Google Ads account structure in 2026: fewer campaigns, more sense
Splitting an account into hundreds of ad groups was right when you managed bids by hand. Today it starves the algorithm of data and you of clarity. How to build a structure that mirrors the business rather than a habit from 2015.
- Split by what you need to steer separately: budget, return target, geography and language.
- SKAGs and micro ad groups now hurt: they fragment the conversion data automated bidding learns from.
- Brand always in its own campaign. Not for performance, but so the report can be read at all.
Account structure used to be the core craft of a PPC specialist. The finer you split an account, the more precisely you could manage bids — which is why single keyword ad groups (SKAGs) were invented. That world ended the moment automated strategies took over bidding and began learning from conversion data. Fine splitting fragments that data.
At the same time this does not mean “put everything in one campaign”. It means splitting by different criteria than before.
The only rule that holds
Separate what you need to steer separately. Merge everything else.
What “steer separately” means in practice: a separate campaign is justified only where you need a different budget, a different return target, a different territory or language, or where internal reporting demands it (a different product director, a different cost centre).
Everything else — similar queries, match-type variants, fine categories with the same margin — belongs together, so the strategy has enough conversions to learn from.
What to always split
Brand and non-brand
Not for performance, but for legibility. Brand queries have a high conversion rate and a low cost, so mixed with the rest they lift the account average and hide how you are doing at winning new customers. Covered in brand vs. non-brand.
Categories with markedly different margins
When one category runs a 10 % margin and another 55 %, they cannot share a return target. In a combined campaign the budget flows to where things sell most easily — which tends to be the low-margin one.
Territories and languages
Different competition, different prices, different conversion rates. And above all: you need to see whether expansion is earning or merely diluting the average.
What not to split any more
- Match types into separate ad groups. Exact, phrase and broad in separate groups was a bid-management tool. Today it only fragments data.
- One keyword per ad group. SKAG is dead. Modern matching merges semantically close queries anyway; you only chop up your conversions.
- Devices. Separate mobile campaigns make no sense when the strategy can adjust bids by device itself.
- Every shop category. Only those that differ in margin or commercial priority. Fifty campaigns at two orders a month is worse than five at twenty.
Go through the campaigns and count conversions over the last thirty days. If most have fewer than ten, the strategies have nothing to compute from and you have nothing to evaluate. That is a signal to merge, not to tune further.
Example: an online shop
| Campaign | Role | Target |
|---|---|---|
| Brand — search | Serve people searching directly for you, hold the position | Low budget, high ROAS, evaluated separately |
| Non-brand — search, main categories | Clear purchase-intent queries, own ads and pages | Target derived from that category’s margin |
| Performance Max — high margin | Volume and discovery in the profitable range | Lower ROAS target, because margin can take it |
| Performance Max — low margin / clearance | Revenue and stock turnover | High ROAS target, strictly |
| Remarketing | Finishing an unfinished journey | Evaluate separately; platform numbers overstate |
Five to eight campaigns cover most mid-sized shops. A larger number only makes sense once each campaign gathers enough conversions.
Example: lead generation
Here the logic differs, because conversions do not share a value. The split that works:
- Brand separately, as always.
- Services with high deal value separately from cheap ones. An enquiry about a full renovation and one about changing a battery do not share the price you may pay for them.
- Urgent queries (emergency, “24/7”, “today”) separately. They have an entirely different conversion rate and a different sensible schedule.
- Informational queries either separately with a different target, or not at all. The most common source of unqualified enquiries.
More in B2B PPC: generating leads sales will actually take.
Naming and housekeeping
The boring part that pays back in a year, when someone else is digging through the account (often you):
- A consistent naming convention:
[channel] | [brand/nonbrand] | [category] | [territory]. What matters is that it is consistent, not exactly what it is. - Shared negative keyword lists instead of copies in every campaign.
- Label campaigns by margin or priority, so the report can be filtered by the business rather than by name strings.
- Log changes. Who, what, why and when it gets reviewed. Without it you cannot work out six months later what caused a drop. We built ClientDeck for this, but a spreadsheet is infinitely better than nothing.
What to take away
A good structure today does not mirror keywords or shop categories. It mirrors the economics of your business: where you have a different margin, a different budget, a different priority or a different market. Everything else belongs together, because the algorithm needs data and you need clarity. When in doubt, merge — you can always split later, once you have a reason backed by numbers.
Common questions
Do SKAGs still make sense?
No. Single keyword ad groups made sense while bids were managed by hand and needed maximum granularity. Today bids are calculated by automated strategies that learn from conversion data — and SKAGs fragment that data into pieces nothing can be derived from. Modern query matching also merges semantically close searches anyway, so isolating a single keyword is now illusory.
How many campaigns should an account have?
As many as there are things you need to steer separately — that is, with a different budget, a different return target, a different territory or different reporting. For most mid-sized online shops that works out at five to eight campaigns. A good check is to count conversions over the last thirty days: if most campaigns have fewer than ten, the account is split too finely and should be merged rather than tuned further.
Why separate brand campaigns?
Mainly so the report can be read. Brand queries have a high conversion rate and a low cost per conversion, so mixed with the rest they lift the whole account average and hide how you are doing at winning new customers. A separate campaign lets you see the non-brand result on its own and also decide separately about brand budget, whose real contribution tends to be far lower than platform attribution suggests.
- Performance Max Campaigns: What You Can Actually Control, iMark InfotechOverview of the available controls from which sensible campaign splitting follows. An agency source, not Google documentation.