Skip to main content

How will ad campaign measurement change as privacy regulation makes attribution much more difficult? As an ad measurement enthusiast and someone who eats and sleeps cookieless, this question is near and dear to me. But it’s also complex — so let’s use a metaphor to make it simple:

You are tasked with growing a vegetable garden. At the end of the season you might judge success based on efficiency and total production.

  • Number of vegetables per dollar spent

  • Total pounds of vegetables

These metrics are similar to the performance metrics that we see in much of modern advertising. Performance-based measurement — the notion of understanding which ads drove actual changes in consumer behavior — is a key director of ad spend for most advertisers today. In fact over 84% of digital advertisers in the US are using digital attribution models in 2021. The implementation may be as simple as last-touch Cost-Per-Action, or as complex as Multi-Touch Attribution. The objective is the same though: to gain a sense of what different portions of our “garden” contribute to our harvest. If we are producing a high volume of vegetables — or lots of conversions per campaign — then we are running an effective garden.

The good news for performance buffs is that today’s advertising landscape is still half-populated with the key ingredient for driving digital attribution models: passively captured advertising IDs. This includes cookies, mobile IDs, and IP addresses, which allow brands to “close the loop” on a good portion of consumers. We are still living in the tail end of the golden era of ad measurement, so it makes good sense to get as much value out of these legacy systems as possible before it’s too late.

Now put your garden gloves back on and take a look at your metrics. What if your harvest isn’t as bountiful as you were expecting? What if your cost per veggie has you wondering why you didn’t just go to the farmers market? If you have only been tracking results you now have the “what” but not the “why.”

To find metrics that explain the “why”, we should look further upstream in the life of the plants. We could have had the best seeds, the best facilities, and a sophisticated real-time veggie measurement system in place. However, if we weren’t tracking soil conditions, irrigation, or pest control we may have a failed garden with no explanation as to what to do differently next time.

In advertising, this pursuit of upstream metrics that determine downstream impact has at times led to false positives in the past: take video completion rate or display click-through-rate as examples of this. But in the near future where attribution becomes more difficult, upstream metrics are worth revisiting. As Gartner puts it, “Cookie obsolescence will compound existing challenges of digital ad measurement, including transparency and interoperability standards and attribution accuracy, while making others irrelevant”. Sounds intimidating! — and like we may need a backup plan.

As such, we’d be right to prioritize owning a reliable, accessible baseline of core metrics for advertising before we get too discouraged with growing challenges on the attribution front. Was the ad experience good for the consumer? Did we serve too many ads to places where our audience doesn’t live? Did we avoid clutter, or stuff our ad pods full of the same creatives back to back to back? Fundamental questions such as these can help determine “why” a campaign worked, and answering them with clarity will lead to optimization opportunities that can directly improve your media buy.


Today’s advertising landscape — in which a cookieless and cookie-enabled internet coexist — is the perfect time to revisit your measurement framework and reset benchmarks around upstream KPIs that can be acted on during media campaigns, with the goal of building confidence in how we steward buys as we proceed into the fully cookieless world. Consider spending more time here and you may find yourself with a richer harvest down the road.

Sticking to vegetables,


%d bloggers like this: