Open almost any underperforming Google Ads account, go into the search campaign, and click through to the ad groups. There is one. Inside it are forty-something keywords covering everything the business sells, and two ads that have to somehow speak to all of them.
Nothing about that is broken in a way Google will warn you about. The campaign runs. It spends. It even converts occasionally. But that single decision quietly removes most of the control you have over the account, and almost every other problem downstream traces back to it.
We find this in roughly two thirds of the accounts we audit, and in one recent audit of a rehabilitation equipment manufacturer it was the same defect in all four campaigns.
What You Actually Gave Up
An ad group is not a folder. It is the unit where three separate things get decided together, and putting everything in one means all three get decided badly.
Your bids. You bid at the keyword level, but you manage at the ad group level. With one group holding forty keywords, you have one lever for forty different things worth wildly different amounts.
Your ad copy. Ads live inside the ad group and serve for every keyword in it. One ad group means one set of ads answering every query, which forces the copy to be generic enough to cover all of them.
Your visibility into what is happening. Reporting rolls up to the ad group. When there is only one, every report is an average of everything, and averages hide exactly the things you need to see.
That third one is subtle and it bites hard. On one steel manufacturer's account we had to move custom columns down to the keyword level to see anything useful, purely because a single ad group made every ad-group-level number meaningless.
The Test: Do They All Mean the Same Thing?
The rule we use is simple enough to apply in your head.
Every keyword in an ad group should mean the same thing. Not related. Not adjacent. The same thing, phrased differently.
"Chiropractor near me," "chiropractor in Irving," and "local chiropractor" mean the same thing. They belong together. One ad answers all three honestly.
"Chiropractor" and "sports chiropractor" do not. We found exactly that pairing bundled into one ad group on a chiropractic account, and the problem is not tidiness. Someone searching "chiropractor" has back pain and wants relief. Someone searching "sports chiropractor" is an athlete with a performance question. Different person, different worry, different reason to pick you.
One ad cannot speak to both. It ends up saying something bland enough to be true for either, which means it is compelling to neither.
Build It From Your Website, Not From the Keyword Tool
The most reliable way to derive structure is to look at what you actually sell.
Take an antiques dealer selling quilts and jewelry. The keyword tool will hand you hundreds of terms in one undifferentiated list. The website already has the answer.
Campaign: Quilts. Ad groups for crib quilts, doll quilts, Amish quilts, wool quilts, 19th century quilts. Someone searching "Amish quilt" wants a specific thing, and an ad that says "Amish Quilts, Authenticated, Free Shipping" beats "Antique Quilts For Sale" every time.
Campaign: Jewelry. Ad groups for bracelets, earrings, rings, necklaces. Same logic. The ad copy lives in the ad group, so the ad group boundary is where copy stops being generic.
For a service business it works the same way. Your services page is your ad group list. If you offer prenatal care, sports injury treatment, and auto accident recovery, that is three ad groups minimum, because those are three different people with three different problems.
This is also why the structure survives. It is built on what your business actually does, not on how a tool happened to cluster words on the day you exported them.
How Many of Each
Two numbers, and they hold up across most accounts we build.
Five to eight keywords per ad group. Sometimes ten. If a group has grown past that, it usually contains two ideas rather than one, and splitting it is easy. If it holds thirty, it is not an ad group, it is a filing cabinet.
Ten to fifteen ad groups in a well-built search campaign. For a wellness client we mapped out around fifteen. That number is not a target, it is what falls out of doing the exercise honestly: every distinct thing you sell, to a distinct person, with a distinct reason to buy.
There is a limit in the other direction. One keyword per ad group was a fashion for a while and it creates enormous administrative overhead for very little gain, and it starves each group of the volume it needs to learn anything. Group by meaning, not by string.
A practical note on match types, since the two decisions interact. Tight ad groups matter more, not less, when your keywords are loose. A broad match keyword can pull in a wide spread of queries, and if it sits in a group with thirty others there is no ad in there written for any of what it drags in. Structure is what stops loose matching turning into incoherent ads.
And if you genuinely sell one thing to one kind of person, you may only need three or four ad groups. The number falls out of the business. It is not a score to beat.
The Bid Control Is the Whole Point
Here is what tight ad groups actually buy you, and it is the part that gets lost when people treat structure as housekeeping.
Some of what you sell is worth far more than the rest. A chiropractic practice may find prenatal patients return for months while a one-off adjustment does not. A manufacturer may have one product line at triple the margin of the others.
When those live in separate ad groups, you can bid them separately. You push hard on the valuable ones, holding the top positions on those specific searches, and let the marginal ones sit lower. That is the entire mechanism by which a small budget beats a bigger one.
When they live in one ad group, you cannot. You have a single bid across everything, so you are either overpaying for your worst work or underbidding on your best, and usually both at once.
This is also why we launch new accounts on manual CPC with exact match. Not because manual bidding is better in the long run, but because in the early phase the objective is page position on a small number of searches we already believe matter, and that requires being able to bid them individually. That phase and what follows it is covered in chapter two.
What It Does to Your Quality Score
Ad relevance is one of the three components Google scores, and it is the most directly controllable of them. It measures how well your ad matches the query.
An ad written for forty keywords cannot closely match any of them. An ad written for five keywords that all mean the same thing matches all five. That is the entire mechanism, and it is why tightly themed groups tend to lift quality score quickly rather than gradually.
You can see it in the account without doing any math. Add the Ad Relevance column to your keyword view and read it. Google grades every keyword as Above average, Average, or Below average, and a single sprawling ad group tends to produce a column of Below average with no obvious cause, because the cause is not the keyword. It is that the ad was written for thirty-nine other keywords as well.
Split the group and those grades move, usually within a couple of weeks, because nothing about the keyword changed. What changed is that the ad now answers it.
What that saves you in actual cost per click, and how the ad rank arithmetic works, is chapter nine. For now it is enough to know that the structure decision and the price you pay per click are the same decision.
Rebuilding Without Blowing Up the Account
If you are looking at one ad group and forty keywords, you do not have to start over.
1. Export your search terms report first. What people actually typed tells you which themes exist in reality rather than in theory. Sort by cost.
2. Group the keywords by meaning on paper. Put every keyword into a bucket where everything in it means the same thing. You will usually find between eight and fifteen buckets, and a handful of terms that belong to nothing, which is its own useful finding.
3. Build the new ad groups alongside the old one. Create them in the same campaign, paused, with their keywords and their own ads.
4. Write ads per group before you turn anything on. This is the step that makes the whole exercise pay, and skipping it means you did the filing and not the work.
5. Enable the new groups and pause the old one. Not delete. Pause. If something unexpected happens you want the history.
6. Expect a short dip. New ad groups have no history, and the account needs a couple of weeks to settle. Judge it after a month, not after a week.
The Twenty Minute Version
- Open your search campaign and count the ad groups. If there is one, that is the finding.
- Count the keywords in it. Anything past ten is holding more than one idea.
- Read your ads and ask honestly whether they answer every keyword in the group.
- Group your keywords by meaning on paper. Same thing, phrased differently, goes together.
- Aim for five to eight keywords per group and ten to fifteen groups.
- Check whether your ad group structure matches what your website actually sells. If it doesn't, the website is right.
Structure is unglamorous and it is the thing most accounts are missing. It is also the cheapest fix available, because it costs nothing but an afternoon and it changes what every other lever in Google Ads is capable of doing.
We rebuild account structure all day long and would happily take it off your hands. If you would rather do it yourself, do the paper exercise first. The buckets are usually obvious once you stop looking at the account and start looking at your own services page.
