Most of the underperforming Google Ads accounts we inherit actually don’t have a bidding problem, or an ad copy problem.
They have a Google Ads campaign structure problem.
Every account we audit that has been carved into a dozen tidy little campaigns is quietly telling Google’s algorithm to learn twelve times slower than it could.
Back in the day, account structures were more thorough and hierarchical, nowadays your campaign structure isn’t an organizational chart.
Your campaign structure needs to feed Google’s algorithm. Do NOT spread it too thin.
Google Ads has spent the last decade moving decision-making from you to the machine. Which auctions to enter, which search terms to match, which placement to serve, what to bid at 11:42pm on a Tuesday. A lot of those are algorithmic decisions now.
You don’t make them. You inform Google.
The main thing that informs them is conversion data.
So when we look at an account, the first structural question we ask isn’t “is this organized?” It’s “how much conversion data does each campaign actually receive?”
Because a campaign is the unit at which Google’s bidding models learn. Conversions pooled inside one campaign compound into a signal.
The same conversions scattered across a handful of campaigns become a handful of weak signals. Spread too thin and none of which reach the volume the model needs to get confident.
Fragmentation is one of most damaging thing we see people do to their own accounts.
If your data is splitting across multiple campaigns and you feel like google’s algo isn’t learning, a second pair of eyes will spot it fast.
Get Your Account Structure Reviewed
Say your account generates 60 conversions a week.
Now say you’ve split the account into six campaigns.
One per service line, or one per product category, or one per city you serve, because that’s how your business is organized internally.
Each campaign now receives roughly ten conversions a week.
Ten conversions a week is not enough for Maximize Conversions or Target CPA to do anything intelligent. The model spends its lifecycle in learning mode, over-reacting to noise, never accumulating enough signal to distinguish a good click from a lucky one.
You’ll watch CPA swing wildly week to week and conclude that smart bidding “doesn’t work for your business.”
Consolidate those same six campaigns into two, and each one gets 30 conversions a week.
This way the AI algo can actually test which search terms convert, which audiences convert, which placements convert.
And it gets those answers faster, because it’s running its experiments against a denser dataset.
This is also why we’re so insistent about getting conversion tracking right before touching anything else.
Consolidation increases the volume of signal flowing to the algorithm. If that signal is wrong — firing on the landing page instead of the thank-you page, double-counting, optimizing for an add-to-cart when you care about purchases — consolidation just delivers bad information more efficiently.
What good structure actually looks like
Run as few campaigns as your account can get away with, and make every split you do keep earn its place.
Each of your campaign should have a reason to exist. Otherwise simplify it. Or use manual CPC instead of target CPA/Max Conversions.
Most of the campaigns we delete during a restructure have no job. They exist because someone wanted a tidy row in a reporting spreadsheet.
When a separate campaign is genuinely justified
In our experience, the list is short. A campaign earns its own existence when it needs:
- A different budget you must genuinely control. Not “prefer to see separately” but really need to control. Brand-only campaign is an example.
- A different bid strategy or target. A tROAS of 400% and a tROAS of 150% cannot coexist in one campaign.
- A different geography with genuinely different economics. Serving a city where your close rate is half what it is elsewhere is a real reason. Serving two adjacent suburbs is not.
- A structurally different campaign type. Search, Shopping, and Performance Max aren’t interchangeable.
Notice what isn’t on that list: reporting convenience. You can segment a report by ad group, keyword, product, or asset group without paying the algorithmic tax of a separate campaign.
Reporting granularity and campaign granularity are not the same thing.
Don’t use campaign structure to make your reporting easier.
Duplicate intent makes you bid against yourself
The other cost of over-segmentation is one advertisers rarely trace back to its source: you start competing in your own auctions.
When two campaigns are eligible for the same search term — and they will be, because match types are looser than they’ve ever been and exact match no longer means exact — Google picks one to enter.
You don’t get two shots at the auction. You get one, chosen by ad rank, while the losing campaign contributes nothing but a diluted data trail.
Worse, the two campaigns are now learning from partial, overlapping slices of the same intent. Neither builds a clean model.
And because you’ve raised competitive pressure inside your own account, you frequently pay more per click for the privilege.
Duplicate intent is the tell we look for first. If you can’t explain, in one sentence, why a search term belongs in Campaign A rather than Campaign B, the two campaigns should probably be one.
We’ll map exactly where your campaigns overlap and show you what a consolidated structure would look like.
What we’ve seen with our clients
The pattern is consistent enough across our portfolio that we now treat structure as a first-pass audit item rather than an optimization we get to later.
Local service accounts arrive the most fragmented. A dental practice will typically come to us with separate campaigns for every little thing.
We often have accounts come to us with overlapping campaigns that were cannibalizing each other, textbook duplicate intent.
We re-organized the campaign structure and corrected the bidding so the budget could actually be deployed, and the practice went from zero new patients a month to twenty and then to fifty-four new patients in a month.
The practice never changed. The account simply stopped competing with itself.
Ecommerce accounts fragment along product categories rather than service lines.
Lead-gen accounts split across multiple limited by budgets, because conversions are scarce to begin with.
A telepsychiatry clinic doesn’t have the volume to feed six campaigns. It has the volume to feed one or two extremely well.
To be clear, structure is never the only lever.
Keyword research, ad copy, landing pages and conversion tracking all did real work in the results we’ve published for a small dental clinic that added 80+ new patients in five months on a limited budget, a fashion brand that went from 0 to 6.4X return on ad spend, and a telepsychiatry clinic now generating 100+ new patients every month.
But structure is the lever that determines whether any of the others get the chance to compound. You need to give google enough data so google can find the pattern in them.
Consolidating without losing control
The objection we hear most is that consolidation means surrendering control. It doesn’t. It means moving control to a different layer.
Negative keywords still shape which searches you’re eligible for. Ad group structure still governs message match between query and ad copy.
Audience exclusions, device modifiers, location targeting, ad schedules and brand controls all survive consolidation intact.
What you give up is the ability to set six different bid targets… and if you were being honest, you weren’t setting from data anyway, because none of those campaigns had enough data to justify a target.
What you gain is a bidding model that reaches statistical confidence in weeks instead of never.
If you want the deeper treatment of how targets and strategies interact once your data is pooled, we’ve written that up separately in how we approach Performance Max optimization.
Common questions about Google Ads campaign structure
How many campaigns should my Google Ads account have?
Rather than targeting a number, work backwards from conversion volume: aim for enough weekly conversions per campaign that your bid strategy can exit learning and stay out.
If a campaign can’t clear a meaningful weekly conversion count on its own, it shouldn’t be its own campaign.
Does this apply to Performance Max as well?
Yes, and arguably more so. PMax is even hungrier for conversion data than standard Search, because it’s optimizing across more surfaces simultaneously.
Running several thin PMax campaigns is one of the most reliable ways we’ve seen to make PMax look like it doesn’t work.
How long before a restructure shows results?
Plan for a genuine re-learning period rather than an instant lift. You’re deliberately resetting the bidding model’s dataset, and it needs to re-accumulate signal.
That’s precisely why we don’t recommend restructuring an account every quarter. Just do it once, correctly, and then leave the structure alone.
Is there a case for a many-campaign account?
At high volume, yes. An account generating thousands of conversions a month can support far more segmentation, because each campaign still clears the volume threshold.
The rule isn’t “fewer campaigns always.” It’s “never more campaigns than your conversion volume can feed.”
Fix the structure before you optimize anything else
If your account is fragmented, nothing downstream will work as well as it should. Better ad copy, sharper landing pages, and smarter bid targets all sit on top of a bidding model. If your bidding algo doesn’t have enough data to learn, no amount of surface-level optimization compensates for a starved algorithm.
Restructuring is unglamorous work. It’s also, in our experience, the single highest-leverage change available to most accounts, and one of the few that costs nothing but the courage to delete campaigns you were proud of.
If you’d like us to look at how your account is structured and where the conversion data is leaking, that’s exactly what we do.
Let’s find out whether structure is the bottleneck holding your account back.
