls you across the line.

Data Shows Last-Touch Attribution Is Just as Effective

It seems obvious that a cost-benefit analysis would show that it’s not worth owning and operating a sophisticated attribution system for most retailers, but we decided to test this theory with one of our customers, an outdoor apparel company. We ran a sophisticated multi-touch attribution analysis that gave various weights to every one of the company’s marketing efforts to determine which drove the most revenue. We also ran a simple analysis in which all revenue from a given purchase was attributed to the last marketing touch before the actual purchase. When we compared the two, it looked like this:

The three most important revenue drivers were exactly the same in each analysis, and in the two instances where the results differed, the rank order was only one rung off. Furthermore, when we looked at revenue attribution in actual dollar values, the greatest difference amounted to less than 10% variation, and most channels showed up with almost the same exact same value.

All this goes to show that marketers probably need to spend less time worrying about how to measure attribution and more time optimizing conversion rates across every channel. Last-touch attribution gives you almost the exact same information as a complex multi-touch analysis, and as consumers’ decision cycles continue to get faster and faster, last-touch attribution will only become more accurate. Unless you’re selling cars, health insurance or mortgages, keep it simple and stick with last-touch attribution.

Omer Artun is the CEO of AgilOne. He holds a Ph.D. in Computational Neuroscience and Physics from Brown University where he studied under Nobel Laureate Physicist Leon Cooper on pattern recognition, data mining and complex systems modeling at the Institute for Brain and Neural Systems. Between 1999 and 2002 he was a consultant with McKinsey & Company, consulting high-tech and retail companies on strategy development. He specialized in analytical areas such as pricing, direct marketing, customer segmentation. Between 2002 and 2004, he was VP of Strategic Marketing at CDW/Micro Warehouse, a $6B direct reseller of technology products. Between 2004 and 2006, he was Sr. Director of B2B Marketing at Best Buy at its newly founded Best Buy For Business division. Omer was an Adjunct Professor of Marketing at NYU Stern School of Business, teaching graduate-level relationship marketing and analytical marketing courses.