Attribution is a serious challenge for marketing. It always has been and will continue to be as cookies go away and privacy regulations grow stricter. There is an answer to the attribution problem, and it's called media mix modeling. Matt Hertig, the CEO, CTO, and co-founder of ChannelMix, explained how it works and showed me a new feature in the ChannelMix platform to ensure you stay on track.
The end of attribution
Marketers struggle to understand which channels drive the most leads, the highest quality leads, and the most conversions. That's what attribution is supposed to tell us. And there are multiple attribution models to work with - first-touch, last-touch, linear, multi-touch, and more.
Attribution isn't as easy as saying that for every dollar you spend on a specific channel, x number of leads can be obtained. That's because it's rare that a person sees an ad in one channel and then automatically converts. And it's even more rare that they've only seen your brand in a single channel. There is no straight line to conversion. As Hertig explained:
The reality is that that line doesn't exist.
People move across channels and use multiple devices, so it's impossible to see the entire journey from start to finish. It's, at best, an educated guess for someone good at attribution modeling. And now that cookies are being deprecated and people are taking their data privacy seriously, that educated guess is becoming nearly impossible.
A solution that relies on first-party data and the ability to understand the past and use it to predict the future is required. That’s what ChannelMix does.
Enter market impact modeling
ChannelMix doesn't care which channel directly leads to a lead or conversion. Instead, it looks at what combination of channels supports a desired pipeline.
It lines up all your first-party data and builds a pipeline, showing you how much you spent across all your channels in a defined period and how many customers you gained in that period, giving you an overall cost per acquisition.
It’s a planning tool. If you want to know how much to spend on each channel in a future time frame based on a defined goal (e.g., leads, website leads, impressions, cost, customers), it will use data from the previous period or previous year and using AI and machine learning start crunching the numbers. If you pick a goal that is not attainable, it will tell you.
What it then provides is a set of predictions that you can review and choose from, including top, great, and good predictions, as well as maximize target and not recommended predictions. If you select a prediction, you can see the breakdown it recommends for channel spend (“a recipe on how to spend money”). Pick the prediction you like and then start activating on it.
Track your plan against the actual effort
Because ChannelMix integrates all your first-party data to create a single source of truth for your marketing spend (traditional and digital marketing), it can track the results of your efforts and show you how you are doing. This is a new feature of the platform. It's called Pacing and gives you a real-time view of how well you are executing against the plan you selected.
Pacing will show the breakdown of your channels and how much you have spent against your plan for the period. It's continually updated as more data comes in, so you can easily see where you need to increase or decrease your budget and where you are on target.
As you implement your plan, ChannelMix shows performance, but it also updates the model to accurately reflect where you are at any point in time and how to get back on track. And that may change as the model updates. Hertig explained that machine learning is constantly learning and making predictions based on backtested scenarios.
The Pacing capability means marketers don’t have to track and report on actual performance in each channel they work with separately. They don’t have to use multiple tools and try to piece everything together, something that often happens after the budget is spent. It allows marketers to be more proactive and adjust budgets as needed to get the results they want.
The other feature ChannelMix offers is a confidence score. Marketers are not data scientists, so talking about R-squared and P-value often goes over their heads. The confidence score tells them how confident the machine is in its predictions. And it includes what Hertig called the default metric to measure marketing - marginal return on ad spend.
Every channel has a role to play. It doesn't matter which channel brings the lead in ultimately, because every channel has an impact - whether it's simply building awareness, or was one of many steps in the journey. This is key for the ChannelMix approach because awareness plays an even more significant role in supporting the buyer journey, especially with generative AI pushing less traffic to actual websites. Hertig said that even multi-channel marketing can only measure the conversion - it can't measure awareness generated.
ChannelMix ticks the boxes for how marketers need to plan and track marketing performance. I didn't go through all the features of this platform; instead, I hit the points around predicting the best channel mix. The new Pacing feature packages things up nicely and gives marketers one analytics tool for everything they need to do to ensure they are doing what they can to meet their goals.