Back to Blog

Understanding Attribution in Paid Digital Media


Attribution. It’s the gift and curse of digital advertising. The oft quoted marketing quandary, “I know half my marketing is working, I just don’t know which half,” is unfortunately still a very real problem. How is it possible that in 2019 most companies haven’t answered this for themselves? It all comes back to attribution and the systemic issue of running marketing campaigns in an increasingly fragmented environment of multiple devices, advertising platforms, walled gardens and the ambiguous online to offline impact.

Let’s start by defining the limitations to determine what’s even possible.

Tracking offline marketing

No good solution exists to date to track this accurately. Which I believe is a main reason traditional media is still purchased and used so extensively today. It has a legacy that is difficult to get around for advertisers fearful of going all in on digital. It can’t be proven that it doesn’t work, and it’s been used by every major brand in the past, so advertisers feel it’s supposed to be part of a marketing strategy.

There are solutions, like media mix modeling, which apply some semblance of logic to a very fuzzy statistical model of estimates and averages, which is just as likely to give the wrong information as it is to provide the right data. They are also very expensive and so full of caveats and assumptions versus measurable data that they can provide a dangerous false sense of security when evaluating your marketing performance. To me, lack of attribution cuts both ways, I can’t point a finger and say it definitely isn’t working, but I also can’t directly say it is working and therefore find it difficult to justify more funding.

To be clear I’m not advocating cutting traditional out of the equation. I think it’s an important channel for reaching an audience at scale in a way that people are comfortable with. But the skeptic in me prioritizes the ability to know, for certain, that media is working. Hence I lean towards a solution that lays digital media as the foundation and then looks to layer on traditional advertising to support the strategy.

Tracking online marketing

Inversely, there are tons of solutions to track online, both to the benefit and detriment of online marketing channels. How can tracking possibly be a negative? When it’s evaluating the wrong metrics associated to a particular targeting tactic.

The best example of this is display advertising. Display advertising is all of those banners popping up all over the internet, following you around, showing you ads for things that may or may not be relevant to you at that moment in time. As advertisers we can see the data, ALL the data and that means having metrics tied to every single ad that’s ever served.

The problem occurs when we take a targeting tactic like “audience prospecting” through display banners, which is an upper funnel tactic designed to drive awareness and brand recognition, and evaluate its performance based on how many conversions it drove.

I can tell you right now the answer’s a very, very small number. Uneducated advertisers look at that and think, this isn’t working, I’m wasting my budget, abort, abort! But in reality, the intention of that targeting tactic was not to drive conversions at all. So looking at that metric to evaluate if it worked or not is incorrect. “But it’s trackable, if it was really a good ad/impression they would have clicked the banner and converted,” advertisers might say. I say wrong. That’s just not the way users interact with that particular medium.

A great analogy I heard was tied to Out of Home (OOH) billboards along the freeway. Expecting a majority of users to click on banners and convert is the same folly as expecting people driving down the fast lane of the freeway to see a billboard, swerve across four lanes of traffic to exit and visit a brick and mortar location. It just does not happen and that’s okay.

Now if we focus on those targeting tactics that are meant to drive conversions (brand search, retargeting, CRM lists, etc.) there are still problems in truly knowing what’s working and it can be frustrating for marketers who feel helpless that they are spending their limited resources effectively. A huge consideration is the attribution model which is the methodology used to give credit to a marketing solution.

It’s very likely that a converting user was exposed to several of your ads, across multiple days and from various targeting tactics you implemented. Problem is your CRM only has one field for where a lead source came from. What do you do? Unfortunately the industry has adopted this same line of thinking and therefore typically defaults to last click/touch attribution. That means that the last thing that a user did before becoming a customer is what gets credit.

Let’s walk through an example. Say a parent-to-be is looking for a crib and that’s the business you’re in. You’ve got a Display and Search strategy and a user just converted through brand search, congrats. But did you realize there were ten touch points before that last branded search? Too bad it’s the brand search that gets credit and so you only see attribution tied solely to that ad. If we follow this logic, and cut off all marketing but brand search we’ll save so much money and have a ton of efficient conversions, right? Probably not. The numbers will look good for a few days, maybe even weeks or months if we are lucky, but there is a big ugly truth coming after that.

Remember those ten touch points before the brand search? Well they actually came from display and non-brand search as the user was searching for solutions to their problem. They were looking up articles and doing Google searches to try and figure out who they should go with for highly-rated baby cribs. This sent off intent signals to display partners that served your banner and in Google your ad showed up and they even clicked! But they were just researching…still had more research to do. Most advertisers would see this as a click, cost accrued, no conversion, what a waste. But this was a crucial step. You are now in the consideration set, congratulations!

So what happens when they finished their research and decided to go with you? Wait for a banner to be served? Do another non-brand search? No, they just went to Google and searched your brand name, the same brand name they had never heard of (remember, this is their first kid). Imagine this same scenario playing out months later when you were only running brand search. You wouldn’t be in the consideration set at all. You couldn’t possibly show up in that user’s brand search because they didn’t know about your brand to search for you. Eventually your brand search volume will die a slow death. Don’t let that be you.

Stay tuned. In Part 2 we’ll explore business performance vs advertising and my recommended solution for true attribution.