Hourly stats and Day-parting of Amazon ads

Analysis

The following analysis has been done solely on the US market since it’s the biggest marketplace where the feature is available and we wanted the data to be as relevant as possible.

First, we looked if there were significant changes across the different weekdays. We noticed nothing of the sort meaning we could focus our analysis on an average of all days.

In order to find room for improvement, we needed to find ACOS variations across the day and we found them !

After seeing these results, one thing’s for sure : there is something to do to improve our performance. Nevertheless, we should keep in mind that the AdSpend at the time the ACOS spikes occur is very low (between 0% and 5% depending on the account) which means even fewer sponsored sales.

‍

Day-parting feature

While we’ve just seen that there is a way to improve the ACOS by using the hourly stats, how should we design the new feature ?

The goal of the day-parting feature is to avoid spending when the actual ACOS is way above the average ACOS. The most reliable solution that comes to mind is to stop the Ads when this happens, and that’s what we’ve done.

Since some of our users would not like to use this feature fearing a loss of potential sales, the activation is manual on a strategy level.

Here is a concrete example :

The strategy “US Products brand 1” has an ACOS target of 14% but we can see a clear spike between 1 AM and 3 AM :

We can use the Day parting feature to stop the ads during the spike :

By doing so, we will save the 2,17% of the spend that has an ACOS 19 points higher than the target, but we will also miss 0,81% of the sponsored sales.

Dear users, we are pleased to deliver this much-requested feature. We hope this article was helpful and we wish you all a nice optimization!