Just like auctions, account structure is often seen as one of the less sexy topics within paid search. This is a huge shame though, as how you structure your account is arguably one of the most fundamental and impactful decisions you can make while running paid search. But first:
A paid search account has various different levels of hierarchy:
We can summarise the relationships above as:
Note that one of the campaigns' ad groups have multiple keywords, while the other campaign's ad groups have only one. It's not common to see a mix of these approaches in the same campaign, which brings us to the question:
If you've followed the above, you may realise that there are a couple of areas in which you can be flexible in when designing a campaign. In particular:
This is the easiest question to answer, simply because it doesn't matter. Your campaign can have as few or as many ad groups if you like.
If you find yourself in a situation where you have 100s of campaigns with only a few ad groups each, it may be worth thinking about consolidating these into fewer campaigns. This doesn't have any performance benefits, it'll simply make your life easier from an admin perspective.
To even run a campaign, your ad groups need to have at least one ad inside of them. If you don't, you don't have anything to show on all the search result auctions that you win, so what's the point?
The questions then becomes a case of whether you should have more than one ad per ad group, and the answer to this is typically yes.
Having multiple ads in an ad group allows you to test ads against each other. Testing is typically run by creating two ads which are identical to one another except for one small feature. Isolating this one feature as the difference between the ads means that you can attribute differences in performance between the ads to that feature, a process called A/B testing. We'll look in more detail about how to run A/B tests later on, I just wanted to get us in a testing mindset.
There are reasons that you might want to put more than 2 ads in an ad group. One of these reasons is to run A/B/n test, which are A/B tests with more than 2 variants. For example, if you wanted to test 3 ad variants this would be called an A/B/C test, and require 3 ads per ad group.
Despite its seemingly simple wording, this is perhaps one of the most contentious topics in all of paid search. The debate over it breaks down into two sides:
The main argument for a STAG approach is that it makes building and management easier. Say that your campaign might has 100 keywords. Grouping them into 10 themed ad groups of 10 keywords each is arguably easier to manage than having 100 ad groups with 1 keyword each.
STAG stands in opposition to:
To understand the motivation behind SKAG, we have to remember that each keyword can only serve ads that are within the same ad group. Because there's only one keyword in a SKAG ad group, it creates a one-to-many relationship between keyword and ad group, as opposed to a many-to-many relationship as there exists within a STAG ad group.
Having a one-to-many relationship between keyword and ad group means you can tightly tailor the ad copy that gets displayed alongside each keyword, because you know those ads will show on one keyword and one keyword only.
Going back to our section on auctions, this is beneficial because it means that we can write ad copy that's highly relevant to that one keyword. This improves our ad relevance and quality scores, in turn giving us better performance in ad auctions.
Contrast this to STAG, where you have a many-to-many relationship between keywords and ads. This means that not all ads in your ad groups will be relevant to all of that ad group's ads. Some keywords will have lower ad relevance because of this, and so will incur higher costs for the same number of clicks.
This is essentially the argument for SKAG. The argument against SKAG is that while it's certainly nice to be able to tailor ad copy at a keyword level, it requires an unreasonable amount of resource and management, and isn't efficient at scale.
While I'd agree that SKAG campaigns do take more effort to build by hand, there are plenty of pieces of tech which can help with this. I'm currently working on one, called Ads Studio, which lets you build SKAG campaigns and tailored ad copy at scale.
Whichever approach you decide to take, do be conscious that your decision has implications. While I'd argue that SKAG is better for performance purposes, there are resource-driven reasons why you might want to use STAG.
After reading the above, you should have an understanding of the basic aspects of paid search account structure. You know what what campaigns, ad groups, keywords and ads are, and how they relate to one another at a primitive level. In the next few sections, we're going to drill into these topics in more detail. To do this, we're going to start at the most basic part of the paid search hierarchy, keywords.