Guide · Structure
How to structure Google Ads for multiple locations
Written by Saoirse Brennan · 14 October 2025 · Refreshed 21 June 2026 with the current LocaliQ medians
A multi-location account has one structural job: produce cost and conversion data that names a branch. Everything else, bidding, creative, budgets, works or fails downstream of that. This guide is the structure we install in month one of every engagement, the reasoning behind each choice, and the migration path for getting there from a blended account without torching your history.
Why do blended accounts fail quietly?
Because averages are comfortable. Across our audits, the median gap between a brand's best and worst branch is a 2.7× difference in cost per lead, and a blended account shows none of it. The dental group in this diary read a healthy blended number for two years while three practices booked patients at more than double the group median. The account was fine on average and broken in the particulars, which is the default state of blended structures.
The failure compounds through Smart Bidding. A shared tCPA target across twelve branches optimizes toward wherever conversions are cheapest this week, which usually means the strongest branches absorb ever more budget while weak ones starve quietly. Nobody decided that; the algorithm did, because the structure gave it one pool of money and one target. Per-location structure is how you get your hands back on the wheel without micromanaging bids.
What does the reference structure look like?
- One campaign set per location with radius or postcode targeting; border overlaps decided deliberately and written down
- Budgets from branch economics: capacity, margin, lead value; reviewed monthly, moved in steps
- Shared negative lists at account level (jobs, DIY, informational), plus local negatives per branch
- One landing page per location: address, hours, staff, reviews, and the branch's own conversion tracking
- Location assets linked to the right Business Profile so ads carry the correct branch details
- Per-branch conversion data, calls included, before any Smart Bidding target is set
Naming conventions carry more weight than they look like they should. We name campaigns brand / branch / intent, so a 50-campaign account still reads at a glance and the monthly export pivots cleanly by branch. Accounts where campaign names accreted over years ("Search 2 NEW final v3") fight their owners at every report; renaming costs an afternoon and pays forever.
How do you migrate from a blended account?
Carefully, and in this order. First, install per-branch conversion tracking while the old structure still runs, so the new campaigns launch with real location data instead of guesses. Second, split brand from non-brand if that has not happened yet; brand clicks subsidize every blended report they touch. Third, carve locations out of the blend in tranches, three or four branches at a time, starting with the biggest spenders, so learning phases stagger instead of hitting the whole estate at once.
Expect two to three soft weeks per tranche while bidding relearns. The dental engagement carved twelve practices in three tranches across weeks five to eight, and the diary honestly records the flat patch in the middle. What you should not expect is a cliff: the old campaigns keep running until each tranche's replacement has exited learning, which is why the migration takes a month rather than a weekend and why it costs nothing in lost lead volume when sequenced properly.
Which numbers anchor the targets?
Vertical medians first, branch history second. The 2026 LocaliQ benchmarks give the bands: dentists at 10.67% conversion and $72.97 CPL, health and fitness at $67.36 CPL, home and garden at $8.33 CPC, against a cross-industry median CPL of $66.69. A branch outside its vertical band has a findable, usually structural, reason. Targets set from wishes starve delivery; targets set within about a fifth of demonstrated performance tighten stepwise as the data improves.
Worth noting from the same report: the year-over-year trend favored advertisers in 87% of industries on conversion rate, and overall cost per lead fell for the first time in five years. Rising tides like that hide sloppy structure, because a blended account can improve on market drift alone while its worst branches keep leaking. Judge each branch against its vertical band and its own history, never against the estate's blended trend.
How does the monthly ledger keep it honest?
One line per location: spend, leads, cost per lead, conversion rate, trend. The format matters because it removes the blend as a hiding place; a branch that misses for two months triggers a fix list, and a branch that beats target at full budget earns more, per the reallocation discipline in geo expansion. The counting layer underneath, calls especially, is covered in the call-tracking guide, and skipping it invalidates the whole ledger: the gym chain's three "failing" sites were merely phone-heavy.
Two ledger conventions save arguments later. Every number carries its window (trailing 30 days, never calendar months, so a five-weekend month cannot flatter anyone), and every budget move gets a one-line reason in the same row. Six months in, that column of reasons is the institutional memory most multi-location brands never had: why Cork got cut in March, why Galway earned its increase, what the Limerick experiment concluded.
What are the most common structural mistakes?
- Even budget splits. Political peace, economic nonsense. Budgets follow branch economics or the ledger is theatre.
- One landing page for everyone. Local intent converts on local proof: address, hours, faces, reviews. A generic page taxes every branch's quality score.
- Location extensions pointing at the wrong Business Profile. Ads that show a rival branch's hours erode trust in ways no bid can buy back.
- Shared bid strategies across unlike branches. A portfolio strategy mixing a saturated city-centre site with a growing suburban one serves neither.
- Structure without counting. Per-branch campaigns bidding on blended or form-only conversion data reproduce the old problem with more campaigns.
Fair questions
Should each location get its own campaign?
From roughly five locations up, yes, or at minimum its own ad groups with location-split budgets. The deciding factor is attribution: you need cost and conversion data that names a branch. Below five locations the overhead can outweigh the insight, and tight geo targeting inside shared campaigns can carry it.
How should budgets be split between locations?
From branch economics, never evenly: capacity, margin, and lead value per site. An even split is a political convenience that overfunds saturated branches and starves growable ones. Review the split monthly against per-branch cost per lead, and move budget in steps with controlled windows.
One website or one landing page per location?
One site, one landing page per location, each with the branch's address, hours, staff, and reviews. Location pages convert local intent measurably better than a generic contact page, and they give each campaign a destination whose quality score reflects the branch, not the brand average.
How do you stop locations competing against each other?
Radius or postcode targeting per campaign with deliberate overlap rules for border areas, shared negative lists, and one bid strategy per branch rather than a portfolio fighting itself. Where two branches genuinely share a catchment, pick one to own the area and record the decision.
Does this structure work with Performance Max?
Only with discipline. PMax reports at campaign level, so a national PMax blurs every branch back together. Run it per region at most, feed it store-level location assets, and keep brand excluded. If a location's economics cannot survive its own campaign, PMax will hide that rather than fix it.