Conversion labels: report per keyword

How to generate conversion reports per label per keyword

[Dlocks]

I have two conversion types:

1. signup
2. purchase/sale

For a service where customers can upgrade their free account I would like to see per keyword the amount of conversions per conversion type/label (and of course other metrics like clicks, costs etc).

Example:

Keyword_____| Signup conversions | purchases/sales
widget________|_________50_________|_______10
red widget_____|_________30_________|________6

Is this possible?

[rwilson]

You could do it in analytics by setting up goals and a custom report.

Create a goal for your url that has your conversion tracking on it set the goal type as "URL Destination".

Name your goal ex)signup.

Then create a custom report with "goal completions" as your metrics
and campaign and keyword as your dimensions.

[Dlocks]

That seems to me like a lot of work because using this method the data is not visible within the campaign/adgroup itself. So you have to look at different places at the same time. In other words: looking up a good converting keyword in Analytics, then find that keyword in AdWords adgroup and then make adjustments.

I was thinking about looking at 1-per-click conversions versus many-per-click conversions.

If I have 50 signup conversions and 10 purchases then the 1-per-click conversions should show 50 and the many-per-click conversions should show 60. Then I could conclude that 60 - 50 = 10 purchases.

There is however one big problem with the above method. My experience is that there is relatively a large percentage of visitors that will refresh the page containing the conversion code or hit the back button to go back to the thank you page at some point. So the many-per-click conversions is always incorrect.

A solution to the above problem is perhaps tracking (for example via a cookie) if the signup conversion code has already been served to a visitor. If so don't embed the signup conversion code again. Same for the purchase conversion code (only embed when visitor has not yet 'seen' the purchase conversion code).

Should that solve the problem of incorrect many-per-click conversion statistics?