David Kennedy is VP of Digital Advertising at Wheelhouse DMG. David joins us from REI, where he led their Paid Search team. Prior to REI, David was a Vice President with RKG, where he built and led a digital advertising team of 30. He has a proven track record as a digital advertising strategist, having devised and executed successful campaigns for many top national brands. David is deeply experienced in both lead generation and e-commerce, and has particularly advanced skills in using bidding technology to maximize performance.
This guest post first appeared on the Wheelhouse DMG blog. If you attended our webinar with Facebook on cross-channel attribution, it offers further, valuable insights on the pros and cons of using various attribution models to measure your marketing success.
Digital marketing continues to expand, both in scale and opportunity. A decade ago, digital marketers were limited in the ways they could connect with customers, but now have a multitude of display opportunities: paid search (text and shopping), marketplaces with Amazon and eBay, video, and myriad social channels.
With the increase in digital investment and, more importantly, the diversity of channels where we are investing, it’s essential that we know which investments are actually generating revenue. And these insights, to a degree, can be found through attribution models.
A good attribution platform can really help digital marketers solve this problem. Data siloed in multiple platforms results in duplicate credit being given to multiple marketing efforts. However, attribution solutions only give credit based on set rules—and it’s up to us to make smart assumptions as to how credit should be given (first-click, last-click, etc.).
These models were born from our need to determine which traffic sources drive value before a conversion takes place—to be able to place a fractional percentage to each individual purchase. This is what we now have, but don’t be fooled. Industry wide, marketers have a concept in their mind that attribution is telling them whether their money is well spent. And to a degree this may be true, but it’s an incomplete picture.
A smart friend of mine, Philip Chiappini, data scientist at REI, reminded me a few years back: “Attribution only tries to measure fractionally how to give credit. It does not measure the incrementality of a marketing effort.”
I can’t stress this enough—and it’s an essential concept to confront when deciding how important an attribution platform is for your business. Attribution platforms cannot answer how much revenue you will lose if you turn off your remarketing efforts, or what lift in revenue you’ll see if you double your marketing budget. Only mixed market analysis or even more simple A/B, on/off tests can tease out those answers.
The primary purpose of an attribution tool is to avoid double counting of conversions and to algorithmically give fractional credit. Attribution models are measuring the value of each touch within a path before a conversion (or micro moment). Measuring that path’s conversion rate to a similar path, minus one touch point, gives you the fractional value of that touch point in the first path.
When compiling millions of different paths, an attribution algorithm can begin to place value on each marketing channel. Because this is how credit is valued, systems tend to overvalue channels where touch paths are more prominent.
What marketers need are attribution systems that continue to do the heavy lifting through computer learning algorithms, but are also customizable enough that a smart person can differentiate a remarketing display ad from prospecting ads.
Until these systems can take into consideration the purpose of the ad and not just the ad type, attribution tools will fall short of delivering what we really need. A system that can not only provide fractional credit but also test for true incrementality would be a welcome, actionable addition to the digital marketer’s tool belt.
