Marketing Attribution in the Data Tech Age

Share this post

Marketing attribution is easily compared to cooking – You wouldn’t dump everything in a pot without carefully measuring the amount and impact of each ingredient, right? So why are you doing that with your attribution tactics?

Measuring what we need and dishing it out appropriately is the right approach. Marketers must implement accurate attribution efforts to find the best multichannel solutions. Those that succeed will have a strategic advantage over competitors. It’s time to start bringing unbiased measurability to cross-channel campaigns.

CMOs now have the power to rely on person-level identifiers – and not just gut-instinct decisioning – to accurately quantify the impact of media mix and optimize marketing spend. As MarTech and AdTech converge and brands begin to embrace true people-based marketing, CMOs must implement the necessary resources to harness real-time data, segmentation and targeting, media execution and frequency management. All of these strategies lead to a better way of optimizing your media mix and maximizing ROI.

So, what’s required for successful attribution today?

  • Correctly assembling all of the relevant data
  • Running the actual attribution analysis
  • Making the analytics outputs actionable (in order to improve ROI)

A successful attribution strategy can be boiled down to a three-step process:

  1. Gather conversion data from all of your addressable channels for attribution and subsequent activation. That doesn’t mean taking it from the areas where you have the most, it means connecting online and offline data to your campaigns. If you can’t account for all sources of conversions, it is extremely difficult to do accurate attribution.
  2. Use high-quality identity resolution assets to attribute media spend at the person level across offline and online channels and across multiple devices. If possible, incorporate rich first-party data from your internal data lakes to better understand your consumer.
  3. Incorporate predictive modeling and statistical analysis for added value in the process – with the required balance of analytics professionals and technology in order to achieve the optimal outcome.

If you are currently using a model-based approach for your attribution strategy, you have a very good theoretical view of the ecosystem. Unfortunately, your interpretability is subjective, and your findings may not be directly actionable.

Approaches like click-through rate, last click and view-through are plagued with problems. For example:

  • They may amplify false correlations between engagement and purchase
  • They may undervalue channels
  • They are easily gamed
  • They lack industry standards

By using a data-driven, controlled-experimentation framework, marketers can remove bias and subjectivity and measure true campaign ROI. The use of person-level identifiers (PII) eliminates reliance on cookies and allows us to quantify true campaign impact not only in email but in display and social channels as well.

This may sound a bit intimidating, but it’s actually easy to achieve. Essentially, this approach starts and ends with an email address. If you can build your target universe using only email addresses, you can measure accurate results by splitting your targets into control (CTR) and treatment (TRT) groups. When the campaign is over you create a deterministic match between your customer records and the treatment and control groups using email addresses. The difference between the two match rates is the true impact of the campaign. This reliable and transparent approach using deterministic identifiers gives you the ability to measure campaign impact even for cookie-based and walled-garden ecosystems.

Let’s apply this approach to the real world. For this example, let’s say you are an advertiser running 4-5 channels trying to determine the optimal mix for your campaign. Here’s how you can apply deterministic measurement with causal attribution.

  1. Identify conversion or brand KPIs of interest.
  2. Isolate a target population of high-propensity prospects with known email identifiers.
  3. Split that target population into treatment (TRT) and control (CTR) groups.
  4. Assign TRT into equal budget, homogeneous groups with each group corresponding to one channel.
  5. Run channel-specific campaigns with similar creatives and messaging for a specific time period.
  6. Calculate lift in conversion and brand KPIs for each channel using incrementality framework.
  7. Calculate optimal channel mix by normalizing incrementality measurements for KPIs of interest.

Person-level Attribution in Action

Here’s a real-world case highlighting the value of person-level attribution:

We recently helped a leading ride-sharing company quantify the true impact of their campaigns using our deterministic measurement framework. They were successfully executing conventional CPM and CPL campaign at scale. Significant leads were being generated, but the advertiser’s media optimization team wanted to understand the true impact of their individual channel campaigns, as opposed to the probabilistic.

We devised a test and control experimentation framework like the one outlined above. We then built our TRT and CTR groups. Next, we ran our campaigns on the TRT group. At the end of the experimental campaign, we quantified the difference in conversion rates between TRT and CTR groups. It was 67%!

After quantifying this, the company was able to increase the size of its campaigns with us even more, by drawing spend from other channels.

Don’t these types of results increase your appetite for thinking beyond what might stick with your attribution strategy?

Share this post
No Comments Yet

Leave a Reply

Your email address will not be published.