The old advice was to segment everything. Single Keyword Ad Groups (SKAGs). Dozens of campaigns. Hyper-granular control.
That advice is outdated. Google's AI needs data to learn. Over-segmentation starves it.
Here's how to structure accounts in 2025.
Why Structure Still Matters
"If AI handles everything, why does structure matter?"
Because structure determines:
- •How you allocate budget across goals
- •What data the algorithm gets to learn from
- •How clearly you can analyze performance
- •How easily you can scale
Poor structure = wasted spend + confused algorithms + unclear insights.
The Core Principle: Consolidate for Learning
Google's Smart Bidding needs conversion data to optimize. Each campaign, each ad group needs enough volume to learn.
The math:
- •Smart Bidding needs ~30 conversions/month to function
- •Optimal performance requires ~300 conversions/month
- •Split that across 50 ad groups and most get nothing
The shift:
- •Old approach: Many small, tightly controlled segments
- •New approach: Fewer, larger segments with enough data to optimize
Consolidation isn't laziness. It's feeding the algorithm.
Campaign-Level Structure
Separate campaigns for genuinely different purposes:
By business goal:
- •Lead generation campaigns
- •Sales campaigns
- •Brand awareness campaigns
By product/service category:
- •Different margins need different ROAS targets
- •Different products need different messaging
- •Keeps reporting clean
By geography (if relevant):
- •Different regions may need different budgets
- •Local messaging and offers
- •Time zone considerations
By brand vs. non-brand:
- •These perform completely differently
- •Need separate budgets and strategies
- •Keeps brand from inflating non-brand metrics
What NOT to separate:
- •Every keyword into its own campaign
- •Minor product variations
- •Audiences that behave similarly
Ad Group Structure
Within campaigns, group by intent theme — not by exact keyword.
Good ad group themes:
- •"Running shoes for beginners"
- •"Marathon training shoes"
- •"Trail running shoes"
Bad ad group themes:
- •[running shoes] as its own ad group
- •[best running shoes] as its own ad group
- •[buy running shoes] as its own ad group
The goal: Ad groups large enough for data, specific enough for relevant messaging.
Rule of thumb: 5-15 ad groups per campaign. Each should get meaningful traffic.
Naming Conventions
Sounds boring. Saves hours.
Use a consistent format:
[Campaign Type] | [Goal] | [Product/Service] | [Geography]
Examples:
- Search | Leads | Enterprise Software | US
- PMax | Sales | Women's Shoes | National
- Brand | Awareness | Company Name | All
When you're managing 20+ campaigns, clear naming is the difference between clarity and chaos.
Performance Max Structure
PMax is a single campaign type that spans all Google inventory. But you still need structure.
Separate PMax campaigns for:
- •Product categories with different margins
- •Different conversion goals (leads vs. sales)
- •Different geographies requiring unique budgets
Asset groups within PMax:
- •Organize by product theme or audience
- •Each needs enough volume to learn
- •Don't create 50 asset groups with no data each
Guardrails to set:
- •Brand exclusions (prevent PMax from stealing brand traffic)
- •URL exclusions (control which pages receive traffic)
- •Negative keywords (now supported, 10K limit)
The Hybrid Model
Most accounts in 2025 benefit from a hybrid approach:
PMax for:
- •Broad reach across all channels
- •Products with strong feeds
- •Goals with clear conversion signals
Standard Search for:
- •Brand protection
- •High-value non-branded keywords where you want control
- •Exact match campaigns for your most important terms
Standard Video for:
- •Intentional brand building
- •Building remarketing audiences
- •Creative testing before feeding to PMax
Why hybrid: PMax is powerful but opaque. Standard campaigns give you visibility and control where it matters most.
What to Avoid
Over-segmentation:
- •SKAGs are mostly dead
- •Too many campaigns = not enough data each
- •Complexity doesn't equal performance
Under-segmentation:
- •One campaign for everything = no budget control
- •Can't optimize what you can't separate
- •Mixed goals = confused algorithms
Inconsistent naming:
- •"Campaign 1" tells you nothing
- •Makes reporting and analysis painful
- •Creates confusion when teams grow
Ignoring the data threshold:
- •Campaigns with <30 conversions/month can't optimize
- •Either consolidate or accept manual management
- •Don't expect Smart Bidding to work without data
A Starter Structure
For most accounts, start here:
Search campaigns (3-5):
- Brand (exact match, always on)
- Non-brand high-intent (your best converting terms)
- Non-brand discovery (phrase/broad for expansion)
Performance Max (1-3):
- Main products/services
- Separate campaign if margins differ significantly
Video/Demand Gen (1-2):
- Awareness/consideration
- Remarketing
Total: 5-10 campaigns for most small-to-mid accounts. Scale from there based on data, not assumptions.
Restructuring Existing Accounts
If your account is a mess:
- 1.Audit current performance — What's actually working?
- 2.Identify consolidation opportunities — Which campaigns have too little data?
- 3.Merge similar campaigns — Keep conversion history intact
- 4.Pause (don't delete) — You may want historical data
- 5.Give it time — Restructuring causes learning periods
Don't restructure everything at once. Migrate in phases.
The Bottom Line
Structure in 2025 is about balance:
- •Enough consolidation for algorithms to learn
- •Enough separation for budget control and analysis
- •Clear naming for sanity
- •Flexibility to scale
Complexity isn't sophistication. A clean, logical structure outperforms a fragmented mess every time.
This week: Count your campaigns and ad groups. How many have fewer than 30 conversions/month? Those are candidates for consolidation.
