Campaign Architecture Builder
Account structure sets the ceiling on everything that happens after it — what you can bid, budget, and read in a report. Set your channel, how many themes and markets you run, and how you treat brand; the builder returns a clean skeleton with names already applied.
Good structure follows one rule: one campaign for each thing you genuinely want to fund and measure separately. On Search that means brand split from non-brand (their economics have nothing in common) and one campaign per match-type cell — RGM uses exact and phrase only, leaving broad off by default for control. Within each campaign, ad groups map to tightly themed clusters. The builder applies a portable CHANNEL_TIER_GEO_MATCH naming convention so every report filters cleanly.
Campaign Architecture Builder inputs and result
| Campaign | Ad / asset group | Tier | Match |
|---|
How to use this calculator
- Pick the channel you are structuringSearch builds keyword campaigns with match-type cells; Performance Max builds asset groups; Display, Video, and Demand Gen build audience-themed ad groups. The skeleton shape changes with the channel.
- Set your themes and geographiesThemes are the tightly related product lines or intents that become ad groups. Add a geo only where you genuinely fund or measure a market on its own — needless geo splits fragment signal.
- Choose how brand is handledKeep the default to ring-fence brand in its own campaign. Brand and non-brand economics differ so completely that mixing them blurs both your bids and your reporting.
- Set the match-type policyRGM runs exact and phrase only, each in its own campaign, so you keep control of which queries spend. Choose exact-only for the tightest control on a new account.
- Copy the generated skeletonUse the tree and the table to build the account in Google Ads Editor, or drop it into your media plan. The naming convention is already applied to every campaign.
RGM Expert Says
Structure is the decision that quietly governs every later decision, so we spend real time on it before a dollar moves. The most common inherited account we see is a single non-brand campaign with broad match and forty mismatched keywords; Smart Bidding cannot learn anything useful from that soup. Splitting it into intent-tiered, match-controlled campaigns — with brand pulled out — usually lifts efficiency before we touch a single bid.
We run exact and phrase only on purpose. Broad match has improved, but it hands query control to the auction, and on most accounts that means paying for searches that were never the point. By breaking match types into their own campaigns, we can fund exact aggressively, let phrase expand under a watchful eye, and keep the whole thing legible in a report. The naming convention is part of that legibility: when every campaign is named the same way, a client can read the account at a glance.
The discipline we coach hardest is restraint on geo and theme splits. Every extra split divides your conversion signal, and thin signal is the enemy of automated bidding. We only break out a geo or a theme when the client will actually fund or judge it separately — otherwise it is structure for its own sake, and it costs you learning.
How it works
The builder encodes a structural rule used across well-run accounts: create one campaign per cell you want to fund and measure independently, then map tightly themed clusters to ad groups inside each. On Search the cells are tier (brand vs non-brand) x geo x match type; on other channels they are tier x geo, with themes as audience-scoped ad groups.
- Channel — sets whether cells are match-type (Search) or audience/asset based.
- Themes — tightly related clusters; each becomes an ad group or asset group.
- Geos — markets funded or measured separately; multiply campaign count.
- Brand handling — brand is always its own campaign so it is never starved.
- Match policy — exact and phrase only by default; broad omitted for control.
Match-type and structure guidance reflect RGM’s house rule (exact + phrase, broad avoided) and Google Ads Help: about keyword match types. The naming convention is a portable RGM convention, not a platform requirement.
Why structure decides what the rest of the account can do
How you organize an account sets the ceiling on everything Smart Bidding can learn. Lump high-intent buyers and casual researchers into one campaign and you blur the signal, hand the algorithm a muddy goal, and lose the ability to fund them differently. Split them cleanly and every part of the account can be bid, budgeted, and measured on its own terms.
The single most valuable split is brand versus non-brand. Brand terms convert cheaply because the demand already exists; non-brand has to create it. Funding them from one budget lets brand soak up spend that should be buying new demand, or lets non-brand bids inflate the price of clicks you would have won for pennies. Separate campaigns end that argument.
Match type is the other lever structure unlocks. RGM runs exact and phrase only, each in its own campaign, so you control which queries spend and can read performance by match type at a glance. Broad match can work, but only as a deliberately small, fenced learn tier — never as the default that quietly absorbs the budget.
Structural rules of thumb
Practical defaults RGM uses when building a Search account. These are conventions, not platform rules.
| Decision | RGM default | Why |
|---|---|---|
| Brand terms | Own campaign, always | Cheap, must not be starved |
| Match types | Exact + Phrase, separate campaigns | Query control + clean reporting |
| Broad match | Omitted by default | Avoid unless a fenced learn tier |
| Geo split | Only where funded separately | Preserve conversion signal |
| Ad groups | One per tight theme | Tight ad-to-query relevance |
What practitioners say about structure
The cleanest accounts win not because of clever bids but because every campaign means one thing, so the algorithm and the operator both know what they are looking at.
Brand and non-brand belong in different campaigns for the same reason savings and rent belong in different accounts — mixing them hides the truth.