Soren | 22 Dec 2007
The Yahoo Search Marketing Blog recently published another article on keyword insertion, one of my least favorite PPC strategies.
The reason I am down on keyword insertion is that it is a fundamentally lazy way of improving campaign performance. It was designed so that novice PPC managers could get better clickthrough rates without a lot of effort. The theory being that seeing the keywords emboldened in the ads will attract the attention of the searcher and thereby improve clickthrough rates.
So first, the disclaimer. If you have a campaign with large ad groups full of diverse search terms, it is possible, or even likely, that keyword insertion will improve your conversion rate. However, there is a limit to how much this can help.
The problem with keyword insertion is twofold. First, it can often result in fairly odd ads, for example:
What does this mean? Are they supporting charm sales or bicycle sales? Bicycle charm sales? If so, they don’t seem to be attempting to qualify a target audience and the ad copy doesn’t really work for any of these potential audience segments.
Second, it is tempting to create very large and diverse ad groups or avoid dealing with the issues that these types of ad groups can cause. The problem is that large and diverse ad groups don’t perform as well as small and cohesive ad groups in building quality score and in identifying diverging behavior in audience segments.
To use Yahoo’s example, they suggest combining the keywords “chardonnay” and “wine” to use in the same ad. However, if we create a user profile for these terms where the user attributes are:
“wine” - low wine education, searching for gift, cost-conscious
“chardonnay” - oenophile, searching for self, quality-driven, bulk-buyer
We can easily see that if we limit optimization to what works for both these audience segments, we may easily miss massive opportunity for increased clickthrough and conversion rates.
Keyword insertion can work, particularly where the ad groups are conceptually tight and the resulting ad text isn’t nonsensical, however, we would not consider it a good best practice in most situations.
