- Attribution is the process of valuing each touchpoint along the journey to a purchase or conversion.
- Start with the data: organize online sources, track offline behavior and merge into a single view.
- No model works for everyone — select an approach that best aligns to your marketing mix and funnel.
- Dig into the data. You must personally understand what’s going on before leveraging it with others.
- It will never be perfect. Start simple, iterate, and expand from there.
“Half the money I spend on advertising is wasted; the trouble is I don’t know which half.”
It was true in the nineteenth century when marketer John Wanamaker allegedly coined the phrase, and — unfortunately — is still true for many brands today. While our shift online has provided marketers with substantially more data, reconciling online and offline shopping behaviour can make tracking painstakingly complex.
A recent Salesforce study revealed that today’s average consumer uses ten different channels to communicate with a company. The ability to track omnichannel behavior has major implications on how companies should invest in growth — but effective marketing attribution can be difficult.
With this challenge in mind, it was my pleasure to host a discussion with two leaders in this field: Michael Kuznetsov, Head of Growth at Redesign Health, and Ron Jacobson, Co-Founder and CEO of Rockerbox. Together we shared best practices and pitfalls from our first-hand experiences as marketing leaders. Our key takeaways are summarized below.
Fundamentals: What is Marketing Attribution?
Marketing attribution is the process of valuing each touchpoint that leads to a customer conversion — whether that be a sign-up, a download, or a purchase. It helps you understand where your growth is coming from, so you can focus resources on the channels and campaigns that perform best. Without effective attribution, marketers are flying blind.
“Attribution is the process of valuing each touchpoint that leads to a customer purchase or conversion”
Unique tracking codes like UTMs have made it much easier to identify online conversions. In fact, an entire cohort of marketers have grown up successfully scaling brands on channels like Google and Facebook. But as these channels become increasingly saturated, incorporating spend from offline channels including podcasts, billboards, and television becomes more important. Tracking non-digital channels requires a different approach outside of built-in platform tools (like Google Analytics) and an effective attribution model to tie everything together.
Start with the Data
Creating an attribution system might seem like a daunting task, but here’s where to begin:
- Organize your online data. Map out your various sources for digital data (e.g. Google, Facebook, direct traffic, etc.), and leverage first-party identifiers like email to track cross-device behaviour.
- Track your offline data. Add post-purchase “how did you hear about us” surveys directly into your sales funnel and/or leverage custom promo codes (#subway20) and URLs (/podcast) to monitor the performance of offline channels.
- Bring it all together. Centralize all your data into a single view with a visualization tool like Tableau or Looker, where you can easily customize dashboards and generate reports. Make sure to determine your main conversion objective: are you optimizing for an email, an app download, an online purchase?
For more information on combining online and offline channels, explore Michael’s deep-dive on this topic.
Selecting the Right Attribution Model
Next, decide what model works best for your business. An attribution model is the set of rules that assign credit to touchpoints leading to a purchase or conversion. Rockerbox has a great article that breaks this down, and here are the basics:
- First touch models give 100% credit to the first marketing touchpoint (e.g. A customer first interacts with you by clicking your Facebook ad).
- Last touch models give 100% weight to the last marketing interaction (e.g. The customer was aware of your brand for months, but clicked on a paid search ad right before they purchased).
- Multi-touch models enable you to distribute weight among different touchpoints throughout the journey.
Many brands use a “1-day view, 28-day click” attribution window — meaning they will track actions that occur within 1 day of someone seeing your ad, or within 28 days of them clicking it. More considered products like financial services may use longer windows. There’s no right or wrong approach, but as a general rule of thumb: the more complicated your marketing mix and sales funnel, the more sophisticated your attribution model needs to be.
“The more complicated your marketing mix and sales funnel, the more sophisticated your attribution should be.”
Practical Advice on Marketing Attribution
Here are a few lessons we’ve learned from years of experience and many mistakes made:
Dig into the data. It’s critical that you personally take the time to understand the underlying logic of your attribution model. It can be tempting to outsource this thinking to a teammate or third-party. But without a solid understanding of how your model works, you will continue to stab in the dark when it comes to tracking and channel performance.
Secure buy-in. When building out your model, solicit input from the team who will use the data. With department heads and respective channel-owners weighing in, there can be politics involved with changing systems of tracking and evaluation. Key stakeholders must understand how the model works and agree that it is a fair, accurate, and single source of truth for marketing KPIs.
Consider the context. If you don’t consider reality, data can be deceptive. While marketing tracking continues to improve, it is never completely accurate. Even using lauded tools like post-purchase surveys — which often have response rates of +85% — you can receive false positives. There are periodic examples of customers who mention they’ve heard of a brand from a certain channel — but the brand has never advertised on that channel. It’s important to take a step back and think about the numbers on the screen.
It won’t be perfect. If you’re feeling a little overwhelmed, now is a good time to take a deep breath and realize you’re not alone. Marketing attribution isn’t a perfect science — and never will be. Your goal shouldn’t be to create the perfect model. Modeling is an iterative process, so start somewhere and refine as you go. At its core, attribution is about providing better data to make better decisions than you did before.
If you find yourself saying “less than half the money I spend on marketing is wasted,” you’re on the right track.