Google Search campaign structure

The three structural decisions

Every Search account makes three structural decisions, explicitly or by accident:

1. How to split branded from non-branded. Always separate. Branded gets its own campaign, non-branded gets its own campaign(s), and they use different bid strategies and budgets.
2. How to tier non-branded by intent. High-intent ("buy [product] online") vs research-intent ("best [product] reviews") vs awareness ("what is [category]"). Each tier wants different bidding, different ad copy, and different landing pages.
3. How granular to make ad groups. Modern Google (2026) rewards broader ad groups than the old SKAG (single keyword ad group) era. The right granularity is theme-level, not keyword-level.

Branded vs non-branded: always separate

Branded campaigns capture demand that already exists; non-branded campaigns create or capture demand that's open to multiple providers. Mixing them in one campaign is the most common Search structure error.

	Branded	Non-branded
Intent level	Highest — searcher knows your brand	Variable — depends on tier
Cost per click	Usually $0.50-$3	Often $3-$30+
Conversion rate	Often 10-30%	Often 2-8%
Bid strategy	Target Impression Share (top of page)	Target CPA or Target ROAS
Budget priority	Defend the brand at any cost	Maximize within CPA/ROAS target
Ad copy	Brand reinforcement, sitelinks	Differentiation, social proof
Landing page	Homepage or branded campaign LP	Category or product LP

The reason for separation isn't reporting; it's bidding. Mixing branded and non-branded under one Target CPA strategy averages the two CPAs, which under-bids branded (where you should be at top-of-page no matter what) and over-bids non-branded (which has worse intent and shouldn't get the same bid).

Intent tiering for non-branded

Non-branded search splits into three intent tiers. Each tier needs its own campaign.

Tier	Query pattern	Example	Treatment
High-intent (commercial)	"buy", "price", "near me", comparison, brand+product	"running shoes for plantar fasciitis"	Aggressive bidding, product landing pages, retargeting pixel on every visitor
Mid-intent (research)	"best", "reviews", "vs", "how to choose"	"best running shoes for plantar fasciitis"	Comparison/review landing pages, modest bidding, follow-up nurture
Low-intent (awareness)	"what is", "why does", problem-symptom queries	"why do my heels hurt when I run"	Educational content, lowest bids, focus on email capture not direct conversion

The intent tiers shouldn't be in the same campaign because they have wildly different conversion rates and average values. A high-intent campaign with a $30 target CPA should not share budget with a research-tier campaign whose CPA is naturally $80+. Separate campaigns let each tier run its own bid economics.

Ad group granularity: theme-level, not keyword-level

The SKAG era (one keyword per ad group) ended around 2020 when Google's machine learning got good enough that ultra-granular ad groups hurt performance — they fragment the data and prevent the algorithm from learning. Modern best practice is theme-level ad groups: 5-30 closely related keywords per ad group, sharing intent and ad copy.

For a running-shoes brand, themes might be: trail-running shoes, road-running shoes, racing flats, beginner shoes, plantar-fasciitis shoes, wide-foot shoes. Each theme becomes one ad group with all its keyword variants.

Match types: broad + exact, no phrase

Google's match types have evolved. Current best practice for most accounts:

• Exact match. Use for proven-performing keywords. Bid more aggressively. Captures intent precisely.
• Broad match. Use with Smart Bidding (tCPA or tROAS) for discovery of new variants. Without Smart Bidding, broad match is unmanageable.
• Phrase match. Largely deprecated since the 2021 phrase/broad-modified merger. Most accounts can skip it.

The 2026 pattern: 60-70% of spend on exact match for proven keywords, 30-40% on broad match for discovery. Pull search-term reports weekly to harvest new exact-match candidates from broad-match queries.

Negative keywords: the rigor that compounds

Every campaign needs:

1. A negative keyword list of competitor brand names (unless conquesting is part of the strategy).
2. A negative keyword list of irrelevant intent terms ("jobs", "free", "download", "DIY", "how to make").
3. Mutually-exclusive negative keywords between campaigns — if you have a high-intent campaign and a research campaign, the research keywords are negatives in the high-intent campaign and vice versa.
4. A standing weekly review of search-term reports for new negative candidates.

Mature accounts maintain negative keyword lists with thousands of entries. The compounding effect is real — every negative added prevents a small percentage of future budget waste.

Bidding strategies by campaign type

Campaign type	Recommended bid strategy
Branded	Target Impression Share (95-100% on top of page)
High-intent non-branded	Target ROAS or Maximize Conversion Value
Mid-intent research	Target CPA at a softer threshold
Low-intent awareness	Maximize Clicks within budget, or Manual CPC with hard caps
Competitor conquest	Manual CPC with strict caps — automated bidding overpays on competitor terms

The structural template that works for most accounts

1. Brand — one campaign, defend at top-of-page.
2. Competitor conquest — one campaign, manual CPC, separate budget.
3. High-intent non-branded — one or more campaigns, segmented by product line, tROAS bidding.
4. Research / mid-intent — one campaign, tCPA bidding.
5. Awareness / low-intent — one campaign, Max Clicks bidding, content/educational landing pages.
6. Dynamic Search Ads — one campaign for keyword discovery, broad-bid, capture queries you didn't think of.

Do I still need SKAGs?

No. Google's machine learning works better with broader ad groups (5-30 keywords). SKAGs fragment the data and prevent the algorithm from learning. The exception is when you have wildly different landing pages per keyword — but the answer there is usually different ad groups for different intents, not different ad groups per keyword.

How many campaigns is too many?

If you can't articulate why each campaign exists, you have too many. A well-structured account for a mid-sized advertiser typically has 8-15 campaigns; very large accounts may have 50+ but with clear taxonomies. Most accounts have far more than they need because campaigns were added by accident rather than design.

What's the deal with broad match in 2026?

Broad match combined with Smart Bidding (tCPA or tROAS) is now the recommended way to discover new query variants. Broad without Smart Bidding is still wasteful. Run broad-match keywords in their own ad groups with smart bidding turned on, then harvest the high-performing search terms into exact-match campaigns.

How often should I review search terms?

Weekly minimum. Each review: add new negatives for wasted spend, add new exact-match keywords for high-performing broad terms, and watch for theme shifts that indicate audience or competitive changes.

Should Search and Performance Max share keywords?

No. Brand keywords should be excluded from PMax via brand exclusions. Non-branded keywords work in both — PMax discovers them via broad-match-like discovery, Search captures them with exact-match precision. The two campaign types serve different functions: PMax for breadth across surfaces, Search for precision on known-converting queries.

How do I handle long-tail keywords?

Two patterns. One: broad-match with Smart Bidding to capture the long tail dynamically. Two: Dynamic Search Ads (DSA) that pull headlines and landing pages from your site automatically. Both work. Manual long-tail keyword research has diminishing returns vs the automated approaches.

Operating checklist

1. Define the conversion event and value model before campaign creation.
2. Onboard first-party data via Customer Match and offline conversion imports.
3. Set the bid strategy and target based on actual margin, not aspiration.
4. Validate creative requirements per ad type before launching.
5. Monitor first 7 days closely; resist the urge to edit before 14 days.
6. Pull search-term and asset-group reports weekly.
7. Document the campaign architecture in a runbook for the next operator.