The Complete Guide to Google Ads Account Structure in 2026

Most Google Ads accounts you inherit are a mess. Forty-three campaigns with names like “Test 3 - FINAL v2”, ad groups that haven’t been touched since 2022, and conversion actions that double-count purchases against leads. The advertiser is spending money — but no one can confidently say what’s working.

A clean account structure isn’t a vanity exercise. It is the single biggest determinant of whether Google’s machine-learning bidding will work for you. Smart Bidding optimizes against signals; if your structure muddles those signals, the algorithm muddles its bids back at you. Get the structure right, and you can spend 40% less on a clean account than on a messy one with identical creative.

This guide is the blueprint we use to architect new accounts at Digitelia, whether the client is a $5K/month SMB, a B2B SaaS company running ABM-style campaigns, or an ecommerce brand pushing $5M through Performance Max.

Why account structure matters more than it used to

In the manual-bidding era — pre-2018, basically — structure existed to give the advertiser granular control. You wanted a separate ad group for every match type and every keyword variant so you could bid $1.85 on “blue widgets exact” and $0.92 on “widgets phrase”.

That logic is dead. Smart Bidding sets bids per query, and it ignores ad-group bids unless you override them. The reason to structure your account in 2026 is different and arguably more important: structure controls what signals the algorithm sees, and it controls what data you can actually report on.

If you bundle B2B leads and ecommerce purchases into one Performance Max campaign with one conversion goal, you’ve blinded the bidder. It will optimize for the cheaper, higher-volume conversion (probably the lead) and starve the higher-value one (the purchase). You will then spend three months wondering why ROAS dropped.

A well-structured account separates intent, product line, margin, and geography clearly enough for the algorithm to optimize each on its own terms — without splitting volume so thin that the bidder can’t gather signal at all. That balance is the art.

The 2026 hierarchy: account → campaign → asset group / ad group → asset → audience signal

Google’s hierarchy has subtly shifted with Performance Max. Here is the modern mental model:

Manager (MCC)
└── Account (one per legal entity, brand, or domain)
    ├── Conversion goals (account-level)
    ├── Audiences (account-level remarketing + lookalikes)
    └── Campaigns
        ├── Performance Max
        │   └── Asset groups (themed by product/audience)
        │       ├── Assets (text, image, video, logo)
        │       └── Audience signals
        ├── Search
        │   └── Ad groups (themed by intent)
        │       ├── Keywords (broad + exact)
        │       └── Responsive Search Ads
        ├── Shopping (standard, if still applicable)
        └── Demand Gen (replacing Discovery)

Below the account, the most consequential decisions are: how many campaigns, how many asset groups per Performance Max campaign, and what conversion goal each campaign reports against.

The three structural archetypes

Every account fits one of three archetypes, depending on what you sell. Picking the right archetype is the most important decision you’ll make.

Archetype 1: SMB service business (plumbers, dentists, law firms, agencies)

You have one offer, one geographic radius, and one funnel: search ad → lead form → phone call. Your structure should be brutally simple.

• One campaign: Search, with broad match keywords and Maximize Conversions bidding (target CPA after 50 conversions).
• 3-5 ad groups: themed by service line if you have multiple (“emergency-plumbing”, “drain-cleaning”, “water-heaters”), or by intent if you don’t (“commercial-keywords”, “branded-keywords”, “competitor-keywords”).
• One Performance Max campaign: for remarketing and brand defence, with a Customer Match list as audience signal.
• One conversion goal: phone calls + form submits, weighted by your closed-won rate.

Total: 2 campaigns. Resist the temptation to make more.

Archetype 2: B2B SaaS (lead gen for high-value, long-cycle sales)

Your conversion is a demo request or trial signup. The closed-won is 30-180 days away. Your structure prioritizes audience signals over keywords, because B2B intent often shows up in non-obvious queries.

• Search campaign: 4-8 ad groups by buyer-intent stage (problem-aware, solution-aware, vendor-aware, branded, competitor).
• Performance Max campaign: separate one per ICP segment if your TAM is large enough (e.g., one for “fintech CFOs”, one for “ecommerce ops leaders”). Audience signals: Customer Match lists of past leads + lookalikes.
• Demand Gen campaign: top-of-funnel awareness, optimized for engaged-view or site visit.
• Conversion goals: demo request (primary, value-tracked), trial signup (secondary), MQL conversion fired server-side from CRM.

Total: 3-6 campaigns depending on segment count.

Archetype 3: Ecommerce (transactional, multi-SKU)

Your conversion is a purchase. ROAS is the only metric that matters. Performance Max with asset groups themed by margin or product category will dominate spend.

• Performance Max campaigns: one per product category or margin tier. High-margin best-sellers in their own campaign with aggressive target ROAS; clearance in another with looser ROAS to move inventory.
• Standard Shopping campaign: optional, for products that need finer feed control or where Performance Max underperforms.
• Search campaign: branded keywords (always carve these out so you can defend the brand cheaply) and one “non-brand head terms” campaign.
• Conversion goals: purchase value (primary), add-to-cart (secondary, for audience building).

Total: 4-10 campaigns depending on catalog size.

Naming conventions that survive turnover

A name should tell you, without opening the campaign, what it does. Use a pipe-delimited convention so columns parse cleanly in Looker Studio:

{Type} | {Funnel} | {Audience or Product} | {Geo} | {Variant}

Examples:

• PMax | Bottom | Best-Sellers | US-CA | v3
• Search | Mid | Solution-Aware | Global | v1
• DGen | Top | Fintech-ICP | EMEA | v2

Why this works: when you export the campaign report to Looker Studio, a split-by-delimiter formula gives you five clean dimensions to slice by. You can now answer “what’s our blended CPA across all bottom-funnel campaigns?” with a single filter.

Equivalent ad-group naming:

{Intent stage} | {Theme} | {Match type or signal}

• Mid | Pricing-Queries | Broad
• Bottom | Brand-Defense | Exact

Conversion goals: the single most underrated lever

Conversion goal setup is where 80% of accounts leak performance. Three rules:

1. One primary, multiple secondaries. Smart Bidding can optimize against only one primary goal per campaign. Pick the goal closest to revenue: for ecommerce that’s purchase value; for SaaS it’s qualified lead value; for SMB it’s call-form weighted. Everything else is a secondary goal — useful for reporting, invisible to the bidder.

2. Value-weight every conversion. Don’t bid against “lead = $1, purchase = $1”. Tell Google what each is actually worth. For SaaS, use a closed-won probability × ACV calculation. For ecommerce, use cart value (or LTV if your bidder will respect it). Without values, the algorithm optimizes for volume — which is rarely what you want.

3. Enhanced Conversions are non-negotiable. With third-party cookies effectively dead, Enhanced Conversions for Web (hashed email/phone passed at conversion) recovers 7-15% of conversion volume on average. Enhanced Conversions for Leads (which require offline conversion upload from your CRM) is mandatory for SaaS. If you’re not running both, your Smart Bidding is flying half-blind.

// Example: passing hashed email for Enhanced Conversions
gtag('event', 'conversion', {
  'send_to': 'AW-XXXXXXXXX/AbC-D_efG-h12_34-567',
  'value': 47.50,
  'currency': 'USD',
  'transaction_id': 'order_8821',
  'email': 'user@example.com',          // hashed automatically
  'phone_number': '+15555550100'
});

Audience signals: feeding the algorithm

Performance Max and Demand Gen rely on audience signals — your hint to the algorithm about who matters. They are not strict targeting; they bias the model.

The hierarchy of signal quality, from best to worst:

1. Your first-party data (Customer Match: emails of past customers, MQLs, demo-bookers).
2. Lookalikes of that first-party data (built automatically inside Performance Max).
3. Site visitors (last 30-90 days), segmented by intent (cart abandoners vs. blog readers).
4. In-market audiences (Google’s algorithmic intent signals).
5. Affinity audiences (broad interest categories — weakest signal).

Always combine at least levels 1-3. If you have under 1,000 customers in your CRM, use Customer Match anyway — Google will use the signal to model lookalikes even if the match rate is small.

The “one campaign per goal × audience” rule

Here is the heuristic I give every new client: a campaign should exist for every unique combination of (conversion goal, audience tier, margin tier). If those three are identical between two campaigns, merge them. The bidder will work better with more volume in one campaign than the same volume split across two.

Concrete example for an ecommerce brand selling supplements:

Campaign	Goal	Audience tier	Margin tier
PMax-BestSellers	Purchase value (tROAS 6.0)	All	High
PMax-Clearance	Purchase value (tROAS 2.0)	All	Low
PMax-Subscribers	Subscription value	Email signups + lookalike	High
Search-Brand	Purchase value (tROAS 12.0)	All	All
Search-NonBrand	Purchase value (tROAS 4.0)	All	All

Five campaigns. Notice every row is a unique combination. No “test” campaigns, no overlapping audiences.

Geographic structure: when (and when not) to split

Splitting by geography is a common over-correction. The rule: split only if you have different bids, different creative, or different language. If you’re just curious what’s happening in California vs. Texas, don’t split — use a geo report inside one campaign.

Cases where splitting is justified:

• Different currencies (always split).
• Different shipping cost structures changing margin enough to warrant a different tROAS.
• Different conversion windows (B2B sells faster in some regions).
• Different ad copy (local language, local social proof, “Free shipping in NY!”).

If none of those apply, keep one campaign and use location bid adjustments or a Geographic Performance script if you really need granular control.

What about Performance Max asset groups vs. ad groups?

Asset groups are Performance Max’s equivalent of ad groups, but the unit of optimization is different. Inside an asset group, you provide all the creative — text, image, video, logo — and Google’s system mixes them across Search, Display, YouTube, Discover, Gmail, Maps. Asset groups should be themed by what the creative says, not by who sees it (which is determined by audience signals).

So an asset group named “Best-Sellers-Winter” would contain:

• 5+ headlines focused on the winter promotion
• 5+ long headlines and descriptions
• 10+ product images (different angles, lifestyle shots)
• 3+ short videos (15-30s)
• Lifestyle and product logos

A separate asset group named “New-Arrivals” would have all-new creative. They live in the same campaign, sharing budget and bidding, but the algorithm picks which creative set wins per impression.

A good rule of thumb: 3-7 asset groups per Performance Max campaign. Fewer and you starve creative diversity; more and you starve volume per group.

When to use Search-only campaigns alongside Performance Max

A common worry: “If Performance Max can run search ads, do I need Search campaigns at all?” Yes, in these cases:

• Branded keywords: always carve into their own Search campaign. You want full visibility into branded performance, separate budget control, and the ability to use exact-match keywords to defend.
• Competitor brand bidding: legally permitted in most markets (bidding on names, not trademarks in ad copy). Use a dedicated Search campaign with phrase or exact match, manual review of search terms.
• High-margin head terms where Search outperforms Performance Max consistently. Test by running both for 60 days and comparing incremental ROAS.

Don’t run a Search campaign that overlaps Performance Max with no clear reason. They will fight each other for query share and your CPC will rise.

Account-level conversion goals vs. campaign-level

Set conversion goals at the account level when every campaign should optimize against them (e.g., “purchase” for an ecommerce account). Override at the campaign level when a specific campaign has a different objective (e.g., a Demand Gen campaign optimizing for “engaged session” instead of “purchase”).

The 2026 Google Ads UI has improved this: you can mark a goal as primary at the account level, then override per campaign with a custom goal set. Use this — it keeps account-level goals consistent for reporting while letting upper-funnel campaigns chase different metrics.

A 90-day account rebuild plan

If you’ve inherited an account and want to restructure without losing performance:

Days 1-14: Audit. Tag every campaign as keep, kill, or restructure. Build a tagged spreadsheet of all conversion actions and their volume. Identify duplicate or stale ones.

Days 15-30: Set up new accounts-level conversion goals with proper values and Enhanced Conversions. Don’t delete old ones yet — let them run in parallel.

Days 31-60: Build new campaign skeleton according to the right archetype. Launch in parallel with existing campaigns at 30% of expected budget. Monitor CPA/ROAS against the legacy structure.

Days 60-90: Ramp new structure to 100% of budget. Pause legacy campaigns. Watch for week-over-week regression — if you see one, the most likely culprit is conversion value misalignment, not structure.

Frequently asked questions

How many keywords should each ad group have? For broad-match Smart Bidding campaigns, you want 5-15 keywords per ad group, all thematically tight. The myth of “one keyword per ad group” (SKAGs) is dead; Smart Bidding works better with thematic clusters because the bidder can pool query data across related variants.

Should I use Performance Max for lead generation? Yes, but only after you have at least 30 conversions per month from existing campaigns and you’ve implemented Enhanced Conversions for Leads with offline conversion uploads. Without offline data, Performance Max will optimize for cheap form-fills, not qualified leads.

What’s the minimum budget to run Performance Max effectively? Roughly 10× your target CPA per week, or 50 conversions per month, whichever is higher. Below that, the algorithm doesn’t have enough signal to converge.

Can I have one MCC for multiple brands? Yes — that’s exactly what manager accounts are for. Keep separate accounts per brand for billing, conversion isolation, and clean reporting. Use cross-account conversion tracking if customers convert across brands.

Do I still need separate campaigns for mobile vs. desktop? No — device bidding has been deprecated and replaced with bid adjustments. Use one campaign and adjust by device if your data shows a strong difference in CPA.

---

A clean Google Ads account is one of the highest-leverage marketing assets you can build. The structure choices above compound over time: clean conversion data improves Smart Bidding, which improves performance, which improves the data quality further. If you’re staring at an inherited account and don’t know where to start, an account audit is the first step — and one of the most cost-effective marketing investments you can make this quarter.