For small clients (as well as for clients whose multi-channel is difficult to analyze), I follow the net CPC (clicks, CTR, cost per click, refusals).
Task : to understand which pk works more efficiently and, based on this, edit the rates.
To do this, I use the cost per useful click (CUC) in analytics. This indicator takes into account the cost per click, and the bounce rate.
')
Formula : Cost / Clicks * ((100-BounseRate) / 100)
I will explain in simple language:
We received 200 clicks for 2000₽, the percentage of failures is 20%. So really useful clicks we bought 80pcs,
$ 2000/80 = $ 25
Also, this metric helps to analyze statistics in small samples, where conversion cannot be decided.
At the entrance we should already have a ready-made DataFrame with statistics from the advertising system.
Enter a new column in the statistics.
Python doesn’t do math in the same way as in mathematics, so we’ll do each action on a separate line:
#f['CUC'] = f['Cost']/f['Clicks']*((100-f['BounceRate'])/100) f['CUC'] = 100-f['BounceRate'] f['CUC'] = f['CUC']/100 f['CUC'] = f['Clicks']*f['CUC'] f['CUC'] = f['Cost']/f['CUC']
We get the following:
Looking at this indicator, we can see weak points in a few seconds.