Google Ads Structure for Service Businesses: The Lead Gen Blueprint

The governing rule of the playbook is simple: Google can only optimize toward what you tell it to. Feed it real signals (qualified leads, accurate exclusions) and it scales. Feed it junk (raw form fills, leaky broad match) and it scales the junk. For service businesses, that means the structure exists to produce clean conversion data, not to look tidy.

Two corrections to old conventional wisdom apply directly to service accounts. First, Demand Gen is not a universal budget line. For B2B niche services it almost always fails, with evidence of GBP 465 cost per lead versus GBP 67 on Search in the same account. Demand Gen sits at zero percent for B2B. Second, raw form fills as the primary conversion train the algorithm on junk. The primary conversion must be a qualified lead.

A service account does not need many campaigns. It needs four, each with a clear job. The Core campaign captures high-intent demand. The Competitors campaign intercepts buyers evaluating named rivals. The Branded campaign protects your own name cheaply. The Remarketing campaign re-engages warm traffic. Demand Gen is excluded for B2B.

• Core High-Intent Search is the engine: most of the budget, mixed broad and phrase match, moving to Target CPA once data builds
• Competitors uses exact match on rival brand terms and routes those queries here rather than into Core
• Branded protects your own name at low cost and high conversion rate
• Remarketing / RLSA re-engages warm visitors with Target CPA bidding
• Demand Gen stays at zero for B2B niche services; it rarely beats Search on cost per lead

Inside each campaign, build ad groups around themes of intent, not around match types. The current standard is STAGs 2.0: 5 to 15 tightly themed keywords per ad group, never one keyword per group. Mix broad, phrase, and exact within the same ad group rather than separating them, and theme by what the searcher wants, not by how the keyword is matched.

Naming keeps the account legible. Use clear, descriptive campaign names such as Leads_Core_Broad_2026 so anyone auditing the account can read its structure at a glance. The discipline that matters most is restraint: every ad group and campaign you add divides your conversion data, so add only what earns its place.

• 5 to 15 tightly themed keywords per ad group, not single-keyword ad groups
• Mix match types within the same ad group; do not split broad, phrase, and exact into separate ad groups
• Theme ad groups by search intent, not by match type
• Name campaigns descriptively (for example Leads_Core_Broad_2026)
• Resist over-segmentation: each campaign should clear 30 or more conversions per month

Bidding is a sequence, not a setting. Enhanced CPC is deprecated, so skip it. The correct order is Manual CPC (only for brand new accounts with zero conversion data), then Maximize Conversions with no target, then Maximize Conversions with Target CPA, then Portfolio bidding once volume is high enough.

The exit gates are specific. Move off Manual CPC after 15 or more conversions with tracking discrepancy under 3%. Move from Maximize Conversions to Target CPA once a campaign reaches 30 or more conversions in the last 30 days, setting the first Target CPA 10 to 20% above recent actual CPA to give the algorithm breathing room. Adjust by no more than 10 to 15% per change, wait 7 to 14 days between changes, and never change bidding and structure at the same time.

Broad match only works for service accounts when it is fenced in. The formula is Broad Match plus Smart Bidding plus ruthless negatives plus audience signals; remove any one and it becomes waste. Before launch, load 100 to 150 starter negatives covering wrong geography, informational intent ("how to", "what is"), job seekers ("jobs", "careers"), wrong audience tiers, and competitor brands you are not targeting.

Then run the weekly ritual. Every Monday, export the search terms report sorted by cost descending. Anything with 3 or more clicks and zero conversions becomes a priority negative. Terms with the right intent that are converting get added as exact-match keywords. Own-brand terms appearing in non-brand campaigns become negatives; competitor brands move to the Competitors campaign. Structure negatives at three levels: an account-level shared list for universal exclusions, campaign-level for specific exclusions, and ad-group level for fine-grained intent separation.

• Launch with 100 to 150 starter negatives before broad match goes live
• Monday ritual: sort search terms by cost, block 3-plus-clicks-zero-conversions, promote converting terms to exact match
• Use a three-level negative structure: account shared list, campaign, and ad group
• Do not over-negative: "free" can block "free consultation"; check before adding broad negatives
• Move competitor terms to the Competitors campaign and own-brand terms out of non-brand campaigns

This is the part that separates service accounts that scale from ones that stall. The primary conversion must be a qualified lead, not a raw form fill. Set up the Google Tag through GTM (never manual code), enable Enhanced Conversions for Leads, and make the primary conversion action a Qualified Lead imported offline from the CRM with currency values where possible.

The loop runs in five steps: a visitor clicks the ad and the GCLID is stored in the CRM; the sales team qualifies the lead; qualified leads are exported with their GCLID and timestamp; the CSV is uploaded to Google Ads weekly (or via CRM integration); and the algorithm learns to find more people who look like those qualified leads. The conversion action must be Import type, not website or pixel, and new conversion actions take 15 to 30 minutes to propagate before an upload is accepted.

The four-campaign structure holds for both, but the details diverge with audience size. B2B service accounts serve small target audiences, so broad match is too risky: use exact and phrase match only on core terms, build niche negative keyword lists, and track downstream conversions like sales-qualified leads and closed-won rather than stopping at the first form fill. Decision-maker targeting can be layered in through Customer Match and LinkedIn audience imports.

B2C service accounts have larger audiences, so broad match becomes viable once a campaign clears 30 or more conversions per month. Track the full pipeline from lead to appointment to sale, treat geographic targeting as critical for local services, and set up call tracking as a separate conversion action from form fills.

• B2B: exact and phrase only on core terms, niche negatives, track SQL and closed-won, add Customer Match and LinkedIn audiences
• B2C: broad match viable past 30 conversions per month, track lead to appointment to sale
• B2C local services: geographic targeting is critical, and calls are tracked as a separate conversion
• Both: the primary conversion stays the qualified lead, imported offline

Allocate 60 to 80% of budget to proven campaigns that have held a stable cost per lead for 7 or more days, and 20 to 40% to testing new keywords, ad groups, and formats. Scale winners by no more than 20 to 30% every 7 to 10 days, and never cut more than 20% in one move, since larger cuts trigger re-learning. Scale only after efficiency is stable, never to fix a broken cost per lead.

When a campaign fails (cost per lead spikes more than 20% or volume drops more than 30%), run the diagnostic in order rather than guessing. Check tracking first: if it is broken, pause and fix it, because everything else is noise until the signal is clean. Then work through search terms, negatives and match types, bidding, structure, ad and landing-page relevance, budget, creative, audience signals, and finally external factors like landing-page speed and competitor activity in Auction Insights.

• 60 to 80% of budget to proven campaigns, 20 to 40% to testing
• Scale winners 20 to 30% every 7 to 10 days; never cut more than 20% at once
• Diagnose failing campaigns in order, starting with tracking
• If tracking is broken, pause and fix it before anything else
• Read cost per lead and Impression Share (lost to budget versus rank are different problems), not raw clicks