Launching Aryma Delta

Measure Incrementality through Information !!

What if you could see Incrementality without running a single experiment?

What if you could detect whether your campaigns are truly moving the needle from the moment they launch, instead of waiting until they end?

Sounds impossible?

Well we just made it possible, after nearly 2 years of R&D of course :)

Introducing Information-Theoretic Incrementality Measurement

At Aryma Labs, we have long leveraged information theory for feature selection in our MMMs - replacing correlation based methods with information based measures like transfer entropy, copulas, mutual information, and KL divergence.

This shift alone led to incredible gains in model accuracy, robustness, and business relevance.

Encouraged by that success, we pushed the boundary further:

Can information theory also measure incrementality itself?

💡 The Core Idea

Unlike correlation, information theory measures how much “information” one variable truly holds about another. For example, how much media spend “tells us” about changes in sales.

By computing information gain (or loss) across time between campaign and non-campaign windows - we can qualitatively detect whether incremental effects are present, without ever running a formal experiment !!

It is a continuous, diagnostic lens on incrementality rather than a post-hoc verdict.

Qualitative ≠ Quantitative

Just to be clear, While Aryma Delta qualitatively measures and reports incrementality, for quantitative estimation we recommend running causal tests such as Difference-in-Differences (DiD), Interrupted Time Series Analysis (ITSA) or Geo hold out tests.

ArymaDelta - A novel approach to Incrementality

ArymaDelta signifies the next leap in our journey to blend information theory and experimentation to measure Incrementality.

ArymaDelta works best with MMM derived contribution numbers and sales. However it can also work well with just raw spends numbers.

Try ArymaDelta for Free now: https://arymadelta.arymalabs.com/accounts/login/

---

Thanks for reading.

For help with MMM, Causal Marketing Experiments and Experimentation, get in touch with us.