Most Google Ads accounts are structured like junk drawers — campaigns accumulated over years, ad groups with 200 unrelated keywords, names like "Campaign 3 (new)". Structure sounds boring, but it's the skeleton everything hangs on: budgets flow through campaigns, targeting lives in ad groups, and Google's automation learns per campaign. Get the skeleton wrong and every optimisation after it underperforms. This guide covers the architecture patterns that work in 2026, the naming conventions that keep accounts manageable, and how to restructure without torching performance.
Why structure matters more than ever
Three reasons. Budget control: budgets are set at campaign level, so campaigns are how you decide where money goes. One campaign per budget decision. Learning: Smart Bidding learns within a campaign — fragment spend across dozens of tiny campaigns and nothing accumulates enough data to optimise. Reporting: clean structure means you can see what's working at a glance instead of excavating it.
The standard pattern
The hierarchy is fixed — Account → Campaign → Ad Group → Ads/Keywords — but how you divide things is up to you:
- Campaigns divide by budget and goal. Brand, generic services, competitor, remarketing, shopping — each gets its own campaign because each gets its own budget and bid strategy.
- Ad groups divide by theme. Keywords and ads grouped by intent: one ad group for "boiler repair", another for "boiler installation", another for "boiler service". Each ad group gets ads written specifically for its theme — this is what drives Quality Score.
- Ads and keywords live at the bottom. 5–15 tightly themed keywords per ad group (fewer is fine), 2–3 responsive search ads with genuinely different angles.
Segmentation patterns (pick what fits your business)
- By intent: separate campaigns for emergency/high-intent vs research-phase keywords. Different intent deserves different budgets, bids and ad copy.
- By geography: separate campaigns per region or city when performance, competition or messaging differs by area. Essential for multi-location businesses.
- By margin: your high-margin services in their own campaign with aggressive budgets; low-margin work contained separately so it can't eat the budget.
- By funnel stage: prospecting campaigns (broad reach, controlled with negatives) separate from remarketing (past visitors, higher bids justified by higher conversion rates).
- By network: Search and Display never share a campaign. Different user intent, different economics, different creative — mixing them means the cheaper network cannibalises the budget.
Don't segment on all five axes at once — that way lies 200 campaigns with £2/day each and no learning anywhere. Segment on the one or two axes that actually change decisions, and no more.
Naming conventions (boring, essential)
Anyone should be able to read a campaign name and know what it is: Search | Manchester | Boiler Repair | Exact, PMax | All Products | Prospecting. Pick a pattern — Network | Geography | Theme | Purpose — and enforce it. Future you, auditing the account at 11pm, will be grateful. Same for ad groups: thematic names, not "Ad group 1".
What died: SKAGs and hyper-granularity
Single-keyword ad groups (SKAGs) were the fashion a few years ago — one keyword per ad group for maximum control. In the Smart Bidding era they're counterproductive: they fragment conversion data so the algorithm can't learn, and modern match types plus responsive ads deliver the relevance SKAGs were built for. Consolidate. The 2026 rule of thumb: fewer, larger, thematically clean.
Shopping and Performance Max: structure notes
Product-led accounts need a different shape. Shopping campaigns structure around product tiers, not keywords: split campaigns by margin or priority (hero products vs long tail), by product category, or by custom labels (bestsellers, high margin, clearance). Your product feed is your targeting — feed quality (titles, images, categorisation, pricing accuracy) matters more than campaign settings. Performance Max for retail follows the same logic: separate asset groups per product category with tailored creative, brand exclusions via negative lists, and new-customer acquisition goals if you're prospecting rather than harvesting.
Account-level settings people forget
- Negative keyword lists shared across campaigns — build one master list (jobs, free, DIY, complaints, competitors-you-don't-want) and attach it everywhere. Update it monthly from search-term reports.
- Location options: "Presence" vs "Presence or interest" — for local businesses, Presence (people physically in your area) is almost always right. The default includes "interest", which shows your Manchester ads to someone in London researching Manchester. Check this on every campaign.
- Ad scheduling: align with when you can actually respond. Ads at 3am for a business that answers phones 9–5 buy clicks that go cold by morning — unless your forms and follow-up are genuinely 24/7.
- Conversion settings: one primary conversion per campaign goal, values where possible, and a sensible attribution window. Review quarterly as the business changes.
The structure audit: scoring your current account in 30 minutes
Before rebuilding, score what you have. Open the account and check: campaign count vs budget — more than one campaign per £300/month of spend is usually fragmentation; ad group themes — can you describe each ad group's theme in three words? If not, it's not a theme; keyword-to-ad relevance — does every ad mention the ad group's core terms? network mixing — any campaign running Search + Display together gets flagged immediately; naming — can a stranger understand the account from campaign names alone? Score each red flag, fix the structural ones before touching bids or budgets. Most accounts we audit fail at least three of these — and fixing them lifts performance before a single bid changes.
Restructuring without breaking things
Rebuilding a live account is surgery — do it carefully:
- Never restructure everything at once. One campaign at a time, starting with the worst performers.
- Keep what's working. A campaign with strong history and good performance gets consolidated into, not deleted. History has value.
- Move, don't duplicate. Pause the old structure as the new one proves itself; run a 2-week overlap, not a 2-month one.
- Expect a learning dip. New campaign structures reset Smart Bidding learning. Budget for 1–2 weeks of wobble and don't panic-tweak during it.
- Document the map. Old → new campaign mapping, so anyone can trace performance history.
Account structure is the first thing we fix in most audits — it's also the cheapest win. Our PPC advertising team rebuilds accounts around clean architecture and consolidated learning. Planning the budgets to flow through it? Read our PPC budget planning guide. Get a free consultation.
