Multi-location SEO operating model for home services
A marketing leader responsible for a multi-location home-services group needs one answer before touching rankings: who has the right to change the brand, domain, location page or Google Business Profile? In Caldrin's August 2026 review of five ranking pages, all five covered profile and location governance. Only two addressed central versus local ownership. The review measured page coverage; it did not observe how operators work or what causes conflicts.
Multi-location SEO is a control system for shared assets and local facts. Group leadership sets the rules. Brand and web teams protect architecture. Local operators supply information that headquarters cannot manufacture. A named owner decides what changes, who approves it and which denominator shows whether the change worked.
Who owns multi-location SEO decisions?
The group should centralise control of shared search assets while assigning local teams responsibility for verifiable location facts. In Caldrin's judgement, central control without local input can raise the risk of generic pages and stale profiles. Local freedom without shared guardrails can raise the risk of duplicate pages, category drift and access problems. These are risks the framework is designed to surface, not outcomes measured in the five-page review.
This is Caldrin's operating model. It does not claim that one organisation chart guarantees better rankings.
| Layer | Accountable owner | Decisions held centrally | Local input required | Change record |
|---|---|---|---|---|
| Group | Portfolio or group marketing lead | Search policy, measurement definitions, access standards, escalation | Material operating changes across brands | Decision, approver, effective date |
| Brand | Brand marketing lead | Brand name, service taxonomy, editorial standards, approved claims | Market-specific services and customer language | Brand standard and exception |
| Domain | Search or web product owner | URL structure, templates, canonicals, redirects, internal links, indexation controls | Location-page requirements and real proof | Source URL, destination, release and rollback state |
| Location | Branch or regional operator | Changes stay within brand and profile policy | Hours, service availability, staff, photos, coverage and customer-facing facts | Location, evidence source, verifier and timestamp |
| Google Business Profile | Group-controlled primary owner; approved managers by location | Ownership, recovery, naming policy and core integrations | Hours, photos, services, reviews and temporary changes | Profile ID, editor, old value, new value and approval |
The primary owner should be an account the group controls, not a departed employee or agency login. Local managers can handle facts that change on the ground, but ownership transfers, business-name changes and other high-risk changes need a central approval path.
Start with one source of truth for every location
A location register should exist before another content calendar. It holds the facts expected to agree across the site, Google Business Profile and other important listings.
For each location, record:
- group and brand;
- public business name;
- real operating address and whether customers can visit;
- local phone number and website destination;
- services actually delivered from that location;
- hours, temporary closure state and service-area boundaries;
- Google Business Profile ID, primary owner and approved managers;
- location-page URL and indexation state;
- last verified date and the person who verified it.
The register is a change-control ledger, not a score. A branch can report a new closing time, but the central owner checks where that fact must change and records completion. When records disagree, the model should show which system is authoritative and who resolves the conflict.
Name, address and phone consistency belongs in this workflow, but a formatting difference is not an all-purpose ranking diagnosis. The practical goal is accurate customer information, cleaner entity resolution and fewer conflicting records.
When a branch can have its own Google Business Profile
A separate Google Business Profile requires a separate eligible location. Google's current guidelines allow one profile per business location and reject virtual offices that do not meet operating requirements. A service-area business without a storefront is allowed one service-area profile. A service business with different locations, separate staff and separate service areas may have one profile for each eligible location.
Use this decision path:
- Confirm the location. Record the address, staff and customer-facing setup. A mailbox is not a branch.
- Decide whether customers can visit. If yes, the profile may show its address when Google's other requirements are met, including permanent signage.
- If the team travels to customers, use the service-area configuration. Hide the address. Google says the overall service area should generally remain within roughly two hours' driving time from the base, while allowing exceptions.
- Separate genuinely separate locations. Distinct staff and service areas can support separate profiles; duplicate profiles for one location cannot.
- Record ownership and verification. Store the profile ID, owner, managers, linked page and last verification.
See Google's guidelines for representing a business before creating, moving or recovering a profile. Primary platform documentation outranks an agency checklist and should be rechecked at implementation.
Govern location and service-area pages with four gates
A location page should represent a real branch or a service-area decision a customer needs to make. Swapping a city name into generic copy creates another URL competing for the same material.
| Gate | Evidence needed | Decision if absent |
|---|---|---|
| Operational reality | Real branch or documented service coverage | Do not create a location claim the business cannot fulfil |
| Distinct customer need | Different services, availability, access, housing conditions or operating detail | Consolidate into a stronger parent page if the page has no separate job |
| Local proof | Staff, original photographs, projects, reviews or verified service facts | Hold the page until the location can supply proof |
| Search purpose | A distinct query and conversion path that does not duplicate a nearby page | Merge, reposition or leave unpublished |
The domain owner controls templates, canonicals, navigation and indexation. The location owner supplies facts and proof. Shared material can explain brand standards; variable fields must reflect real differences.
This page owns ongoing location and profile operations. Decisions about keeping, merging or redirecting acquired brand domains belong to the separately held acquisition-governance workflow.
Reviews need local context and central safeguards
The group owns the approved request method, escalation rules and platform compliance. The local team knows whether a complaint refers to the right crew, job or branch.
- Route each review to the correct profile and location owner.
- Flag policy, legal or safety issues for central review.
- Let a person with local context draft the response.
- Record response status and escalation, not just average rating.
- Report counts and response status by location before aggregation at brand level.
Reviews appeared in three of the five cleaned comparison pages. That supports including review governance here. It does not show that a response template, target or workflow causes ranking, AI visibility or commercial performance.
Measure local search at the level of the decision
A single visibility number hides whether the problem belongs to one brand, one branch, one query or the measurement itself.
| Measurement level | Denominator | Useful reading | Decision it supports |
|---|---|---|---|
| Query by location | Every tracked query-location point | Organic position, pack present and brand present | Where visibility is absent or concentrated |
| Pack-fired view | Only points where a local pack appeared | Brand appearances and pack position | Whether absence reflects the market or the result format |
| Location | Tracked points assigned to one real branch or service base | Coverage by service and market | Which branch needs profile, page or operational work |
| Brand | All eligible locations within the brand | Distribution, not a blended average alone | Whether the issue is systemic or local |
| Portfolio | All brands, with each brand's denominator retained | Comparable issue counts and data coverage | Where group resources and governance attention go next |
| Commercial outcome | Calls, forms, booked work or revenue with valid attribution | Outcome by location and source | Whether visibility changes produce business value |
603 baseline and method: On 9 August 2026, Caldrin measured eight service queries across 61 coordinate test locations in New Hampshire, Massachusetts and Maine, creating 488 query-location points. The raw export contains 21 unique town labels. A local pack appeared at 412 points. 603 Basement Solutions appeared at 10 of all 488 points and 10 of the 412 pack-fired points. A separate 488-point grid on 14 August found Groundworks Foundation Repair Manchester at 38.1% of all points. These are before-picture readings. The uncontrolled five-day movement for 603 remains excluded as an outcome.
What the Caul calibration showed about source recurrence
Caldrin's 30 August 2026 Caul calibration used a frozen hs_multi_location cluster: 10 prompts across rendered Google AI Overviews, ChatGPT's consumer interface and Gemini's consumer interface, repeated 10 times per prompt and surface. That created 300 attempts. Successful trials used United States country-level geography with no geo downgrade. Of the 300 attempts, 265 met the grounding-eligibility rule and 238 distinct cited domains appeared.
Mean pairwise source-set Jaccard was 0.4695 and mean answer-token Jaccard was 0.3258. These values describe overlap under the documented calculations; neither has a supported good-or-bad threshold.
Provider reliability uses the full two-cluster calibration, not the 300-attempt multi-location subset. The full calibration contained 600 attempts, 200 per provider. Gemini returned 200 provider-success rows, but only 137 met the grounding-eligibility rule; 63 were quarantined. Google AI Overview had eight failures in its 200 attempts. Gemini's grounding eligibility remains a methodology failure and the full Caul run remains HOLD.
This was consumer-surface instrument calibration. It did not test an intervention, establish consumer-population share, describe universal behaviour, create a market benchmark or support a causal content prescription.
Run the model as an exception system
Routine facts should be easy to update and risky changes hard to make by accident. Use a monthly operations review to answer five questions:
- Which location facts changed, and are the site and profiles aligned?
- Which profiles have ownership, suspension or duplication risk?
- Which location pages lack a distinct purpose or current proof?
- Which branches are under-represented when the correct denominator is applied?
- Which decision needs group approval rather than another local workaround?
The output is a short action register with an owner, evidence link, due date and rollback path.
Next step: define the partner scope
Use the home-services partner-selection model to decide whether the operating gap needs a local specialist, a portfolio governance partner or a defined implementation workstream.