The first thing you typically do when executing a keyword analysis is deciding what level of intent you’re after. In almost all cases, we want to target the highest intent possible in order to achieve the highest conversion rates possible. This makes sense, since it delivers value for the client, and it therefore should (almost always) be the first thing to initiate in all search strategies.
What we miss out on, however, is to inspire future demand with our clients’ products and services today. On many occasions, we use social channels to trigger interest among current non-customers, but awareness initiatives can also be promoted through search campaigns.
One example is X Shore, a company that manufactures and sells 100% electric boats. Initially, we started off with medium and high intent keywords. Later, we also tapped into the wider electric vehicle market, in order to inspire future demand with X Shore’s products. So, when an electric car enthusiast at a later point decides to purchase their first electric boat, X Shore is the first brand they think of!
It goes without saying that automized campaigning is the future, and we should, as long as the prerequisites allow us, try to utilize machine learning as much as possible. The most important prerequisite is conversion volume. Without it, the algorithm will struggle with its learnings.
In many cases, we need to take action in order to feed back more conversion data to the platform. Some examples that will help with the attribution are:
Note that Smart Bidding uses data from across your account to predict performance, and is not limited to data from a single keyword or campaign. This means that previous recommendations for conversion volume per campaign is no longer applicable. There is, however, still a minimum recommendation of 50 conversions per campaign for Target ROAS bid strategies.
As the Google ads system has become increasingly better at determining the relevance between a user’s search term and the advertiser’s keywords, we have begun moving from SKAG to intent-based ad grouping. The theory is, that in an auction, the systems take the meaning of all keywords in an ad group into consideration before matching a keyword to the specific search term. This should, in the end, produce a higher relevance, and therefore higher performance.
Grouping several keywords together, however, can have a negative impact on quality scores due to poorer ad relevance. So, as ad rank still is considered a primary factor that dictates performance, SKAG can still play an important role in campaign structuring.
As data has become even more important and match types have developed a broader nature, it is hard to argument for campaign splitting based on match types.
If we were to see vast differences in performance between match types, however, it could make sense to separate the match types into individual campaigns in order to achieve control of budget distribution between the two.
Broad match is the only match type that uses all of the signals (when used in automized campaigns) available to understand the intent of both the query and your keyword, while having the flexibility to find the most relevant match that is expected to perform for you. In, other words, this can be a great way of using machine learning for scaling.
Naturally, the system while look for a wide variation of search terms to match to your keywords, so it can be a hit or miss experiment, especially the first couple of weeks. It’s therefore strongly recommended to have stable campaigns with exact and/or phrase match keywords before expanding with broad match.