How to structure a Google Ads account that stays manageable

Three hard constraints force structure. Everything else is preference.

1. Budget lives at the campaign level. If you need to control spend on something separately, it needs its own campaign.
2. Most targeting settings live at the campaign level: bid strategy, locations, languages, networks, ad schedule, campaign type.
3. Reporting rolls up the hierarchy. If you cannot filter to a question, you cannot answer it without manual reconstruction.

The old craft of hyper-segmentation was a rational response to manual bidding. Smart Bidding changes the trade. Splitting traffic into many small campaigns also splits the conversion data the bidding model learns from, and thin campaigns learn slowly or not at all. The working principle now: consolidate for learning, separate for control.

Before you touch the account, write down the questions it has to answer every month. Typical ones:

• What did brand cost compared with non-brand?
• Which service line or product category returns best?
• How do newer markets perform against the core one?
• How much went to capturing existing demand versus creating new demand?

Any question that must be answerable without a spreadsheet rebuild becomes a structural line. Every other question becomes a label, a saved report or a segment. Labels are far cheaper than campaigns: they cost you nothing in learning data and they can be changed without resetting history.

1. Split brand from non-brand. Always. Then apply brand terms as a negative list to the non-brand campaigns and non-brand terms as a negative list to brand, so the two stop stealing each other's queries.
2. Split by campaign type, because the platform makes you. Search, Shopping or Performance Max, Demand Gen, Video and App each have a different job and a different cost profile. Never judge them against one blended target.
3. Split non-brand by commercial unit, not by keyword count. For ecommerce that usually means margin tier or category. For services it means service line. The test is simple: would you willingly move budget between these two things? If no, they need separate budgets.
4. Split geography only when something real differs. Different currency, language, seasonality, or a budget you must protect. Otherwise one campaign with multiple locations and location bid adjustments is easier to run.
5. Keep a handful of ad groups per campaign, each with one landing page and one coherent message.
6. Give broad, top-of-funnel discovery its own campaign and its own budget. Shared with a converting campaign, it will quietly absorb the money.
7. Keep competitor targeting separate. It never earns the same efficiency target as your own brand or category terms.

Performance Max does not sit politely alongside the rest of the account, so decide its boundaries explicitly.

• For the same product in the same account, Performance Max takes priority over Standard Shopping. Pick which campaign owns which products and write it down.
• Performance Max will serve brand queries unless you exclude them. Brand-heavy Performance Max looks excellent and is largely harvesting demand you already had. Apply brand exclusions if you want an honest read.
• Within a single account, only one of your ads is eligible for a given auction. Overlapping campaigns therefore do not double your cost, but they do make reporting unstable because the winning campaign varies by auction.
• Segment inside Performance Max with asset groups and listing group filters first. Add a second campaign only when you need a separate budget, for example to protect a high-margin range.

Pick one pattern and never deviate:

[Type] | [Market] | [Brand/NonBrand] | [Unit] | [Theme]

• Search | UK | NonBrand | Enterprise | Integrations
• PMax | GLOBAL | NonBrand | HighMargin
• Search | AE | Brand | Core | Exact

Rules that save you later: one separator character, a fixed token order, ISO country codes, no spaces inside a token, and no dates in names. Dates belong on labels, because labels can be reassigned and names carry history. If you must rename a campaign with meaningful history, record the change somewhere the next person will find it.

• Negative keyword lists. Keep an always-negative list for universal junk, a brand list applied to non-brand campaigns, and a non-brand list applied to brand campaigns. Three lists, maintained once, applied everywhere.
• Account-level negative keywords for terms that are irrelevant to the entire business.
• Conversion actions defined once, then selected per campaign using campaign-level conversion goals. This is what lets a lead-gen campaign optimize to qualified leads while an ecommerce campaign optimizes to purchases, inside one account.
• Audience lists built once and referenced, not rebuilt per campaign.

The last one deserves emphasiz. Editing a live campaign repeatedly leaves you with performance history that is an average of three different strategies. When the change is significant, build the new version alongside, run both, then retire the old one.

• Weekly: search terms report into negatives, budget pacing, disapprovals and policy flags, overlap check on anything new.
• Fortnightly: asset and ad performance, ad group theme drift, landing page availability.
• Monthly: structure against the question list. Retire campaigns that no longer have a job.
• Quarterly: naming audit, conversion action audit, negative list review for terms that are now relevant.

Structure is not a one-time build. It is a thing you prune. The accounts that stay manageable are the ones where someone removes a campaign roughly as often as they add one.

Structure work usually lands in the first two weeks, after measurement is verified and before budget increases. There is no point scaling spend through a structure that cannot report on where it went. We treat it as part of performance marketing setup: agree the reporting questions, rebuild once against those questions, then run. Scope and cost depend on how many markets and campaign types are in play, which is why pricing is built around the work rather than a package.