Marketing Attribution: 5 Lessons from 5 Models

In the last 18 months, I’ve spent a good (veering towards unhealthy) amount of time thinking about marketing attribution. I’ve pondered important questions like “What’s accurate?” “What’s fair?” “What’s actionable?”

And one thing I’ve realized is: I’m not the only one struggling with this challenge. With the astounding amount of data now available, determining marketing performance is more complex than ever before. Today’s marketers are able to track every online and offline touchpoint—from web visits to asset downloads and booth visits—and then tease apart where they came from and which opportunities and accounts they tie to. But despite all of this data, we’re all still struggling to identify the best way to effectively spend our budget to optimize revenue.

Last year, the Marketing Operations team at Demandbase set out on a mission to answer that very question by overhauling our existing attribution model. 18 months and five attribution models later, we’re proud to say we’ve made some sort of progress.

While we probably haven’t gotten it completely right (if that’s even possible), I’d like to think we’ve learned some things along the way that could benefit our fellow marketers contemplating a similar overhaul…

Get your marketing team on board. It’s important to remember you’re about to change the way you measure your team’s performance. For program owners and managers alike, that’s a big deal. It impacts their perception in the company and potentially their pocketbooks. Don’t just assume you can unveil a completely new way of measuring performance without having them on board.

Compare models. Allay fears up front by being transparent about how channels are going to be impacted. In our case, channels like email and direct web traffic saw a significant bump in their likelihood to be attributed to opportunities. On the other hand, channels like events were less impacted. Change a virtual certainty, but you can avoid a lot of stress by helping your team understand how (and why!) their performance metrics look different.

Report like crazy. There is no such thing as too much information, especially during the first few months of a new model. If you’re lucky enough to be on a data-driven team, marketers will be eager to understand how to forecast with potentially different numbers. Arm them with as much information as you can.

Build a model that fits your business. If there’s one thing we’ve learned during this process, it’s that there are infinite ways to interpret your data. For a company hyper-focused on building out its database, focusing on the first touch might make sense. If you’re more interested in learning what knocks an account into a sales cycle, a last touch model can be effective. If optimizing spend across all channels is your goal, something like a full-funnel model works. The best way to frame up your answer really depends on the question you’re asking.

Keep the bigger picture in mind. Here’s the kicker: we’re back to a single-touch model. I’m almost ashamed to admit that, given how adamantly I believed we needed to switch to a multi-touch model. The thing is, our multi-touch model didn’t help us with our original goal: to understand how to optimize revenue. We needed to have a clear way to communicate which specific programs resulted in which specific opportunities, without splitting apart every opportunity into hundreds of pieces. So, while we still measure all touches across the funnel, we optimize (and set KPIs) around the last touch before opportunity creation. And, at least for now, we’re happy with that decision.

I have no doubt we’re only just scratching the surface when it comes to marketing attribution. The data will only get richer and our ability to process and take action on it will only improve. Here’s to the next five models!

Want to learn more about ABM attribution? Join Demandbase & Bizible on 8/30 at 10am PT for a live webinar, where we’ll discuss how to set short-term and long-term measurement goals for your ABM programs.