Authors: Stephen Holland, SEM Manager & Julia Kung, Director of Paid Acquisition

Hint: Probably not to seed

It’s nearly impossible to identify all of the different keywords people will use to search for a particular thing. There are some keywords that most people use in their searches, and these are typically called the “head terms.” They make up the bulk of search queries, but they are also the most competitive, and therefore the most expensive, to advertise on.

The rest of the keywords people use are incredibly varied, less competitive, and generally less expensive to advertise on. A good SEM strategy will aim to discover and test bidding on these “tail end” keywords. One way to discover these keywords is called the seed approach.

The seed approach relies on broad matched keywords to discover search queries relevant to the keyword. This initial set of broad matched keywords is called a “seed list” of keywords. Queries matching to those keywords are then combed through to discover negative and positive keywords. Negatives are added to the account to prevent matching to those queries.

One of these Nikes is more relevant . . .

Sources: nike.com & wikipedia.org

Positives are added to the account as ad groups with only exact match queries. This means the search has to exactly match the keyword listed. If the search query was “Nike running shoes” the keyword would only match this phrase, not a variation such as “Are Nike running shoes amazing?” If the query didn’t match exactly to a keyword in the inventory, the search would default to the general broad match ad group. This limits the reach of relevant ad copy. Someone searching around Nike related terms sees the same ad as someone searching “Nike running shoes” or “Nike women’s shoes” and anything else not covered with exact keywords.

Traditional matching

The more traditional approach continues to cover everything with broad match ad groups but includes phrase and modified broad match types within ad groups. This covers subtle variations like “Nike running shoes” versus “Nike running shoe.” Negatives are still discovered and applied. The main benefit is the control around serving the most relevant ad, in this case a Nike running shoes ad, to a wider swath of query variations. This also creates a more manageable account, as phrase and modified broad cover more queries, as opposed to constantly building out exact match keywords. That became very cumbersome as we expanded not only the keyword list but also the geo targeted campaign versions.

Here’s another way of putting it:

Our search query is: Nike running shoes

The seed approach matches the ad we set up for Nike running shoes if we have the exact match keyword, [Nike running shoes], that is set up, otherwise a generic ad is served.

The traditional approach matches a small business ad if we have an exact keyword, the phrase matches, or the keywords are included as modified broad matches. Variations could include Nike running sneakers, blue Nike running shoes, Are Nike running shoes good? etc.

Generally speaking, it’s clear that the traditional approach is more flexible, more practical to use in ongoing account maintenance, and encourages account growth. So when does it make sense to use the seed approach?

A seed structure can be useful on non-brand terms with very little variance in the search queries. For brand campaigns it’s usually important to display ads for all queries about the brand (barring negative queries of course). In this situation the vast majority of searches are simply “Nike” but the running shoes-centric long tail variations are the ones we are most interested in. Yes, we could capture all of these in exact if we spent most of our time creating the exact match keywords. But the more important thing is to capture those combinations of keywords and phrases with the most appropriate ad. Building out ad groups that cover these queries with relevant copy is more important than adding countless variations of exact match keywords that could be caught with phrase and modified broad, especially since the brand CPCs are usually lower than non-brand CPCs. At the end of the day we are still catching anything else with the broad match ad groups.

Seed campaigns work well for non-branded terms without much variance in search queries, but that isn’t the most common scenario. At ROI DNA, we come across many accounts set up with a seed structure that aren’t performing to their full potential- the necessary maintenance that seed structures require ends up hindering results! If you want to switch your account over from the seed approach to a traditional matching approach, we suggest switching it out over the course of a few weeks, so that you don’t end up negatively impacting your quality score too drastically.