Structure Is the Strategy
A $3,000-a-month Google Ads account and a $30,000-a-month account fail in exactly the same way: one giant campaign, stuffed ad groups, broad match on by default, and a search-terms report nobody has opened since launch. Structure is not an administrative detail — it determines which queries get which budget, which ads, and which landing page. At Brand Advertisers, we architect search accounts the way a service business actually wins work: by service line, by job value, and by intent. This is the account blueprint we deploy for contractors, HVAC shops, roofers, plumbers, med-spas, and the marketers who run accounts for them.
Campaign architecture by service line and job value
Never run one campaign for the whole business. A plumbing company that does $300 drain clears and $15,000 repipes in the same campaign is bidding against itself: the cheap, high-volume queries will consume the budget and the profitable jobs starve. Split campaigns by service line, then by job value within a line when economics differ:
- High-ticket / priority campaigns: installs, replacements, repipes, remodels, full systems. Highest budgets, tightest match types, dedicated landing pages.
- Mid-value service campaigns: repairs, tune-ups, diagnostics. Healthy budgets, ad copy that pre-qualifies before the click.
- Low-ticket / volume campaigns: drain clears, filter swaps, entry treatments. Capped daily budgets so they never cannibalize the priority campaigns.
- Brand campaign: your company name and close variants, small budget, cheap insurance against competitors bidding on your name.
- Competitor conquesting: optional, separate, and capped hard — these clicks are expensive and convert worse. Never blend them into a priority campaign where they'll distort your numbers.
Budget pacing follows the same logic: allocate monthly budget by service-line priority and gross margin, not by which campaign happens to be active. Check spend against targets weekly, and use shared budgets only if you genuinely don't care which service line gets the spend — usually, you do.
Ad Groups and Match-Type Discipline
One intent per ad group
Each ad group should represent exactly one intent, with keywords, ads, and the landing page all speaking to that intent. "Emergency water heater repair" and "water heater replacement cost" are different jobs with different economics; they belong in different ad groups with different ads. A practical ceiling is roughly 10–20 tightly themed keywords per ad group. When you find yourself writing an ad that has to cover two ideas, split the group.
Match types: where broad match is and isn't safe
Start with phrase and exact match on every campaign that carries real budget. Broad match in 2024-and-later Google accounts is smarter than it used to be — but "smarter" means aggressive expansion, not precision. It can work when all three conditions hold:
- Conversion tracking is feeding real outcome data back into Smart Bidding (see below) — broad match without strong conversion signals burns budget fast.
- You have a mature negative-keyword list built from weeks of search-terms data.
- You're running a volume campaign with slack budget, not a high-ticket campaign where every wasted click is expensive.
In our experience, broad match on a high-ticket install campaign with thin conversion data is the single most common way local accounts quietly lose four figures a month.
The Negative-Keyword Weekly Habit
The search-terms report is the account's truth serum. It shows the actual queries your ads matched to — including the ones you'd never knowingly bid on. "Free," "DIY," "jobs," "salary," "training," "how to" (when you sell done-for-you work), "wholesale," "used," nearby city names you don't serve, and competitor brand names you don't want to conquest are all standard additions.
Make this a standing weekly appointment — same day, same person, thirty minutes:
- Open the search-terms report filtered to the last 7 days, all campaigns.
- Flag any query with spend and no conversion that doesn't describe a buyer you'd send a truck to.
- Add negatives at campaign level for terms specific to one service line; add to a shared negative list (applied across campaigns) for universal junk like "free" and "DIY."
- Review once a month for conflicting negatives that are accidentally blocking good traffic.
Thirty minutes a week consistently outperforms most "optimizations" agencies bill for.
Location Targeting and Ad Assets
Set targeting to 'presence,' not 'presence or interest'
Google's default targeting includes people whose interest is your service area — someone in another state researching plumbers for a relative, or a competitor checking ads. For a local service business that dispatches trucks, that's pure waste. In campaign settings, change location targeting from the default "Presence or interest" to "Presence": people physically in (or regularly in) your service area. Verify the setting on every campaign — Google applies the expansive default to new campaigns.
Assets that pre-qualify the click
Ad assets (extensions) aren't decoration; they filter buyers before you pay for the click.
- Call assets: enable call reporting, schedule call assets only during hours someone actually answers. A tracked call that rings out is worse than no call — it trains Google's bidding on the wrong signal.
- Location assets: linked to the correct Google Business Profile, so the ad shows address, map, and directions — confirmation to nearby searchers that you're local.
- Sitelinks: structured as service lines and proof, not generic pages: "Financing," "Reviews," "Service Areas," "Water Heater Repair." Each sitelink is a filter — "Financing" screens for budget-conscious buyers early.
- Structured snippets and callouts: license numbers, "Free estimates," "Same-day service," "Serving [county] since [year]" — concrete qualifiers, not adjectives.
Conversion Tracking That Counts Real Outcomes
Clicks and form fills are not conversions; a qualified booked job is. The account should track, at minimum:
- Qualified calls: calls lasting longer than your threshold for a real conversation (often 60–120 seconds for service businesses), not calls under 30 seconds that are mostly misdials.
- Booked jobs via CRM import: the gold standard. Import offline conversions from your CRM — deal created, estimate booked, job won — back into Google Ads so Smart Bidding optimizes toward revenue, not leads. Many CRMs (ServiceTitan, Housecall Pro, Jobber, HubSpot, GoHighLevel) support this via offline conversion import or a GCLID parameter captured on your forms — verify your platform's current docs, since integrations change.
- Primary vs. secondary conversions: mark only true outcomes as "Primary" (what bidding optimizes for); keep page views and soft engagements as "Secondary" so they don't contaminate the signal.
The Account Review Cadence
Structure gets built once; it drifts weekly without a cadence. Run the account on this rhythm, with the same owner and a standing calendar invite:
- Weekly (30–45 min): search-terms negatives, budget pacing vs. plan, disapproved ads or assets, call-recording spot checks on a sample of tracked calls.
- Bi-weekly: ad copy rotation — pause the loser in any A/B pair once statistical direction is clear; check auction insights for new competitors.
- Monthly (60 min): cost per qualified lead and cost per booked job by campaign and service line; reallocate budget toward the service lines winning at target economics; review landing-page conversion rates alongside ad metrics.
- Quarterly: re-architecture review. New service lines, seasonal demand shifts (tune-ups before summer, generator installs before storm season), match-type experiments, and landing page refreshes.
The pattern across every account we rebuild: the structure, not the bids, was the leak. Fix the architecture first — campaigns mapped to job value, one intent per ad group, disciplined match types, ruthless negatives, presence targeting, pre-qualifying assets, and CRM-fed conversion data — and then bidding, copy, and budget tuning actually have something to work with. A search account built this way stops being an expense line and starts behaving like what it should have been all along: a sales engine with a dispatcher.