To try and answer this question, I conducted a test to see where I could get the most bang for my buck. This is what I found:
Approach 1. Spread Thin
Spreading PPC budget thinly across adgroups casts a wide, but shallow net; you cover all bases, but end up "going dark" early on in the day. In addition, the quality of engagement from these visitors is lukewarm.
Approach 2. Go Deep
Focusing budget on the top performing adgroups produces a higher level of conversions, with visitors clustered more towards the buy phase. However, as the net is narrow, there are much fewer visitors being caught. In addition, the keywords belonging to top performing adgroups tend to have a lot of competition, and consequently have a much more expensive cost-per-click (CPC).
Looking at these two approaches, the road you take would probably depend upon the goal you are trying to achieve:
- If you want to drive traffic to your site, then the "Spread Thin" approach makes sense
- If you want to drive conversions to your site, then "Go Deep" delivers
Considering the Customer Journey
There are additional, important implications of taking a "Go Deep" approach that needs to be considered. In particular, the matter of assist attribution.
My eyes were first opened to the concept of assist attribution in a talk given by Jim Sterne at SES San Jose. The idea is that although general keywords used at the start of the customer journey have a low conversion rate, they can actually assist higher converting keywords further down the keyword funnel. Therefore, by removing assist keywords, you reduce conversion of your higher performing keywords later on.
Here's an illustration to demonstrate this point (adapted from an example given by Jim):