Macaroni impact

We have 14 clients with at least 20 actions implemented as of 1st May 2024.

'martins_chocolatier': 456, 'moderntribe': 319, 'erogenos': 216, 'ukmedi': 118, 'wwmake': 75, 'kokoso': 69, 'bigjigs_toys': 43, 'nuts_pick': 42, 'procoffee': 37, 'twelvesixtynine': 33, 'gilda_pearl': 33, 'sosimply': 29, 'mensuas': 22, 'she_selected': 20,

Narrowing down down the actions to just those 14 clients, we have a total of 1949 actions impacting 953 pages.

Method

In order to investigate the impact of these actions, we look at the deviation in a certain metric (impressions, clicks, revenue, etc.) from the “pre-action mean”.

That is, for each impacted page, we

1. Look at the metric for the 60-day period prior to the action being taken.
2. Select a list of “similar pages” from our set of non-actioned pages, those which have similar (±20%) values to the impacted page during this period.
3. Calculate the mean value during the 60-day pre-action period. This is our base-line for this page.
4. Calculate the deviation from this base-line over the next 60 days (post action).

Results

The chart below (Fig. 1) shows this analysis applied to the impressions metric. We can see that, in the post-action period, the deviation (blue line) increases steadily, up to over 250% (2.5) after 60 days. The “similar” non-actioned pages also see an increase, although less significant. It’s important to note that some of these pages may have been impacted by other improvements to the site. Many of those clients, including Martin’s Chocolatier, the site with the largest number of macaroni actions, are also Blink clients. Some non-actioned pages may also experience a knock-on effect from the site being generally more well-optimised.

Fig. 1. Mean change in Impressions over the pre-action mean.

In Fig. 2 the same analysis is applied to Revenue. This is generally more patchy than Impressions data, and less reliable since it is difficult to attribute revenue directly to a particular page. We are simply looking at the landing_page and associated purchase_revenue in GA4. The data in this case has been smoothed by taking the mean in a 7-day rolling window.

Fig. 2. Mean change in Purchase Revenue over the pre-action mean. Smoothed by taking the mean over a 7-day rolling window.

The revenue, on average, increases by a greater degree than impressions, up to a 400% increase on some days around 50 days post-action, and we note that the non-actioned pages also seem to experience a much larger increase (up to 200%).

It is important to note that these two charts give an overview by taking the mean over all clients and all actions. Our macaroniImpact app built in Retool shows the results of the same analysis on a per-client, per-action level and can provide more detailed insight.

https://blinkseo.retool.com/apps/b2e71d2e-0899-11ef-a3b9-f7897205c9a3/macaroniImpact

Summary

It is clear overall that the macaroni actions are positively impacting site performance, and that this positive impact is, on average, steadily increasing over time, even 60 days after the action is taken.