I'm attempting to create a weighted ranking of content served in emails. I have done

some work already, but I'm not sure how to proceed. Thank you in advance for you help!! :tup:

I've anonymized the data here: https://docs.google.com/spreadsheets/d/1-iSQfqOVr9Dr1P_uIA5LH_gz6hlXzvXfVbefz-E_jLg/edit?usp=sharing

See the 'Stats' tab for what I've done already. All numbers and percentages are made up, but based on real data and a real sample size. Assume this is my data as I would apply the techniques that we would use here directly to my real data.

**Goal:**

Rank content by 'performance'. Our measures of performance are open rate, click rate, and the negative indicator of unsubscribe. However, I just have a very elementary understanding of stats.

My goal is to take my entire sample size here, rank the measures of performance listed above in a way that

**does not factor in the size of the sends**so as to not favor those with large send volumes.

I will then weight the ranks by importance of the measure - so we've decided to rank them with the following weight out of 100. Sorry if I'm not using the proper terminology here also, I'm way out of my element.

Open rate: 20

Click rate: 70

Unsubscribe: 10 (this will ultimate be a negative factor in the score, so we'll probably multiply by -1 while we're summing the scores)

So a 1.5% click rate is much stronger than a 1.5% open rate.

Once I have the ranking system implemented we'll only rank those content items that have only

**What I've done:**

This is me just struggling through and thinking of what might be helpful. By all means, this could all be useless, or maybe I've hit on what I need to do! I have no idea. That is why I'm posting.

So, what have I done. I've normalized the open rate and click rate. I've taken the average of those two numbers. I will then add the average of those two numbers with the unsubscribe number* to get the 'final score'.

**What I still need to do:**

1) most importantly - is that the right way to 'rank' this stuff?!

2) I need to weight the scores, which I am just a google search away from learning how and doing.

3) Be able to explain this in simple english to non-technology and statistician types.

*I don't have the unsubscribe rate numbers, but I assume I'll be able to find them, weight them, and apply the negative score in the end.