How to Structure a Search Campaign (Ad Groups, RSAs, Settings) — the foundation an audit checks and a budget plan assumes is sound

A Search campaign is more than keywords and a budget: how you group keywords into ad groups, what you leave in the RSA, and which settings you never revisit (network, location targeting, ad rotation) all determine whether the account has a shot at working before anyone starts optimizing it. This is the checklist to get that foundation right at setup, in the order Google's own documentation frames it.

Group Keywords Into Tightly Themed Ad Groups

For each ad group, pick a narrow theme and use only keywords related to that theme — support.google.com/google-ads/answer/6372655. Google explicitly recommends organizing ad groups around your website's structure, so each ad group maps to a different page or product category — same source. Ads should mention at least one keyword from the ad group in the headline, since that increases perceived relevance — same source. The grouping principle is topic/intent, not match type: Google's own guidance emphasizes thematic structure over grouping by match type alone — support.google.com/google-ads/answer/6372655. See Keyword Match Types in 2026: Broad, Phrase, Exact — and Why None of Them Is Literal Anymore for how match type behaves once keywords are grouped, and Account, Campaign, Ad Group Hierarchy — and Why Getting It Wrong Burns Budget for where ad-group-level settings sit versus campaign-level ones.

Do not reorganize a live account casually

restructuring after you've already created several campaigns and ad groups wipes out all the accumulated performance data, which can hurt how ads perform — support.google.com/google-ads/answer/6372655. Get the theme grouping right at launch; it's expensive to redo later.

Write Responsive Search Ads That Compete on Ad Strength

A Responsive Search Ad (RSA) can hold up to 15 headlines and 4 descriptions — support.google.com/google-ads/answer/6167122. Google's own guidance is to add as many unique, high-quality assets as possible for optimal performance — same source. Best practice is to implement at least 2 RSAs per ad group rated "Good" or "Excellent" Ad Strength — same source; advertisers who move Ad Strength from "Poor" to "Excellent" see roughly 15% more clicks and conversions on average — same source.

Composition that works

headlines of varying length (to fit different placements and audiences) while keeping messaging consistent across devices; keywords integrated naturally into the copy, since users engage more with ads that look relevant to their search (e.g. a headline "Buy Digital Cameras" if that's one of the ad group's keywords); specific calls-to-action ("Call now for a free quote") instead of generic ones, which show decreased engagement; messaging on user benefits, not just features; at least 4 unique image assets; and business logos, location/phone assets, and sitelinks as part of the asset set — all support.google.com/google-ads/answer/6167122. Full asset taxonomy is in Ad Assets (Sitelinks, Callouts, Asset Groups) — the free lever that moves Ad Rank without raising the bid.

Pin only when you have to

pinning headlines or descriptions to fixed positions limits the combinations the RSA can test and reduces flexibility — support.google.com/google-ads/answer/6167122. Only pin when there's a specific messaging requirement (e.g. a legal disclaimer); avoid pinning on general campaigns — same source.

Campaign Settings You Will Forget to Check

These live at campaign level and default to values that are easy to leave unexamined — see Account, Campaign, Ad Group Hierarchy — and Why Getting It Wrong Burns Budget for why they can't be set per ad group.

Location targeting — Presence vs Presence or Interest

the default and Google-recommended setting is "Presence or Interest," which shows ads to people likely to be in the targeted locations *and* to people who have shown interest in them (detected via Google Maps searches, custom location settings, and past visit patterns) — support.google.com/google-ads/answer/1722038. "Presence" is the more restrictive option: it targets only people likely to be located or regularly located in the targeted locations — same source. Advertisers who switched from Presence to Presence or Interest in the Travel, Real Estate, and Education verticals reported +5% more conversions on Search — same source. Presence targeting is preferable for sensitive verticals with strict targeting limitations — same source.

Search partners are on by default

new Search campaigns include Google search partners automatically — support.google.com/google-ads/answer/1722047. These are third-party sites where ads can appear in search results, site directories, product pages, and YouTube search — same source. Clicks from search partner sites may not always reflect highly targeted traffic, so weigh this against your campaign goals before leaving it on — same source. Opt out any time from campaign settings, via the Networks dropdown, by unchecking "Include Google search partners" — same source.

Ad rotation — Optimize vs Do Not Optimize

"Optimize" optimizes ads for clicks in each individual auction, using signals like keywords, search terms, devices, and locations; as performance data accumulates, better-performing ads get more impressions — support.google.com/google-ads/answer/112876. "Do Not Optimize" delivers ads more evenly over time without prioritizing high performers, and Google does not recommend it for most advertisers, since ad quality still affects placement — same source.

Set an Initial Budget and Bid Strategy Matched to Available Conversion Data

Most new advertisers use automated bidding for its effectiveness — support.google.com/google-ads/answer/6242914. The two primary automated strategies are "Maximize Conversions" (driving sales or leads) and "Maximize Clicks" (driving traffic) — same source. The campaign's average daily budget is the cap on overall spend — same source.

Don't switch to Maximize Conversions too early

Google's recommended minimum is at least 15 conversions in the last 30 days before applying the Maximize Conversions strategy — support.google.com/google-ads/answer/7381968. This threshold matters at structure time because ad group granularity and campaign count directly determine how conversion volume gets split — over-fragmenting a new account starves every individual campaign of the data a Smart Bidding strategy needs. See Bidding Strategies: Manual, Smart Bidding, and When Each Applies — matching the strategy to the conversion data you actually have and How to Set Budgets and Bid Strategies by Funnel Stage (Without Starving Smart Bidding of Data) for the budget-sizing math this feeds.

Even with little or no conversion data, Smart Bidding can still use query-level signals to build more accurate initial conversion-rate models, and it can draw on data from all of a marketer's campaigns — so a brand-new campaign in an account with existing history isn't starting from zero — support.google.com/google-ads/answer/10970825.

Negative Keywords: Campaign-Level Lists vs Account-Level

Account-level negative keywords apply from one place across all relevant campaign types — Search, Performance Max, App, Shopping, Smart, and Local — up to 1,000 per account — support.google.com/google-ads/answer/11396330. They're the right tool for terms that should never trigger an ad anywhere in the account, such as competitor names or job-seeker indicators — same source. Campaign-level negatives apply to a single campaign; you can add up to 5,000 per list and create up to 20 negative keyword lists per account — support.google.com/google-ads/answer/7102995. Both levels support broad, phrase, or exact match type per negative keyword — support.google.com/google-ads/answer/11396330. In manager (MCC) accounts, shared negative keyword lists from the shared library can be applied across multiple client accounts at once — support.google.com/google-ads/answer/7519927. Decide the level at setup — moving exclusions later is straightforward, but a campaign that launches with zero negatives at either level is exposed from the first auction. See How to Do Keyword Research and Build a Negative Keyword List — closing the leak Smart Bidding can't see for how the list gets populated on an ongoing basis.

Avoid Policy-Triggering Patterns at Launch

A handful of editorial rules cause disapprovals often enough to check before the first ad goes live: correct capitalization is required (incorrect use is a policy violation); spelling and grammar must be commonly accepted and comprehensible; punctuation and symbols must be used correctly, with no gimmicky or excessive use; non-standard or unnecessary repetition of names/words/phrases is barred; business names must reflect the actual domain, advertiser name, or app, with no promotional language; and phone numbers cannot appear in ad text (they have a dedicated asset field instead) — all support.google.com/adspolicy/answer/6021546. Every ad must clearly identify what's being promoted, and the destination page must offer unique value and be functional, useful, and easy to navigate — same source. The full policy taxonomy behind these rules, plus the restricted-category and suspension side, is in How to Write Ads That Don't Get Disapproved (or Worse) — the checks that separate a fixable rejection from an account-ending one and Google Ads Policy Taxonomy: What Gets an Ad Disapproved vs an Account Suspended.

What does NOT work

Reorganizing a live account's ad groups after it has accumulated data — the restructuring itself wipes the performance history that made the account worth optimizing — support.google.com/google-ads/answer/6372655. Grouping keywords by match type instead of topic/intent, which runs against Google's own thematic-structure guidance — support.google.com/google-ads/answer/6372655. Pinning RSA assets without a specific messaging requirement — it caps the combinations the ad can test for no benefit — support.google.com/google-ads/answer/6167122. Leaving "Do Not Optimize" ad rotation on for a typical account — Google itself says this isn't recommended for most advertisers — support.google.com/google-ads/answer/112876. And switching to Maximize Conversions before the account has the recommended 15 conversions in 30 days — the strategy is built to need that baseline — support.google.com/google-ads/answer/7381968.

Why this matters for the objective

this page is the setup layer everything downstream depends on — How to Audit an Existing Google Ads Account (Priority Order) — the sequence that stops you from trusting broken data checks these exact settings for drift, and a broad match keyword or a Smart Bidding switch made against a badly-grouped, data-starved account structure is how budget gets burned before any optimization even starts.

Related

- Account, Campaign, Ad Group Hierarchy — and Why Getting It Wrong Burns Budget — the account → campaign → ad group hierarchy that explains why network, location, and budget settings live at campaign level and can't be set per ad group. - Keyword Match Types in 2026: Broad, Phrase, Exact — and Why None of Them Is Literal Anymore — how match type behaves inside the themed ad groups this page builds, and why it isn't the grouping axis itself. - Ad Assets (Sitelinks, Callouts, Asset Groups) — the free lever that moves Ad Rank without raising the bid — the full asset taxonomy (sitelinks, callouts, images) that RSA composition here only partially covers. - Bidding Strategies: Manual, Smart Bidding, and When Each Applies — matching the strategy to the conversion data you actually have — the bid-strategy detail behind the Maximize Conversions/Clicks choice made at setup. - How to Do Keyword Research and Build a Negative Keyword List — closing the leak Smart Bidding can't see — the ongoing workflow that populates the negative keyword lists this page says to decide the level for. - How to Write Ads That Don't Get Disapproved (or Worse) — the checks that separate a fixable rejection from an account-ending one — the full editorial and restricted-category ruleset behind the launch-time policy check here. - How to Audit an Existing Google Ads Account (Priority Order) — the sequence that stops you from trusting broken data — the checklist that later verifies whether these same settings (network opt-ins, budget, bid strategy) drifted from intent. - What Does NOT Work: Broad Match and Smart Bidding Without Guardrails — where automation burns budget instead of saving it — what happens when a structure built here is paired with broad match and Smart Bidding on too little conversion data.