Multi-Touch Attribution Is Dead
Multi-Touch Attribution is dead. For those of you who haven’t even deployed your first multi-touch attribution model, skip past Go and consider yourself lucky- you just avoided an analytics adventure with a bad ending. Multi-Touch Attribution- or “MTA” for the uninitiated- is a method of attempting to measure the effectiveness of marketing campaigns by tracking users across devices and ads using their identities and clicks to determine which ads “caused” the customer to make a purchase. The methods in MTA were always subject to serious limitations and accuracy issues, but several recent (and looming) events are putting an end to MTA for good.
Let’s start with the premise that a company could actually track unwitting consumers with cookies across all of their devices and stitch together a digital picture of advertisements’ ability to drive a sale- and let’s set aside whether consumers even wanted to participate in this scheme even if they knew they were part of it (hint: most people don’t like this). The premise stated has serious flaws:
– How does the marketer know the sale wasn’t driven by TV, Radio and Print ads that were also running at the same time? You can’t drop a cookie on these ads (at least not yet).
– What happens when the customer goes into the store to make the purchase instead of ordering on their phone? Marketers- open up your wallets- you’ll need to buy a bunch of third-party consumer identity data to piece this puzzle together. Or worse, you can send your entire customer file to Facebook and have them tell you whether their ads were responsible for the sale. Gulp.
– Oh wait- what if the customer paid in cash?
– You might have been running a sale at the time the purchase was made. How does the marketer know if seasonal or day-of-week effects weren’t responsible for the sale? Black Friday, Cyber Monday, Q4 Holiday impacts matter? How about Saturdays when many people go out to shop? (Yes, this still happens…)
– Also, the impact of promotions is often a significant driver of sales and traffic. Not accounting for these effects will substantially overstate the value of certain marketing campaigns.
MTA promised to measure the mighty and mythical “path to purchase” and the magical sequence of advertisements that unlocked the treasure chest of consumer purchases. Never mind the inconvenient fact that over half of online purchases have two or fewer clicks prior1. And, if the cookies and 3rd party data get the consumer matching wrong even by a small percentage, the models and measures become highly inaccurate in a hurry. Think this doesn’t happen? It most certainly does (read more here:
Most of the issues above are well known to cross-channel marketers and analytics professionals. Despite this, many still pursue the tantalizing siren song of consumer tracking. There’s actually a better use for MTA but most conflate tracking a consumer with “marketing measurement”. Tracking via MTA can be a good method for consumer experience and personalization objectives. It just isn’t particularly good for measuring the contributions of an ad. Even the MTA vendors knew about these problems and added other techniques (marketing mix modeling) to overcome the deficiencies and accuracy issues. The problem is, they still use MTA to measure digital campaigns, despite calling themselves “Unified” measurement vendors. There isn’t anything “unified” about these approaches.
On top of this Apple has entered the ring, with Intelligent Tracking Prevention (“ITP”). ITP 2.1, which rolled out recently, puts major restrictions on consumer tracking on all Apple devices as well as any device using the Safari browser. ALL cookies are deleted after 7 days regardless of whether the cookie is 1st party. So, if you sell goods that have a pre-purchase consideration window of a week or longer, forget about using MTA to understand marketing impacts. It simply won’t work any longer. The latest data on ITP effects is substantial: 29% of all internet browsing in the US uses Safari, and over 47% of US smartphone users have an Apple device2.
MTA for marketing measurement is dead. It never worked very well anyway, so it’s OK to move on and seek better solutions.
Need a marketing measurement solution that isn’t MTA? There’s a better way: OptiMine. We can help you- no cookies, no compromises, no MTA.
1. Coleman, Alex (2016, Sep. 30). The Path to Purchase in the Context of Time [Web Log Post]. From https://www.wolfgangdigital.com
2. http://gs.statcounter.com April, 2019