Google Ads Account Structure in 2026: How Many Campaigns You Actually Need

Open a Google Ads account that has been running for three years and you can usually read its history like rock strata. There’s the layer from when someone read that every keyword deserved its own ad group. A layer of campaigns split by match type. A Performance Max campaign added in a hurry because a rep suggested it. Three campaigns named “Test” and one named “New Test (2)”. Nobody planned the whole thing. It accumulated.

Most people who search for help with account structure aren’t starting from scratch. They have a mess, or a nagging suspicion they have one, and they want to know how many campaigns they should actually be running and what belongs in each. The advice they find is often contradictory: one article says split everything out for control, another says consolidate everything so the algorithm has data. Both can be right, depending on what you’re splitting and why.

This post is an attempt to give you a way to decide, rather than a template to copy. We’ll go through what structure is actually for, what Google’s automated bidding does with the shape you give it, where splitting helps and where it quietly hurts, and how to audit an account you’ve inherited or built up over time. I’ll be clear about which claims come from Google’s own documentation and which come from agencies and tool vendors writing about it, because those are not the same thing, and the numbers floating around are often softer than they sound.

What account structure is actually for

A Google Ads account is a hierarchy: the account holds campaigns, campaigns hold ad groups (or asset groups in Performance Max), and ad groups hold keywords and ads. That much is in every beginner’s guide. What gets skipped is why the hierarchy matters, because each level controls something different.

• Campaigns control budget, bidding strategy, locations, languages, networks, schedules and, for the most part, the conversion goals being optimized.
• Ad groups control which keywords share which ads and, in practice, which landing page a search gets sent to.
• Keywords, audiences and assets control the individual matching and messaging decisions inside those containers.

So structure is really a set of decisions about where you want to be able to say different things. If two groups of searches need different budgets, they need different campaigns. If two groups need the same budget and bid target but a different message, they can share a campaign and live in different ad groups. If two keywords would happily share the same ad and the same page, there’s no structural reason to separate them at all.

Put that way, the question “how many campaigns should I have?” is the wrong one. The better question is “what do I genuinely need to control separately?” The count falls out of the answer.

There’s a second function that people forget: structure is also how you read your own results. If your account is one big campaign, you can’t tell whether the brand terms are carrying the numbers while the generic terms lose money. If it’s forty campaigns, you can’t read anything because every number is based on six clicks. A good structure gives you a level at which the data is both distinct enough to mean something and large enough to be believable.

The old advice and why it stopped fitting

For years the orthodoxy was granularity. The most extreme form was the single keyword ad group, usually shortened to SKAG: one keyword, often in three match types, each in its own ad group, with an ad written to echo that exact keyword. The logic made sense in the era of manual bidding and exact-match-only thinking. You wanted tight relevance between the search, the ad and the page, and you wanted to set a bid for each keyword individually.

Two things changed. Match types loosened, so an “exact” keyword no longer meant only that exact string. And bidding moved from something you set per keyword to something the system sets per auction, using signals about the person searching that you can’t see and couldn’t act on if you could.

When bids are set per auction, the reason for slicing your keywords into tiny buckets mostly disappears, and a real cost appears. Every bucket holds only a sliver of your conversions. Several of the current guides on this topic, from agencies and from tools that sell account management, repeat the same conclusion: Smart Bidding needs volume, more granularity means less data per campaign, and less data means slower and less stable learning. Google’s own account structure guidance, as summarized in those same guides, has moved toward themed ad groups and away from single keyword ad groups. I’d treat the exact wording with some caution since I’m reading it secondhand, but the direction isn’t controversial. Almost nobody credible argues for SKAGs any more.

The replacement that keeps showing up is the single-theme ad group: a handful of closely related keywords that you could all answer with the same ad and send to the same page. The numbers quoted for “a handful” vary by source. You’ll see 5 to 15 keywords, or 5 to 10, and 7 to 10 ad groups per campaign as a rough ceiling. Those are rules of thumb from practitioners, not limits from Google. They’re useful as a sanity check, not as a target to hit.

The test that does the work is simpler than any number. Take the keywords in an ad group and ask whether one ad could speak directly to every one of them and whether one landing page would be the right destination for every one of them. If yes, the group is fine regardless of size. If you find yourself writing “and also” in the ad to cover the stragglers, split it.

What Smart Bidding does with the structure you give it

It helps to be concrete about the mechanism, because the consolidation argument gets repeated as dogma and then applied in places where it doesn’t fit.

A bid strategy like Target CPA or Maximize Conversions learns from the conversions attributed to the campaign (or portfolio) it’s attached to. It builds a picture of what a converting click tends to look like, and then it bids higher or lower in each auction based on how closely the current searcher matches that picture. More conversions in the pool means a more reliable picture. Fewer means a noisier one, and noisier estimates produce wobblier bidding. We covered the data side of this in more detail in how much conversion data Google Ads actually needs before Smart Bidding works, and the short version is that the commonly quoted thresholds are softer than they sound.

The structural consequence is this. When you split one campaign into two, you don’t get two campaigns with the same quality of learning. You get two campaigns that each learn from roughly half the data. If your account produces 200 conversions a month, that’s an easy trade. If it produces 25, you’ve just made both halves weaker to gain a reporting view you could probably have gotten another way.

You’ll see specific conversion thresholds quoted in guides: fewer than 30 monthly conversions, run one campaign; 30 to 80, maybe two; aim for 30 to 50 per campaign as a floor. One source I read puts the floor for fragmentation at about 15 conversions a month per campaign. These are practitioner heuristics and they don’t agree with each other. Google doesn’t publish a hard minimum that flips a switch. What they’re all pointing at is a real gradient: below a few dozen conversions a month per bidding unit, you’re in a region where results get erratic and every change is hard to evaluate.

There’s also a budget side. Google’s help documentation for some campaign types gives a rule for how large the budget should be relative to your target cost per conversion. As I understand it, the guidance for Demand Gen is a daily budget of at least ten times your target CPA, and for Performance Max it’s at least three times. Those ratios differ by campaign type and I’m citing them from summaries of Google’s pages, so check the current help page for the type you’re running. The principle is worth keeping regardless: if your target CPA is $60 and your daily budget is $40, the campaign can’t even afford one conversion a day at target, and it will spend much of its life starved. Splitting a small budget across many campaigns makes that problem worse for every one of them.

If you’ve ever watched a campaign flip into a “Learning” status after an edit and stay there, this is part of the story. Campaigns with thin data take longer to come out of it. We dug into what actually triggers it and how to avoid resetting it in the piece on the Smart Bidding learning phase.

When you should split things anyway

None of this means one campaign is the goal. Consolidation has limits, and the guides that push it hardest tend to bury the exceptions. The cleaner way to state the principle is: split only when something genuinely has to be different. A reasonable list of reasons that hold up:

• Different budgets. If you want to cap spend on one product line, service or region without touching another, they need separate campaigns. A budget is a campaign-level setting and there’s no way around that.
• Different targets. A high-margin service and a low-margin one can justify very different cost per acquisition. If one bid strategy has to serve both, it will compromise and get neither right.
• Different geographies or languages. These are campaign-level settings too, and mixing them usually makes both the reporting and the bidding murkier.
• Different conversion goals. If one campaign is chasing purchases and another is chasing newsletter sign-ups, they shouldn’t optimize toward a blended goal.
• Brand versus non-brand. This one is so consistently recommended that it gets its own section below.
• Compliance or policy reasons. Regulated categories sometimes need isolation so that one disapproval doesn’t drag everything around it.

Notice what isn’t on the list. Match type isn’t there. Device isn’t there. “So I can see it in the reports” isn’t there, because there are segmentation views in the interface that let you slice a single campaign by device, network, day or audience without fragmenting the bidding. Splitting a campaign to get a report that a segment column would give you for free is one of the more common forms of self-inflicted damage I see described.

The pattern to watch for in your own account is splitting for the sake of control you never actually use. If you create a campaign for mobile and a campaign for desktop and then give them the same target and the same budget and never touch either, you’ve bought nothing and paid in data.

Brand and non-brand: the split that almost always earns its place

Searches for your own company name behave differently from everything else. They convert at much higher rates, cost less per click, and represent demand you’d largely get anyway. Blend them into a general campaign and they quietly flatter it. The campaign looks efficient because a slice of it is people who were already looking for you, while the expensive generic terms are doing worse than the average suggests.

Keeping brand in its own campaign lets you set a separate budget, choose a bid strategy that fits (several guides recommend Target Impression Share or a modest Target CPA for brand, which fits how brand behaves), and read your non-brand numbers without contamination. We looked at whether bidding on your own name is worth doing at all in should you bid on your own brand name in Google Ads. Whatever you decide there, if you do run brand, it belongs in its own container.

The reverse problem matters just as much once Performance Max is in the account. Performance Max can and does serve on branded searches, and because those searches convert well, they can inflate the campaign’s reported return while it does little to find new customers. Google provides brand exclusions for Performance Max, and the guides I read describe them as generally more reliable than plain negative keywords, because exclusions can catch misspellings and variants in other scripts that a negative keyword list would miss. One widely repeated claim says search term overlap between Search and Performance Max shows up in the large majority of accounts. I couldn’t trace that to a primary source, so I wouldn’t quote the percentage, but the mechanism is real and easy to check yourself in the search term insights.

A workable setup for most accounts that have both: a brand Search campaign, a non-brand Search campaign (or several, following the rules above), and a Performance Max campaign with brand exclusions applied, plus a shared negative keyword list carrying your brand terms on the non-brand campaigns so they don’t cannibalize the brand campaign’s traffic. The mechanics of building those lists are covered in our guide to negative keywords.

Where Performance Max and AI Max change the picture

Performance Max and AI Max for Search both blur the old boundaries between campaign types. They deserve a separate look because a structure that made sense before them can work against them.

Performance Max works on asset groups, not ad groups. An asset group is a bundle of images, headlines, descriptions and videos tied to a theme and, optionally, an audience signal. The most common structural error reported in the guides is putting everything into one asset group: every product, every service, all in one pile. The system then has no clear theme to pair with each creative, and you can’t tell which part of the offer is earning anything. The better approach is to organize asset groups around a product category, a service line or a customer type, so that each has its own coherent creative. What audience signals do and don’t do is worth understanding before you lean on them. We wrote about that in Performance Max audience signals.

Because Performance Max offers so little visibility, your structure is one of the few levers that gives you any readability at all. If you run one campaign per distinct product line with its own budget, you can at least see which line is doing the work. If you run a single campaign for everything, you see one number. That’s an argument for splitting Performance Max by business goal or margin tier, subject to the same data-volume limit as anywhere else. If you can’t feed two Performance Max campaigns enough conversions to learn, run one. Our walkthrough of how to audit a Performance Max campaign when you can’t see inside it covers how to check what a campaign is actually doing.

AI Max for Search adds another wrinkle. It expands matching beyond your keywords using your landing pages and assets, which means the keyword-level structure you carefully built matters a bit less and the quality of your landing pages and the coherence of your ad groups matters a bit more. If your ad groups are a grab bag of loosely related themes, AI Max has more ways to drift. Tighter themes plus negative keywords give it a narrower lane. We went through the practical side in AI Max for Search campaigns: what it actually changes.

One honest caveat about all of this: the automated campaign types change quickly, and some advice on how to structure around them has a shelf life measured in months. Anything I say about them should be checked against what your account is actually doing, and against current Google documentation. The durable principles are the ones about data volume and separating things that genuinely differ.

Match types, ad groups and the keyword layer

Inside a Search campaign, the keyword layer has gotten simpler too. The old habit of duplicating every keyword across exact, phrase and broad in separate campaigns or ad groups is one of the things Google’s current guidance steers away from. The reasoning in the guides is that the same keyword in three match types competes against itself, splits your data three ways, and forces you to manage three sets of everything. A few sources go further and say to consolidate match types into one ad group and lean on broad match with Smart Bidding. That’s a bigger bet than it sounds, and it depends on having decent conversion tracking and a good negative list. If your tracking is shaky, broad match will happily optimize toward garbage. I’d consider that step only after the tracking is sound. We covered how the match types behave now in what broad, phrase and exact actually mean now.

For ad groups, here is a practical process that doesn’t depend on any published ratio:

• Export your keywords along with their search terms for the last 90 days.
• Sort them by the landing page that would be the right destination. Keywords that belong on the same page are candidates for the same ad group.
• Within each of those, check whether one headline could plausibly promise what every keyword is asking for. If a dental clinic has “emergency dentist” and “teeth whitening” on the same page, they need different ad groups even though the page is the same, because the promise in the ad should differ.
• Merge any ad group that has only a single keyword and less than a few dozen clicks, unless there’s a hard business reason to keep it separate.
• Check that every ad group has at least one responsive search ad with enough distinct headlines to be useful, rather than a few near-duplicates.

That last point connects to ad strength, which is easy to over-weight. A “Poor” rating on an ad in a well-themed group isn’t a structural problem. A group so loose that no ad could be strong for it is. The piece on what ad strength actually measures goes into why the two get confused.

Shared budgets and portfolio strategies

When you have several campaigns that share a goal and each is too small to learn well alone, portfolio bid strategies are the tool that lets you keep the separation while pooling the learning. A portfolio strategy groups campaigns under one bidding strategy and one target, so the system learns from the combined conversions. One third-party write-up quotes a Google figure of about 13% more conversions on average for advertisers using shared budgets together with portfolio strategies on Search. I haven’t been able to verify that from Google’s own page, so treat it as a claim, not a benchmark. The logic, though, is sound.

The constraint is that the campaigns in a portfolio need to actually share a goal. Put a lead generation campaign and an e-commerce campaign with very different conversion values in the same portfolio and you’ve given the strategy contradictory instructions. Group by objective and approximate economics, not by convenience. We wrote the long version, including the ways pooling quietly backfires, in portfolio bid strategies: when pooling campaigns together helps, and when it quietly hurts.

Shared budgets are a different thing and a more dangerous one. A shared budget lets several campaigns draw from one pool of daily spend, which sounds efficient until one campaign eats the whole pool by mid-morning and the others never get to run. If two campaigns need different amounts of money to do their jobs, give each its own budget. Use shared budgets only when you really don’t care how the money splits between them.

Three structures for three kinds of account

It’s more useful to see how the principles work out for different situations than to imagine a universal template. These are illustrations of reasoning, not prescriptions, and none of the conversion figures below come from real accounts.

A small local business with a modest budget

Suppose a plumbing company spends a few thousand a month and gets, say, 20 leads a month through the form and phone. There is very little data here. The structure should be as plain as possible: one Search campaign covering the core services, organized into a handful of single-theme ad groups (emergency repair, water heater, drain cleaning), tight location targeting, call and form conversions tracked properly. Brand gets its own small campaign only if competitors are bidding on the company name, otherwise it can sit inside the main campaign with a note to yourself to review it. No Performance Max until the search campaign has enough conversions to be stable and you know what it would add.

Splitting this account into a campaign per service would feel organized and would leave each one with a few conversions a month. You’d get very tidy dashboards full of very unreliable numbers. The approach that works better for this kind of business is covered on our page for local service businesses, and location settings matter more than anything else here, which is the subject of our location targeting piece.

An online store with a product catalog

An e-commerce account has a different shape because the catalog is large, margins vary by product, and Shopping and Performance Max are the main engines. Here the structure usually turns on margin and product type, not keyword theme. A reasonable layout might put high-margin products in one campaign with a more aggressive return target, the long tail in another, and a separate campaign for best sellers that you want guaranteed budget for. Brand search sits separately with Performance Max brand exclusions in place so the store isn’t paying for customers who were going to buy anyway.

The temptation in e-commerce is to build a campaign per category because the product feed makes it easy. Do that only if the categories genuinely have different economics. Otherwise you fragment conversions and create a dozen campaigns that each need babysitting. If your feed has problems, structure won’t save you, since disapproved products can’t serve anywhere, a problem covered in our Merchant Center feed errors article. The broader picture for stores is on our e-commerce page.

A B2B company with a long sales cycle

B2B accounts tend to have low volume, expensive clicks, and a problem much bigger than structure: the thing Google sees as a conversion (a form fill) is a poor stand-in for the thing the business cares about (a closed deal). With few conversions per month the case for consolidation is strong. Fewer campaigns, themed around the problem the buyer is trying to solve, not around product features. Most of the real leverage sits in what you feed back to Google, not in campaign count. If you’re passing qualified-lead data back through offline imports, we wrote about a change that could affect that in offline conversion imports in 2026. For this audience our B2B and SaaS page is the place to start.

How to audit the structure you already have

If you’re inheriting an account or finally looking hard at your own, there’s a sequence that keeps it from turning into a rewrite. The aim is to find the changes that matter and leave the rest alone, because restructuring an account that’s working is one of the quickest ways to reset everything that Smart Bidding has learned.

Step one: count conversions per bidding unit. For each campaign (or portfolio), look at conversions over the last 30 and 90 days. List any campaign with very few. Those are your consolidation candidates. Write down for each one why it was separated in the first place. If nobody remembers, that’s informative.

Step two: list what differs. For every pair of similar campaigns, check whether anything meaningful is different between them: budget, target, location, goal, audience. If the answer is “no” or “barely”, they’re a merge candidate. If two campaigns have identical settings and identical targets, the split is decorative.

Step three: check for internal competition. Look at your search term reports across campaigns. If the same queries are being served by several campaigns, you have overlap, which muddies both data and spend. Brand queries showing up in non-brand campaigns or in Performance Max are the usual culprit. The walkthrough for reading those reports is in how to read a Google Ads search term report.

Step four: check the ad group layer. Find single-keyword groups with thin data, and find the opposite: bloated groups with dozens of unrelated keywords. Both are symptoms. The first wastes data; the second wastes relevance.

Step five: look at budget allocation against performance. Is the money going where the conversions are? Campaigns flagged as limited by budget while others underspend are a structural signal. What that label really means, and how it’s changing, is covered in our piece on “Limited by budget”.

Step six: sequence the changes. Don’t merge five campaigns on the same day. Merge one, let it settle, and look again. Before you start, open your change history and get into the habit of noting the date of every structural change, so that when performance moves you can tell whether it was the restructure or something else.

How to merge campaigns without wrecking performance

Consolidation is easy to recommend and surprisingly easy to botch. A few things that reduce the damage.

First, pick a survivor. When merging two campaigns, keep the one with the most history and the healthiest bidding state, and move the other’s keywords, ads and settings into it. A brand-new campaign starts learning from zero, whereas an existing campaign keeps at least part of what it knows.

Second, pause before you delete. Keep the campaign you’re retiring paused for a few weeks. If something unexpected happens, you can see exactly what the old settings were. Removing a campaign destroys the evidence.

Third, expect some turbulence. Smart Bidding will likely wobble for a short period after a structural change because the pool it’s learning from has changed. Whether it’s a week or several depends on how much data is involved. If you’re in a peak season, don’t do it. Restructure in a quieter window. For accounts heading into Q4, we put together a budget pacing checklist that’s worth reading before you change anything in the next couple of months.

Fourth, tell the system about known oddities. If the restructure coincides with a tracking outage, a sale, or a site change, there are tools for that. The difference between excluding bad data and adjusting for expected swings is covered in data exclusions versus seasonality adjustments.

Finally, keep an eye on the numbers that show whether the merge worked: cost per conversion, conversion volume, impression share, and the share of budget lost. Judge it over a couple of weeks, not a couple of days.

Mistakes that keep showing up

A few patterns come up across nearly every write-up on structure, and they match what you’d expect from how bidding works.

Structure that mirrors your org chart, not your customers. Campaigns named after internal teams or product owners. Customers don’t search by department.

Tracking problems blamed on structure. If conversion numbers are wrong, restructuring doesn’t help. You’ll get a cleaner account pointing at the wrong target. Audit tracking first. There’s a full walkthrough in why your Google Ads conversion numbers don’t add up.

Experiments run as permanent campaigns. A test campaign created “for a few weeks” that’s still spending eighteen months later. If you want to compare two approaches properly, the built-in experiment tools do it with a traffic split, instead of running them side by side and hoping. We covered how to set that up without fooling yourself in Google Ads experiments.

Over-reliance on automated recommendations. Google’s recommendations tab sometimes suggests structural changes, such as adding keywords or switching bid strategies, and auto-apply can make them without you. Some of those are fine. Some will quietly reshape your account. Our look at the optimization score and auto-apply explains which to be careful with.

Treating structure as a one-time project. An account that was sensibly organized two years ago drifts. New campaigns get added, old ones get forgotten, goals change. A structure is a description of your business at a moment in time, and it needs revisiting when the business changes.

Where ongoing monitoring fits in

Structure problems are slow. Nothing alarms when a campaign quietly slips below the conversion volume it needs, or when a brand term begins leaking into a prospecting campaign, or when a new campaign duplicates the targeting of an old one. These things show up as a gradual drift in cost per conversion that’s easy to attribute to the market. Part of the reason they persist is that most people look at an account in bursts and not continuously. We’ve written about how often you should actually check your account, and the honest answer depends on spend, but the general problem is the same.

This is the kind of thing that Growera’s Google Ads manager is built around: it checks the account every day, flags things like campaigns with thin data, overlapping search terms and budget imbalances, and explains what it found and what it would suggest, so you’re not relying on noticing the drift yourself. You can see how that works on the ERA AI page and what it costs on the pricing page. It’s a monitoring and recommendation layer, and the structural decisions described above remain yours to make. Whether you use a tool like that or a calendar reminder and a spreadsheet, the useful habit is the same: look for drift on a schedule instead of waiting for a bad month to force the question.

A short decision guide

If you want something to keep next to the account, here is the reasoning compressed into questions you can ask of any structural choice.

• Does this split let me control something I actually want to control differently (budget, target, location, goal)? If not, don’t split.
• Will each resulting campaign still get enough conversions to learn from? If you’re unsure, err toward fewer campaigns.
• Could one ad and one landing page answer every keyword in this ad group? If not, split the ad group.
• Is brand traffic separated from everything else, and is Performance Max blocked from taking it?
• Do my Performance Max asset groups each have a coherent theme?
• Am I about to change this during a busy season or right after another big change? If so, wait.

Summary

Account structure exists to let you control the things that really need to differ and to give the bidding system enough data in each place to learn. Everything else is decoration. The long-standing advice to slice accounts into the smallest possible pieces made more sense when bids were set by hand. With Smart Bidding setting them in each auction, fragmentation now mostly costs you data, and the guidance from Google and from most practitioners has moved toward fewer campaigns and themed ad groups.

That doesn’t mean everything goes into one campaign. Split when budgets, targets, locations, conversion goals or brand versus non-brand behavior truly differ. Keep ad groups to keywords that one ad and one page can serve. Organize Performance Max around coherent asset groups and block it from brand traffic. When campaigns share a goal but are too small to learn alone, consider a portfolio strategy. And when you restructure, change one thing at a time, keep the old campaign paused instead of deleting it, and avoid doing it in your busiest weeks.

The specific numbers in circulation, such as conversions per month, ad groups per campaign, and keywords per group, are rules of thumb from practitioners and vendors, not thresholds Google enforces. Use them to check your instincts, not to replace checking your own data. If you want to start somewhere this week, pick your smallest campaigns and ask the first question above: what exactly does this one let me control that the others don’t? If there’s no answer, you’ve found your first merge.