The importance of an effective measurement strategy cannot be understated. In his in-depth presentation, Google’s Alex Maksimov provides an overview of the lay of the land as well as tips for choosing your measurement strategy.
As consumers demand more transparency, choice, and control over how their online data is used, advertisers seek to provide ever more personalised messaging and targeting using customer data.
To handle these two conflicting demands, the industry has shifted. From the broadcast era and the precision era, we’re now moving into what’s being known as the predictive era; where marketers use sophisticated algorithms to enable them to do more with less. In this new landscape, how can brands find strong evidence to measure their marketing activity and make better decisions?
In a digital attribution model (DDA), credit is assigned to digital media touchpoints based on their involvement in conversions. This model looks at both converting and non-converting journeys to see the likelihood of specific touchpoints in increasing conversions. However, as third-party data becomes more limited, this model will likely become more probabilistic and we may need to rely more on first-party data instead. While it has a lot of strengths — including powering your bidding strategy, this model cannot show incremental impact, or tell you anything about the impact of something not in the system.
With DDA as the foundation of your measurement strategy, you can start to fill in the gaps. In econometrics, marketers seek a mathematical relationship between different variables both in and outside of your control and the outcome variable — in this case it’s usually sales. This gives you a high level view of how each variable is contributing to sales, but it does take a fair bit of historical data and time to set up.
As any budding scientist will know, you have to test your hypotheses to find out if they work. The same principle applies to your marketing efforts. By designing a test to check your hypothesis, you can show the incremental impact of a change in any metric, or prove the value of a specific channel. However, it can’t deal with many changes at once or give you real-time results.
The future of measurement will likely require multiple solutions to handle such a diverse set of insight requirements. You’ll need a mix of econometrics, DDA and controlled experiments to have a solution that works across all channels and takes external factors into consideration. Who knows, maybe it’ll do the creative one day too.