Why You Should Segment Your Competitive and Brand Data

David Kennedy / 3rd May 2016 / Comment

SEM brand data

At Wheelhouse, one of the first adjustments we generally make when assuming management of a new SEM account is to ensure that brand and non-brand (competitive) keywords are properly segmented by campaign. From a reporting perspective, it is critical to evaluate brand and non-brand keywords separately. There are a few reasons for this – some of which may be more obvious than others.

To begin with, we believe we create the greatest value for clients by helping to increase their non-brand business. Ensuring separation is foundational to doing so – as is clear, in-depth reporting on our performance. But the value of brand/non-brand segmentation lies in much more than just reporting. Our approach combines highly skilled, hands-on PPC management with enterprise-grade bid optimization.

Understanding Your Platform’s Algorithm

When leveraging bid optimization, it can be a huge mistake to allow any algorithmic system to evaluate long tail keywords by aggregating data using both brand and non-brand data. This is because the two data sets almost always respond very differently from one another.

Take as an example a keyword that doesn’t have enough data to be bid on its own. An algorithm will generally model like keywords together based on common attributes. In some platforms, these attributes are determined through some level of human interaction while other platforms determine attributes and groupings wholly through algorithmic means. Either way, if you fail to ensure your platform is not excluding brand data for bidding non-brand keywords, you can run into the type of bidding error I describe (in simplified fashion) below.

Overvaluing Non-Brand Traffic

Brand Image

As the table shows, when using brand data, aggregated values go from a suggestion of $0.75 all the way to $1.59. Combining your brand and non-brand data in this manner could lead you down the path of overvaluing your non-brand traffic. You’ll end up paying much more than it’s really worth.

It’s important to note that the table above isn’t suggesting that the algorithm would just aggregate all keywords and give the same bid across the board. There will generally be a level of “feathering” and thus, the more data a keyword already has, the less it will require “help” from its like partners. Another issue the above shows is that you probably would want some levels of security that wouldn’t allow your brand bids to rise to $3.75, which you may very well want for your more competitive non-brand keywords, but that’s another article.

What is important here is to make sure you have a strong understanding of how your platform algorithmically works and what you must do to ensure you are not supplying that platform with data that could skew the way in which it is calculating the value of your long tail keywords. Separating your brand and non-brand keywords can dramatically help your platform bid the tail and drive the performance you expect.

By David Kennedy