Recommendations on Results: What is it, and how to use it to your advantage

This article summarizes the Recommendations feature's definition, and how to use it in the best way.

We've just released the Recommendations feature onto the measurement dashboard, allowing you to see the recommended action to take, based on the measurement results and the goal defined on the board.

These are simplistic recommendations, based on your target reach vs. the incremental (marginal) performance, and they are there to simplify the interpretation of results.

We've centralized for you all the definitions and best practices, for you to utilize when reviewing measurements you've run.

Technical Definitions

The recommendations are created automatically, based on the defined goal in the dashboard.

For instance, when creating a ROAS goal, the platform will utilize the defined benchmarks and will create the recommendations according to this goal, and the incrementality showing for your ROAS within each result.

The recommendation will appear in every measurement that covers the defined goal (i.e. the specific OS, country, and cohort for the goal, if defined).

How is The Recommendation Determined?

Let's break down the logic used in order to determine the recommendation:

As most goals are based on attribution data, please keep in mind that an incrementality-based result differ from attribution data, such that:

• It is relative to the spend level you are currently at, such that it provides information only to your spend change and not your full spend
• It is describing market contribution (rather than user-based attribution)

We advise to be more flexible with incrementality results.

For now, the sensitivity level we enabled for scaling up recommendations, is for the set goal to be met by at least 80%.

Let's take the following example, where a customer defined a ROAS goal of 24.5% for ROAS d7.

We're seeing in this measurement, that the marginal ROAS that was lost, when Facebook was reduced, is 18.8% - Close to the defined goal, yet slightly below it.

Your budget decision itself could differ based on your sensitivity to goal and based on the current level of spend (the higher your current spend level is, the more flexible you should be).

Best Practices

Below is a list of best practices to apply when utilizing the recommendations:

• Always account for at least 1 measurement per entity, to allow you to see different spend levels and their incrementality level. The more measurements you have, the more data points you can rely on when making your next move.
• As results might vary between different cohorts, and hence - The recommendation might vary, we suggest choosing a cohort you optimize spend based on, and review several measurements within the same cohort.
• Have you come acorss a recommendation that contradicts itself? Be sure to notice the following:
  • Measurements should be in the same granularity level, e.g. comparing campaign level measurements to other campaign measurements.
  • In case you're measuring a channel level change, try filtering to the campaign level, to understand which changes happened on what campaigns at the same date, country, and OS - Should the change be in 2 different campaigns, this might explain the difference in the results.
  • Make sure the country and OS are identical across all measurements

• • [In case relevant] Should you have a major promotion, or another external factor that could impact performance greatly within the same timeframe as the measurement, please add it to the board and rerun the measurement again, to make sure it accounts for the event's value as well.

If you have any further questions, please reach out to us via Support@incrmntal.com and share with us.