Most search advertisers are familiar with this organizational structure:
Account -> Campaign -> Ad Group -> Keyword
That’s the way paid search was logically set years ago and it made complete sense at the time. Targeting/budgeting at the campaign level with keywords grouped with creative at the ad group level is an easy way to work within the platforms. Even though the industry has matured, that structure has virtually remained unchanged.
Maybe it’s time to revisit this architecture and see if there’s room for improvement?
On some accounts I work on, this structure is counterproductive to our managing, bidding, and reporting needs. It actually holds us back from being able to quickly gain insight on various ways we need to analyze the data for optimization. So, as a workaround, we create an offline keyword table to build in our desired functionality. In this table, each keyword in the account has multiple columns of metadata, that when combined with our campaign data, lets us pivot through the information in a much more intuitive way. In essence, this helps us transcend the constraints of the campaign/ad groups structure and see the data the way we want to see it.
This lets us look at our campaigns through different lenses, such as “branded terms,” “general terms,” “product terms,” “promotional ad group,” “testing campaign,” “awareness” vs “direct response” and so on. We can even track each keyword in terms of manufacturer, profit point, color, size, geolocation term, etc. The business questions you can answer quickly are very valuable. “How are my HP keywords doing versus my Dell ones?” or “Are my local/city specific terms doing better than my DMA ones?”
Campaigns and ad groups are logical software segments, not logical business segments.
Organizing your campaigns in this way is especially helpful for tracking keywords from different product categories. Sure, you could split your “keyboards” and “mice” into separate campaigns or ad groups, but maybe computer accessories don’t necessarily deserve their own group due to their lack of importance on your computer retail site. All you need is one “computer accessories” ad group in terms of keeping your account at a manageable size. However, later on, by adding in this data for a pivot table, you can split the keywords out if you want to look at them separately.
Here’s an example of what this table looks like:
All you do is download your keyword level campaign data, perform a vlookup (instructions follow this article) and you’re good to go.
Richard Stokes, President/Founder of AdGooRoo, has some great advice in his book, Mastering Search Advertising: How the Top 3% of Search Advertisers Dominate Google AdWords One of his top tactics is to split keywords into “buy,” “inform,” or “browse” categories to target users with ad copy based on their intent (i.e. where they are in the buying cycle). That’s a very important thing to do to reach users in the most relevant way. It also puts into perspective the importance of each term based on your business objectives at the time. For example, if you’re in the high season of your business, you want to ensure your budget focused solely on buy and browse terms. Or, if you notice you’re not getting enough new unique visitors, you may shift to inform terms to prime your sales pipeline for more future business.
Sometimes it’s just not feasible to actually create enough campaigns or ad groups to handle all of these pairings. What if you have fifty product lines that need their own campaigns because of the geotargeting required? Splitting them all three ways now creates one hundred and fifty ad groups! Plus, now your branded terms are all over the place in each ad group and now you have zero visibility into those keywords. As well, how do you manage your promotions when they affect every keyword in your campaign?
From a technology perspective, Omniture SearchCenter does allow you to create “virtual groups” by tagging keywords, ad groups, and campaigns. This is very useful for monitoring new ad groups or watching your top spending terms from multiple campaigns. However, I’d like to see a tool that allows to add in multiple metadata fields as each level so it would be easy just to pull all of your “awareness terms” or see how your “branded terms” are doing without having to add them all into one campaign.
I’m not sure where this could go, but maybe a paid search platform could be developed taking a concept from Gmail—no folders, just labels. You just have a list of keywords which you could apply multiple ad creative and destination URLs at the keyword or label level and then be able to sort and search through swiftly. In fact, all of the settings such as match type, geotargeting, dayparting, etc could be handled at the keyword (or label) level. You could even quickly define related words for content targeting. Quality score is done at the keyword level so it shouldn’t be affected much.
What format limitations are you running into with your current paid search accounts?
Using Excel’s Vlookup function to organize your campaign
The Vlookup (vertical lookup) is a handy Excel feature that allows you to compare two tables and pull data from either. The requirement is that you have at least one column of shared data so that the two tables can have a point of reference. In my example above, I have one table with keywords and their metadata. In another, I have the keyword level campaign data from my paid search account. I want to “marry” these two tables so that I have the keyword metadata alongside my campaign data.
By using Vlookup, you can perform this task by sending the metadata of matching keywords to the campaign table. Since the keyword is the unique pivot point, Excel knows that these two tables can be linked. I’m not going to go through each step, I just wanted to expose this feature to those of you who haven’t used it yet. Here’s the official Microsoft Office article that walks you through the Vlookup function. Check it out and add it to your arsenal…it can be very handy.
Here’s a great YouTube clip that will make you an instant Vlookup champion:
Opinions expressed in the article are those of the guest author and not necessarily Search Engine Land.