Most underperforming Google Ads accounts we’re asked to rescue in Singapore don’t have a bidding problem or a budget problem — they have a structure problem. Everything is crammed into one campaign, unrelated keywords share an ad group, and a single message tries to speak to searchers who want completely different things. The result is low relevance, wasted spend and reporting you can’t act on. Account structure is the skeleton everything else hangs on: it decides where your budget goes, how relevant your ads can be, how well Smart Bidding learns, and whether you can actually see what’s working. And while Google’s AI has taken over a lot of the day-to-day optimisation, structure hasn’t stopped mattering — it’s changed shape. This guide explains how to structure a Google Ads account properly in 2026, in plain English, for a Singapore SME.
If you’re still getting oriented, our complete Google Ads guide for Singapore SMEs covers the platform end to end; this post zooms in on the architecture that makes the difference between an account that compounds and one that leaks.
The four levels, and what each controls
A Google Ads account has a simple nested hierarchy, and knowing which settings live at which level is half the battle.
- Account. The top level. It holds your billing, time zone, users and permissions, conversion tracking, and account-level negative keyword lists. Settings here apply everywhere.
- Campaign. This is the level that matters most, because the big controls live here: your daily budget, your bid strategy, your location and language targeting, your ad schedule, and which networks you’re on. Anything you want to budget or bid for separately has to be its own campaign.
- Ad group. A container inside a campaign for one tight theme — a set of closely related keywords and the ads written to match them. One theme per ad group is the whole point.
- Keywords and ads. The bottom level: the search terms that trigger your ads (governed by match types) and the responsive search ads themselves.
The single most important consequence of this is that budget and bidding are campaign-level settings. If you want to spend more on your most profitable service, or bid differently in one region, or pause spending on something after office hours — each of those is a reason to make it a separate campaign. Structure isn’t bureaucracy; it’s how you get control over money and message.
Does structure still matter now that Google’s AI runs bidding?
It’s a fair question, and the answer is yes — but differently than it used to. Smart Bidding and broad match have absorbed a lot of the manual tuning that structure once existed to enable. But structure still does four things automation can’t:
- It controls budget. Google will never move budget between campaigns on its own the way you’d want; splitting your priorities into separate campaigns is how you guarantee your best service isn’t starved by a weaker one.
- It drives relevance. Tightly-themed ad groups let you write ads and point to landing pages that genuinely match the search — the foundation of a healthy Quality Score, which directly lowers your cost per click.
- It feeds Smart Bidding the right signal. Modern automated bidding needs conversion volume to learn. Group things so each campaign accumulates enough data — too many thin, fragmented campaigns starve the algorithm; see our guide to Smart Bidding.
- It makes reporting usable. A logical structure means your reports answer real questions — which service, which region, which intent is profitable — instead of one undifferentiated blob.
So the goal in 2026 isn’t the hyper-granular structures of a decade ago. It’s a structure tight enough to be relevant and readable, but consolidated enough that Smart Bidding has volume to work with. That balance is the whole game.
How to split your campaigns
Because budget and bidding live at campaign level, deciding your campaigns is really deciding how you want to control money and targeting. The main ways to split, used alone or together:
- By objective / campaign type. Keep Search, Shopping, Performance Max, Demand Gen and Display in separate campaigns — they behave differently and you never want them sharing a budget. Our campaign types guide covers when to use each.
- By product or service line. The most common and usually the best split for SMEs. A renovation firm might run separate campaigns for “kitchen renovation,” “HDB renovation” and “commercial fit-out,” so each gets its own budget and message.
- By profitability / priority. Put your highest-margin or highest-value service in its own campaign so you can fund it deliberately, rather than letting it compete for scraps inside a general campaign.
- By geography. If you serve different regions with different value or competition — or you want to concentrate on, say, the CBD and east Singapore — separate campaigns let you budget and bid per area.
- Brand vs non-brand. Almost always worth splitting. Searches for your own company name convert cheaply and shouldn’t be blended with the more expensive non-brand prospecting terms, or your reporting and bidding both get distorted.
Resist the urge to over-split. A dozen near-identical campaigns fragments your data and your attention. Start with the fewest campaigns that give you the budget control and reporting clarity you actually need — for many Singapore SMEs that’s two to five — and add more only when a real need appears.
The big shift: STAGs replaced SKAGs
Inside each campaign, you organise keywords into ad groups — and this is where the biggest change in modern account structure has happened. For years the fashionable approach was the SKAG (Single Keyword Ad Group): one keyword per ad group, for maximum control over the ad shown. That logic has broken down. Since Google widened close-variant matching and replaced expanded text ads with responsive search ads, a “single keyword” ad group actually matches dozens of query variations anyway, so the precision SKAGs promised is largely gone — and they fragment conversion data across hundreds of tiny ad groups, which is exactly what Smart Bidding hates.
The current best practice is the STAG (Single Theme Ad Group): group 5–15 closely related keywords that share the same intent into one ad group, with ads written to that shared theme. STAGs keep ads relevant while giving each ad group enough volume for Google’s bidding to learn from. SKAGs still have a narrow place — for a handful of your very highest-value, highest-volume keywords where the extra control genuinely pays off — so many accounts run a hybrid. But the default in 2026 is single-theme, not single-keyword.
How many campaigns, ad groups and keywords?
There’s no universal rule, but these are sensible 2026 ranges for an SME account — guardrails, not laws:
| Level | Sensible range | Why |
|---|---|---|
| Campaigns per objective | ~3–5 (fewer for small accounts) | Enough to control budget by priority without fragmenting data |
| Ad groups per campaign | ~5–10, tightly themed | More than ~10 usually means it should be another campaign |
| Keywords per ad group | ~5–20, one intent | Enough coverage for a theme without diluting relevance |
| Ads per ad group | 2–3 responsive search ads | Google can test and rotate; 3 is the enabled RSA limit |
Smaller Singapore SMEs should err toward the lower end. It’s far better to start with one or two well-built campaigns you can actually manage than ten half-baked ones. You can always expand a winning structure; you can rarely un-tangle a sprawling one without a rebuild.
A worked Singapore example
Take a Singapore aircon servicing company that also installs new units and does repairs. A clean structure might look like this:
- Campaign 1 — Servicing (Search). Highest-volume, bread-and-butter work; its own budget. Ad groups: General servicing (aircon servicing, aircon maintenance), Chemical wash (chemical wash, chemical overhaul), Contract servicing (servicing contract, quarterly servicing).
- Campaign 2 — Installation (Search). Higher-value jobs, so a separate budget and a higher acceptable cost per lead. Ad groups: New installation, Replace old unit, BTO / new home aircon.
- Campaign 3 — Repair (Search). Urgent, high-intent searches. Ad groups: Aircon not cold, Aircon leaking, Aircon repair.
- Campaign 4 — Brand. The company’s own name; cheap, high-converting, kept separate so it doesn’t flatter the other campaigns’ numbers.
Notice what this buys the owner: they can pour budget into installation when they want bigger jobs, protect the profitable repair campaign, see exactly which service line drives leads, and give Smart Bidding a clean signal per campaign. Every campaign also carries a tight set of negative keywords — the repair campaign excludes “DIY” and “how to,” the installation campaign excludes “servicing” — so the three don’t cannibalise each other. That’s structure doing real work.
Naming, shared settings and the plumbing
A few unglamorous habits keep an account manageable as it grows:
- Naming conventions. A consistent scheme — e.g.
[Type] | [Service] | [Region]such asSearch | Servicing | SG— makes an account instantly readable to anyone who inherits it. Names aren’t a ranking factor; they’re a sanity factor. - Account-level negative keyword lists. Terms you never want to appear for (job seekers, “free,” competitor complaints) belong in a shared list applied across campaigns, not re-typed campaign by campaign.
- Conversion tracking and auto-tagging. Set up once at account level and inherited everywhere. Without accurate conversion data, none of the structure above helps Smart Bidding — it’s the foundation of everything we do in performance marketing.
- Shared budgets and bid strategies can pool data across related campaigns when it makes sense — useful for small accounts where individual campaigns sit below the conversion volume Smart Bidding wants.
Restructuring an existing account without losing what works
Most Singapore SMEs reading this already have a messy account, not a blank slate — and the fear is that rebuilding will throw away hard-won conversion history and Smart Bidding learning. That’s a legitimate concern, so restructure deliberately rather than bulldozing. A few principles that keep the transition safe:
- Change one thing at a time. Splitting one bloated campaign into three service-line campaigns is a big change — expect a fresh learning period of one to two weeks per new campaign while Smart Bidding recalibrates, and don’t stack another major change on top mid-cycle.
- Migrate your winners first. Identify the keywords and ads that already convert and rebuild around them, rather than starting from scratch. Account-level conversion history isn’t lost when you move keywords between campaigns — the account keeps learning — but individual campaigns do need to re-accumulate signal, so give them time.
- Pause, don’t delete, at first. Keep the old structure paused for a couple of weeks so you can compare and roll back if something regresses, then archive it once the new build proves itself.
- Do it in a quieter period. Avoid restructuring in the week before a major sale — you don’t want a learning period colliding with your busiest, most expensive traffic.
Done this way, a restructure is low-risk and usually pays for itself quickly in lower cost per lead and clearer reporting. Rushed all at once during peak season, it can cost you a fortnight of unstable performance for no reason. Patience is the whole trick.
Common account structure mistakes
- One giant campaign holding every service, so you can’t control budget or read results.
- Unrelated keywords in one ad group, forcing a single generic ad that matches none of them well and tanks Quality Score.
- Over-fragmenting into SKAGs, starving each ad group of the data Smart Bidding needs.
- Not separating brand from non-brand, so cheap brand clicks flatter your averages and distort bidding.
- Mixing campaign types — letting Search, Shopping and Performance Max blur together instead of running as distinct campaigns.
- No negative keywords, so campaigns overlap and pay for irrelevant traffic.
- Building for a decade ago — hyper-granular structures that fight against, rather than feed, modern automated bidding.
The bottom line
Good account structure in 2026 is a balancing act: tight enough that every ad group has one clear theme and every campaign has one clear job, but consolidated enough that budgets are controllable and Smart Bidding has the conversion volume it needs to learn. Split campaigns by the things you want to control — objective, service line, priority, geography, brand versus non-brand — group keywords into single-theme ad groups of five to fifteen, write two or three relevant responsive search ads each, and keep the account tidy with naming conventions, account-level negatives and clean conversion tracking. Do that, and everything downstream — Quality Score, cost per lead, reporting, automated bidding — gets easier. Skip it, and no amount of clever bidding will rescue an account that’s built on sand.
Not sure whether your account structure is helping or hurting? We’ll audit how your campaigns and ad groups are organised and show you exactly where budget and relevance are leaking. Talk to us, explore our Google Ads management, or see the results in our case studies.
FAQ
What is the structure of a Google Ads account?
It’s a four-level hierarchy: the account (billing, settings, conversion tracking) contains campaigns (each with its own budget, bid strategy, location and schedule), campaigns contain ad groups (one tight theme each), and ad groups contain keywords and ads. Budget and bidding are set at the campaign level, which is why your campaign split matters most.
How many campaigns should a small business have?
Most Singapore SMEs do well with two to five campaigns, split by service line, priority or objective. Start with the fewest that give you real budget control and clear reporting, and add more only when a genuine need appears — over-splitting fragments your data and attention.
What is a STAG, and is it better than a SKAG?
A STAG (Single Theme Ad Group) groups 5–15 closely related, same-intent keywords into one ad group. In 2026 it’s preferred over the older SKAG (Single Keyword Ad Group) because close-variant matching and responsive search ads erased SKAGs’ precision advantage, while STAGs give Smart Bidding enough pooled data to learn from.
How many keywords should be in an ad group?
Roughly 5–20 closely related keywords that share one intent. Enough to cover the theme without diluting relevance. If keywords in an ad group would need different ads to be relevant, that’s a sign they belong in separate ad groups.
Should I separate brand and non-brand campaigns?
Almost always, yes. Searches for your own company name convert cheaply and behave very differently from non-brand prospecting terms. Blending them distorts your reporting and your automated bidding, so keep brand in its own campaign.
Does account structure still matter with Smart Bidding?
Yes, but its job has shifted. Structure now controls budget allocation, keeps ads relevant for Quality Score, feeds Smart Bidding enough conversion volume to learn, and keeps reporting readable. The modern aim is tight themes with consolidated data — not the hyper-granular structures of the past.


