Introducing Causal Attribution: An Evidence-Backed Method for Making Daily Spend Decisions

Attribution tools like in-platform reporting and multi-touch attribution are built on a simple premise: if you stitch together clicks and views, you’ll be able to tell if your media investment is paying off. 

The problem? Click-based attribution assigns credit for a sale by tracking consumer touchpoints – like clicks and views – that occur in close proximity to a sale (correlation) instead of tracking which tactics actually cause a sale (causation). 

Messy consumer journeys just don't lend themselves well to traditional correlation-based attribution. Consumers may see a digital ad that convinces them to buy, but end up buying in store. That purchase is invisible to attribution. Or consumers may have planned to buy your product regardless of every ad you run, and attribution erroneously gives your digital channels credit for sales your ads had nothing to do with.

And that’s just the beginning. Platforms are incentivized to over-report conversions, privacy regulations limit user tracking, and with upwards of 7-9 attribution models on the menu, marketers constantly have to justify their definition of success to the c-suite and investors (it’s tough). Given all of these factors, it’s no wonder why marketers – and their skeptical stakeholders – take even the best attribution models with a (sometimes massive) grain of salt. 

While marketing attribution data is biased and incomplete, it’s still one of the most critical signals marketers rely on to manage spend on a daily basis (and rightly so). With hundreds of thousands or millions of dollars on the line across hundreds of campaigns and ads, having a day-to-day pulse on paid media performance down to the ad level is non-negotiable.

So how can marketers get the daily signal they need to guide micro spend allocations, but mitigate the inherent bias in attribution?

Our answer: Casual Attribution

Causal Attribution debiases your attribution data systematically, down to the ad level

We built Causal Attribution, the first attribution tool of its kind, to give marketers a debiased view of daily media performance. And customers who use it are seeing a tangible impact to their media investments: 

‍

“We've seen an immediate improvement in top-line business performance that neither traditional MTA nor platform reporting alone was able to predict or capture.”

‍Greg Robinson, Head of Data @ GLD Shop

‍

Just like you’d expect from an MTA, Causal Attribution uses a pixel to collect raw attribution data. But unlike any other attribution approach, the inputs aren’t limited to trackable clicks and views. 

Instead, Causal Attribution uses causal data – incrementality experiments – to establish a baseline for marketing impact before it layers on clicks and views (new to incrementality testing? Start here). 

With experiment data, it’s actually possible to cut through the bias of correlation-based attribution. Thanks to a lot of complex math, Causal Attribution can adjust incrementality experiments for seasonality and spend level changes, then apply those incrementality factors dynamically to your raw attribution data.

For channels or campaigns where you don’t have an experiment to go on, we developed the privacy-safe Incrementality Index, a collection of thousands of incrementality experiments, to give you an experiment-backed performance estimate that’s grounded in incrementality.   

The result is incrementality-calibrated metrics like ROAS, CPA, AOV, orders and revenue down to the ad level. It’s the first view of daily performance that comes closer than any other tool at capturing the true impact of both online and offline marketing efforts to influence micro spend allocations. 

There’s another important upside to this attribution approach. Instead of constantly justifying the attribution model you chose – last-click, first-click, multi-touch or a dozen others – Causal Attribution gives marketers a single incrementality-calibrated attribution model they can rely on for decision-making, with first-click, last-click and linear models they can compare to if needed.  

We’re really only scratching the surface in explaining how it works. To double click on Causal Attribution, book a demo with our team so we can give you a detailed walk-through.

Building trust in Causal Attribution’s ability to drive better business outcomes

The best way to evaluate how Causal Attribution will impact your business is to actually use it. And we’ve made the barrier to adopt as low-friction as possible. Causal Attribution is included with Haus experiments by default so that you have the incrementality data you need to make evidence-backed decisions down to the ad level.

A few best practices to help you evaluate how well it’s working: 

• Compare Causal Attribution data against your existing MTA or our standard attribution models to see if it follows the general trend of your attribution data (it shouldn’t match, but the data should follow a similar pattern).
• Run an experiment to see if your MTA or Causal Attribution gives you numbers that more closely align to that experiment result. 
• Monitor business results over time as you use Causal Attribution to get a sense for how it’s impacting performance.
• Start with smaller spend decisions and work your way up to bigger ones as you build confidence.

Millions of dollars of investment, thousands of gnarly science problems and hundreds of conversations with marketing experts later, we’ve built a new approach to attribution that is consistently more reliable than MTA, and delivers better paid performance for the businesses that are using it. 

Causal Attribution works because it starts from a more reliable premise than any model MTAs have spent years polishing. We invite marketers to try it, poke holes, push us to make it even better, and see the results for themselves.

‍