Google Analytics How To MODEL Around Not Provided
Here is an easy step by step model for how to model your keywords around Google's Not Provided. Working on +MoondashAudiodotcom reminded me how to do this:

1. Remaining - Subtract the rest of your keyword revenue from your massive Not Provided total. 

2. % Remaining - Divide your "Remaining" or non Not Provided keys into the Remaining total. This gives you the % for each non Not Provided key of the total revenue associated with non Not Provided keys. 

3. Allocate - Now multiple your Not Provided total by your % Remaining. 

4. Finally ADD your new Allocations by non Not Provided keys to their Google reported income (A + C). 

Now you have used a reasonable method to allocate Not Provided. Is it foolproof? No, but since most websites have a sizable 80/20 rule anyway (80% of the money is being generated by 20% or less of the products) the chances your can screw something up from modeling like this is tiny. 

The VALUE of allocating Not Provided is it creates the RIGHT PRIORITY for different keys. Here is another tip. Once you have your keys modeled stop thinking of them as distinct from one another. Think of your keywords as dancing with each other. 

If, as the chart below shares, you have 3 keys at $40 and $20 it helps to think of the $20 keys as half your $40 or your $40 as 2x your $20. I like to create averages and then compare to those averages as a way of getting in touch with how one key exists in relation to other keys and the entire ecosystem. 

I like to break out clusters and produce graphs since a metrics picture is truly worth a million words. Good luck and hope this little trick helps you know what value your keywords are really bringing to your #marketing  efforts. 

BTW, you can do a similar exercise for branded terms too. If your website's name is winning the attribution race model it out like this and reallocate so you can see the value at the line item level. Sure you are making a modeled assumption, but chances are a very good one. Branded terms, thanks to the mysteries of attribution, can be as tricking to understanding WHAT matters and HOW MUCH as Not Provided. 

Good luck. Marty 
Shared publiclyView activity