Google Ads Campaign Structure: Setup Best Practices

Structure decisions made once at setup keep compounding for the life of an account — a poorly organized campaign doesn't just look messy, it actively works against Quality Score, makes Smart Bidding's job harder, and turns every future audit into a bigger job than it needed to be. This covers the levels worth understanding, a debate that's been settled by how modern Smart Bidding actually works but that a lot of older advice still gets wrong, and a worked example of taking a genuinely messy account and fixing it.

This assumes you already know the basic account-campaign-ad group hierarchy, covered in our beginner's guide — this goes further into the actual structural decisions worth making deliberately, not just what the levels are called.

None of what follows requires anything beyond what's already available inside a standard Google Ads account — these are structural decisions, not premium features requiring a separate purchase.

The three levels, and what belongs at each

Campaigns are where budget, bidding strategy, campaign type, and geographic targeting live. Treat a campaign as a distinct strategic initiative with its own goal, not just a folder for grouping similar keywords — a campaign's budget and bid strategy should reflect a single coherent objective, since mixing objectives at the campaign level (some keywords chasing volume, others chasing efficiency, under one shared budget) makes both harder to manage well.

Ad groups sit inside a campaign and group closely related keywords with the ads meant to match them. The tightness of an ad group's theme directly affects how relevant the ads inside it can be to the searches triggering them, which flows straight into Quality Score.

Keywords and ads live at the ad group level — the actual terms you're bidding on and the actual creative shown. Everything above this level exists to organize these two things sensibly; a technically correct hierarchy with poorly chosen keywords or weak ads still performs poorly, so structure is necessary but not sufficient on its own.

It's worth keeping these three levels' distinct responsibilities in mind whenever you're deciding where a new setting or change belongs — a common structural mistake is trying to solve an ad-group-level problem (a specific keyword's poor relevance) with a campaign-level lever (an overall bid increase), which spends more without addressing the actual, more targeted cause. Match the level of the fix to the level of the problem.

SKAGs vs. STAGs, settled

This debate has real history worth understanding before dismissing either side. Single Keyword Ad Groups (SKAGs) — one keyword per ad group, with ad copy written to match that exact term precisely — were genuinely good advice for years, when manual bidding and tight keyword-to-ad matching were the primary levers available for improving Quality Score and relevance.

The reasoning has changed because the platform has changed. Smart Bidding strategies need a meaningful volume of conversion data per campaign or ad group to make good decisions — covered in more detail in our KPIs guide, where roughly 30 conversions in a recent 30-day window is a commonly cited threshold before an automated strategy has enough signal to perform reliably. Fragmenting an account into dozens or hundreds of single-keyword ad groups splits that conversion data into pieces too small for Smart Bidding to learn from effectively in any one of them, even if the account's total conversion volume would have been more than sufficient pooled together.

Single Theme Ad Groups (STAGs) — 5 to 15 closely related keywords sharing genuine search intent, grouped together — solve this by pooling enough conversion volume per ad group for Smart Bidding to actually learn from, while still keeping the theme tight enough that ad copy can remain genuinely relevant to the group as a whole. This is the right default for most accounts running Smart Bidding today, which is most accounts, given how strongly Google has pushed automated bidding as the default recommendation.

SKAGs still have a narrow, legitimate use case: a small number of extremely high-value, high-volume individual keywords where the search intent is genuinely singular and distinct enough that a dedicated ad group, with ad copy written to match that one term with maximum precision, produces real value — your own brand name is a common example. For the bulk of an account's keyword list, though, STAGs are the better default, and building an entire account on the SKAG model is applying advice that made sense under a bidding paradigm the platform has largely moved past.

If you're managing an older account still built on the SKAG model and considering whether to consolidate, weigh the disruption to Smart Bidding's existing learning (covered in the FAQ below) against the ongoing cost of fragmented data — an account that's been struggling with Smart Bidding performance specifically, with hundreds of single-keyword ad groups each individually too thin on data, is a strong candidate for consolidation despite the short-term relearning disruption. An account performing acceptably well despite a SKAG structure is a weaker case for change purely on structural-purity grounds.

A naming convention worth adopting from the start

A consistent naming convention across campaigns and ad groups pays for itself the moment an account grows past a handful of campaigns, or the moment anyone besides you needs to understand it — including a future version of you returning to the account after months away. A reasonable format: [Campaign type]_[Intent segment]_[Location if relevant] for campaigns, and [Theme]_[Match type if mixed] for ad groups — something like Search_NonBrand_US or Search_Branded, and within it, ThemeA_Exact or ThemeA_Phrase if you're deliberately splitting match types into separate ad groups for the same theme.

The specific format matters less than consistency — pick a convention, document it somewhere accessible (even a single note in the account's shared documentation), and apply it to every new campaign and ad group going forward, rather than letting names accumulate organically with no pattern, which is how a six-month-old account ends up with a mix of "Campaign 1," "NEW - test," and "Spring Promo (OLD)" all sitting in the same account with no clear indication of what's still active or why.

This matters more than it might seem worth the setup effort, specifically because the cost of skipping it is invisible until it isn't — nobody notices a missing naming convention until the day someone new needs to understand the account quickly, or an audit needs to happen fast, and that's exactly the wrong moment to discover the structure only makes sense to the one person who originally built it.

Segmenting campaigns by intent

Branded campaigns — bidding on your own brand name — deserve their own campaign, separate from everything else, almost always. This traffic converts at a meaningfully higher rate and lower cost than any other segment, and blending it into a general campaign lets its strong performance mask weaker performance elsewhere in a shared, averaged budget and bid strategy.

Non-brand (generic category) campaigns need your own brand name added as a negative keyword specifically, so the two campaigns aren't competing against each other in the same auction — without this, budget that should be driving new-customer acquisition ends up capturing branded searches your branded campaign would have caught anyway, for less.

Competitor campaigns — bidding on a rival's brand name, covered in detail in our dedicated guide — belong in their own isolated campaign too, both because the economics genuinely differ (lower Quality Score, different conversion rate, covered in that guide) and because keeping this traffic isolated makes it far easier to judge honestly whether the tactic is actually working, rather than having it blend into a broader campaign's average.

Remarketing campaigns, targeting people who've already visited your site, warrant separation as well — this audience has fundamentally different intent and typically a much higher conversion rate than cold traffic, and mixing remarketing bids and budget into a general prospecting campaign obscures both how well remarketing is actually performing and how much of your "generic" campaign's apparent success is really coming from warm, returning visitors rather than genuinely new demand.

The underlying reason all four of these deserve separation is the same one driving the SKAG-versus-STAG conclusion above: Smart Bidding performs better when each campaign's conversion data reflects a genuinely coherent, consistent intent, rather than a blend of fundamentally different searcher behaviors averaged together under one shared bid strategy.

None of these four segments need to launch simultaneously on day one — the complexity-earned-through-data principle covered later in this piece applies here too. Start with whichever segment genuinely reflects where your current traffic and budget already sit, and add the others as real, separate volume justifies giving each its own dedicated campaign.

Match types within a theme, structurally

Whether to mix match types within one STAG or split them into separate ad groups by match type is worth a deliberate decision rather than defaulting either way. Mixing exact and phrase match for the same theme in one ad group is simpler to manage and keeps conversion data pooled together for Smart Bidding, which is the right default for most accounts, most of the time.

Splitting match types into separate ad groups (an exact-match version and a phrase or broad-match version of the same theme) makes sense specifically when you want to bid differently on the tightly-controlled exact matches versus the broader, more exploratory phrase or broad matches — since bids are set at the ad group level, this split gives you that granular control at the cost of splitting your conversion data into two smaller pools rather than one larger one, which works against the STAG reasoning covered above if either resulting group ends up too small for Smart Bidding to learn from effectively.

A reasonable rule of thumb: keep match types together within a theme by default, and only split them once you have enough volume that each resulting split group would still individually clear a meaningful conversion threshold — splitting for bid control at the cost of starving both halves of adequate learning data usually isn't a good trade.

Check your own ad groups' actual conversion volume before deciding either way rather than applying a blanket rule across the whole account — some themes will have enough volume to support a split, others won't, and the right answer can genuinely differ theme by theme within the same account.

A worked example: restructuring a messy account

Say you inherit an account with one Search campaign called "Main Campaign," containing 40 keywords spanning your brand name, several generic category terms, and two competitor brand names, all sharing one ad group with six generic ads written to vaguely cover all of it.

First pass: split by intent. Pull the brand name and its close variants into a new Branded campaign. Pull the two competitor brand names into a new, isolated Competitor campaign. Everything else — the genuine generic category terms — stays in a Non-Brand campaign, with the brand name added there as a negative.

Second pass: theme the ad groups within Non-Brand. Instead of one ad group holding all 30-plus remaining generic keywords, group them into 3-6 STAGs based on genuine shared intent — if the business sells both a product and a service, for instance, those are two different themes deserving separate ad groups, each with its own ad copy speaking specifically to that theme rather than one set of generic ads trying to cover both.

Third pass: write ad copy for each new ad group specifically, rather than reusing the same six generic ads everywhere. Each STAG's ads should speak directly to that group's specific theme — a genuine, deliberate improvement in relevance that a shared set of generic ads across a mixed keyword list could never achieve, regardless of how well-written those generic ads were.

The result: four campaigns instead of one (Branded, Non-Brand, Competitor, plus Remarketing once conversion volume justifies adding it), each with a coherent budget and bid strategy matched to its own actual performance profile, and 3-6 tightly themed ad groups inside Non-Brand instead of one catch-all group diluting both relevance and Smart Bidding's learning signal.

This kind of restructure is exactly the situation where the sequencing caution from our audit checklist applies most directly — implement the changes, then give the new structure genuine time (the same 30-conversion, 30-day rough window covered in our KPIs guide) before judging its performance against the old one, rather than reacting to the first week's inevitably noisier, still-relearning numbers as if they represented the new structure's real, settled performance.

Budget allocation across a properly segmented structure

Fund branded search first and fully, before anything else. It's your cheapest, highest-converting traffic, and under-funding it (or worse, not running a branded campaign at all) risks losing that position to a competitor bidding on your name, covered in detail in our competitor-bidding guide — defending your own brand terms is rarely optional once you understand the downside of not doing it.

Allocate non-brand budget based on actual, ongoing performance data by ad group, not an even split assumed at setup. Once STAGs have accumulated real conversion data, some themes will clearly outperform others — shift budget toward what's proven rather than leaving it distributed the way it happened to be divided when the account was first built.

Treat competitor-bidding and remarketing budgets as distinct line items with their own performance bar, evaluated against their own realistic economics (covered in the dedicated guides for each) rather than judged against your account-wide average, which would unfairly penalize remarketing's naturally higher conversion rate or competitor bidding's naturally higher CPA as if either were a flaw in execution rather than an expected characteristic of that specific traffic type.

Revisit this allocation on a real recurring schedule — monthly is reasonable for most actively managed accounts — rather than setting it once during the initial restructure and leaving it static. Relative performance between segments shifts as competition, seasonality, and your own product mix change, and a budget split that was correct at restructure time can drift out of alignment with actual current performance within a few months if nobody revisits it.

Shopping-specific structure notes

Shopping campaigns organize around product groups rather than keywords, and campaign priority (High, Medium, Low) becomes the key structural lever instead of ad group theming. A common, effective pattern: a low-priority campaign with aggressive bids for your best-margin or best-converting products, and a higher-priority campaign with more conservative bids covering the rest of your catalog — priority determines which campaign's bid wins when a product could technically show in more than one, letting you direct more aggressive spend specifically toward your proven top performers without needing to manage bids product by product across the whole catalog.

This is a genuinely different structural model from Search's ad-group theming, worth understanding as its own system rather than trying to force Search-style single-theme grouping logic onto a Shopping campaign's product-group structure.

Product group segmentation within a Shopping campaign benefits from the same underlying logic as Search's ad-group theming — group products that share genuine performance characteristics (similar margin, similar demand pattern) together, rather than segmenting purely by the product feed's existing category taxonomy if that taxonomy doesn't actually reflect meaningful performance differences.

When to keep it simple versus when to add complexity

A brand-new account with limited budget and no performance history yet should start simpler than the fully segmented structure above — a single well-themed Search campaign with a handful of STAGs is a reasonable starting point, with the intent-based segmentation covered above added once there's enough real data and volume to justify separate campaigns rather than fragmenting a small budget across four campaigns each too small to learn effectively.

The signal that it's time to add complexity is data volume, not time elapsed — once a specific segment (branded traffic, a particular product theme) has accumulated enough conversions to be worth its own dedicated budget and bid strategy, splitting it out becomes genuinely useful rather than premature. Splitting too early, before any one piece has enough data to perform well independently, recreates the same fragmentation problem SKAGs caused, just at the campaign level instead of the ad group level.

Structure consistency across a Google Ads manager account

If you're applying this structure across several client accounts under an MCC, a consistent naming convention and segmentation approach — the same Branded/Non-Brand/Competitor/Remarketing pattern, applied uniformly — pays off considerably more than it does on a single account, since it lets anyone on your team jump into any client's account and understand its structure immediately, without needing a client-specific orientation each time.

It's reasonable to adapt the specific segmentation to each client's actual business (not every client needs a competitor campaign, for instance, if competitor bidding isn't part of their strategy), but keeping the underlying structural philosophy — intent-based campaign segmentation, single-theme ad groups sized for adequate Smart Bidding data — consistent across the roster makes both onboarding new team members and auditing existing accounts meaningfully faster than treating each client's structure as a bespoke, one-off decision.

Document the shared convention once, somewhere every team member managing accounts can reference it, rather than relying on informal knowledge passed between people — the same documentation habit worth building into account structure generally, covered earlier, matters even more once the account count grows past what any one person can hold in memory.

How many keywords should be in a single-theme ad group?

Roughly 5 to 15 is a reasonable working range for most accounts — enough to pool meaningful conversion volume without diluting the theme's coherence. Fewer than that starts approaching SKAG-level fragmentation; meaningfully more than that often signals the theme itself is too broad and would benefit from splitting into two more specific ad groups rather than continuing to add keywords to an increasingly loose group.

Should I restructure an existing, working account, or only build new campaigns this way?

If an existing account is genuinely performing well, restructuring it purely for structural tidiness carries real risk — significant structural changes reset some of Smart Bidding's accumulated learning, covered in our audit checklist regarding sequencing changes carefully. Restructuring is worth prioritizing when an account is underperforming and a structural audit (also covered in that guide) points at genuine structural causes, not as a routine tidy-up applied to something that's already working.

Does this structure advice apply the same way to Performance Max?

Not directly — PMax organizes around asset groups rather than keyword-based ad groups, and it doesn't offer the same granular, keyword-level structural control Search does. The underlying principle still applies in spirit, though: distinct, coherent asset groups built around genuinely different products or themes generally outperform one generic asset group trying to cover an entire broad catalog, for similar relevance and signal-quality reasons.

What's a reasonable number of campaigns for a small business account?

There's no universal number, but four to six is a common, reasonable range once an account has matured past initial setup — Branded, Non-Brand, and Remarketing at minimum for most businesses with any meaningful branded search volume and returning-visitor traffic, with Competitor and additional product-line or location-based campaigns added as genuinely justified by data volume and strategic priority, following the same complexity-earned-through-data principle covered above.

Will restructuring reset my Quality Score?

Not directly in most cases — Quality Score is tracked at the keyword level and generally carries forward with the keyword regardless of which ad group or campaign it moves to, provided the keyword itself isn't paused and recreated as an entirely new keyword. What does reset, or at least temporarily disrupt, is Smart Bidding's learning phase for the affected campaigns, since significant structural changes alter the data patterns the algorithm has been learning from — this is the more relevant concern when timing a restructure, covered in the sequencing note in our audit checklist.

Should location-based segmentation be its own campaign or just a targeting setting?

It depends on whether the locations genuinely need different budgets, bid strategies, or messaging. If you're a single business with one general offer serving several regions similarly, location targeting settings within one campaign are usually sufficient. If different locations have meaningfully different budgets, competitive intensity, or require distinct localized messaging (different store hours, region-specific offers), separate campaigns per location give you the independent budget and bid control that a single campaign with location targeting settings can't provide.

The short version

Use single-theme ad groups of 5-15 closely related keywords, not single-keyword ad groups — the Smart Bidding era rewards pooled conversion data over maximum keyword-level precision. Segment campaigns by genuine intent (branded, non-brand, competitor, remarketing), since mixing fundamentally different searcher behaviors under one shared budget and bid strategy starves the automation of clean signal. Add structural complexity as data volume justifies it, not on a fixed timeline, and adopt a naming convention from day one rather than retrofitting one onto an account that's already grown unwieldy.

None of this is about structure for its own sake — every recommendation above ultimately traces back to giving Smart Bidding cleaner, better-pooled data to learn from, which is the actual mechanism connecting good structure to better real-world performance.

Is it ever appropriate to run more than one campaign for the same theme?

Occasionally, though it's worth a deliberate reason rather than a default. A genuine case: running the same product theme in both a Search campaign and a Performance Max campaign, where the two serve different purposes (Search for granular, controllable targeting on your best-understood terms; PMax for broader automated reach across additional networks) rather than simply duplicating the same effort twice. Running the same theme in two Search campaigns with no clear distinguishing purpose between them, by contrast, usually just means they're competing against each other in the same auctions — the internal keyword-overlap problem covered in our audit checklist — and consolidating into one campaign is typically the better structural choice.