- A network's problem is not ranking: it is stopping its own pages from competing with each other.
- Central brand and individual branches have different goals and need different pages, with clear rules.
- With rank tracking cannibalisation shows up in days rather than quarters.
- The measurement side is free: semalt.com/authorize.
A network with forty branches has a problem a single-location business never meets: its worst competitors are its own pages. Head office publishes a landing page per city, each franchisee builds their own website, somebody opens a profile per branch, and within a year the brand is competing against itself across half of Spain.
This article is the manual we hand to franchise networks, clinic chains, gyms, workshops and restaurant groups based in Madrid with national expansion. It is not about ranking better: it is about organising who ranks for what.
What cannibalisation actually means in a network
It is not two pages covering the same topic: it is two pages competing for the same search intent. When that happens the engine picks one, and usually picks badly, because neither is clearly the best answer.
In multi-location networks it appears in three very recognisable forms.
| Form | Visible symptom | Usual origin |
|---|---|---|
| Central versus local | The national landing page appears for a specific city search | Generic landing page listing cities |
| Local versus local | The Chamberi page appears for Chamartin searches | Identical local pages with the district swapped |
| Franchisee's own site | Two domains of the same brand competing | No digital agreement in the contract |
| Profile versus website | The business profile takes the click and the site never appears | A poor or non-existent local page |
The architecture that works: two layers and one rule
The structure that holds a forty-branch network has only two layers and one non-negotiable rule: every search intent has a single responsible page.
National layer. Service and brand pages with no city name. They answer generic searches and explain the proposition, reference pricing and how things work. These pages receive most of the links and carry the domain's authority.
Local layer. One page per branch, not per city. The difference matters: in Madrid there are networks with six branches, and a single “Madrid” page cannot answer six different proximity intents.
- ✓Exact address, its own phone number and the branch's real opening hours
- ✓Photographs of the branch, not corporate stock
- ✓The team or manager by name, where the service allows it
- ✓Services available at that specific branch, which are rarely all of them
- ✓How to get there: transport, parking, neighbourhood references
- ✓Reviews or cases from that branch, not from the brand in general
If removing the district name makes two branch pages identical, the engine cannot tell them apart either. It is the simplest criterion and the most ignored.

The conflict with franchisees
The uncomfortable part of this work is not technical: it is contractual. Many franchisees build their own website because they want commercial autonomy, and that site competes with head office using the same brand.
The solutions we have seen work, in order of preference. First, head office provides the branch page inside the main domain with limited editing rights for the franchisee. Second, when an independent site already exists, specialise it: let the franchisee's domain cover content head office does not, and link to the branch page for commercial queries. Third, the worst option but better than nothing, agree which queries each side works and verify quarterly.
“In a network the biggest enemy of ranking is not the competition: it is the absence of a written agreement about who publishes what.”
Reviews and each branch's profiles
In a network, business profiles matter as much as pages, and they are usually managed worse because they are spread across forty different owners. The usual result is a network with excellent branches online and branches that do not exist at all.
What has to be centralised is the criterion, not the execution: the same way of writing the name, the same primary category, the same base corporate photographs plus the branch's own, and a committed response time for reviews. Each branch keeps its listing, but under identical rules.
The indicator worth reviewing by zone is not the number of reviews but their regularity: a branch with a new review every fortnight signals activity, while one with forty published in the same month signals exactly the opposite.
Measurement: what to watch when forty pages look alike
With a network you cannot review every branch weekly. The system that works is exception-based: track a reduced set and act only when something falls outside expectations.
What to do when two branches sit very close together
In Madrid it is common to have two branches two kilometres apart. District differentiation is not enough there, because the catchment areas overlap and both pages compete for the same searches.
The way out is not writing more text: it is differentiating by something real. Different services, different hours, different practitioners, parking at one and transport links at the other. If there genuinely is no functional difference between them, the honest option is a single page with two addresses, letting the engine handle proximity through the business listing data.
Indexing: the silent problem of large networks
When a network publishes forty branch pages at once, it is common for only part of them to be discovered. The ones left out never appear in any query, and because the set as a whole works, nobody notices until a franchisee complains.
Verification is straightforward with the indexing module: submit every branch URL and check in the log which received verified crawler visits. The first hundred URLs are free, enough for a complete mid-sized network.
AI answers and the local question
“What is the best place for X near Chamberi?” is a question now asked of an assistant as often as of a search engine. In networks we see an interesting pattern: answer systems cite branch pages with complete data more readily than brand pages, because they carry verifiable, specific information.
The AI visibility module lets you check, for any domain, which queries the brand appears in and which pages get cited. In a network it is worth reviewing zone by zone: it is not unusual for a brand to be strong in one city and invisible in another where the local competitor publishes better information.
Opening a new city: the first ninety days
An opening is the moment a network wins or loses a year. The sequence we recommend starts before the launch, not after.
Six weeks ahead the branch page goes live with whatever information already exists: address, expected opening date, services, how to get there. Publishing early lets it be discovered and evaluated while there is no rush, and it captures searches from people who have already seen the sign in the street.
In launch week it is submitted for indexing, the business profiles are completed with exactly the same data, and the page is linked from the branch directory and from the national service page.
Over the following three months three queries for that city get watched and compared with the local competitor, who is rarely the same as elsewhere in the country. If after ninety days the branch does not appear for its own district, the problem is either differentiated content or indexing, and both are checked in minutes.
The data head office should require from each branch
Maintenance is where networks decay. A changed summer schedule, a new phone number, a withdrawn service: if the data is updated in only one place, the inconsistency starts eroding credibility and confusing the systems that reconstruct the business identity.
The practical solution is not technological: it is a one-page procedure each branch reviews quarterly - address, phone, hours, active services, photographs, manager - which head office cross-checks against its own website and profiles. Ten minutes per branch, once a quarter, prevents most of the problems that later cost weeks of diagnosis.
When volume is needed rather than order
Fixing the architecture resolves cannibalisation, but it does not create authority. In expansions into new cities, where the brand has no presence or mentions yet, steady signal is required. That is where managed campaigns fit: AutoSEO at $149 a month with AI keyword selection, daily link building from a network of over 230,000 partner sites, on-site recommendations and live reporting; FullSEO from $500 a month with a dedicated team, manual query control and manual donor approval. Conditions are on the pricing page.
The network that grows by acquisition
There is a particularly delicate scenario: the network growing by buying existing businesses, each with its own website, trading name and history. It happens often with clinics, workshops and training centres around Madrid.
The temptation is to unify everything under the new brand at once and shut down the acquired domains. That is the fast route to losing years of accumulated signals. The sequence that works is different: first inventory which URLs on each acquired domain receive real impressions, then create the equivalent page inside the main domain, and only then redirect one by one, keeping those redirects live indefinitely.
During the process it is worth watching queries for the old brand: many people will keep searching the previous name for years, and that demand should land on the matching branch page, not on a corporate homepage that answers none of it.
How work splits between head office and branches
The split we have seen hold over time gives head office the structural work and each branch the specifics. Head office maintains the template, the national pages, data consistency and measurement; the branch supplies real photographs, news, team details and the particularities of its service.
What matters is that the franchisee has a simple, fast way to update their data. When the only route is an email to marketing that takes two weeks, data stops being updated and the network degrades through an accumulation of small inconsistencies.
Frequently asked questions
One page per branch or per city?
Per branch whenever a distinct physical address exists. A city page only makes sense when there is a single branch in that city.
What about a franchisee who already has a website?
Specialise it rather than close it where possible: content head office does not cover, plus a clear link to the branch page for commercial queries. Closing it without agreement creates conflict and loses signals.
How long does cannibalisation take to resolve?
Once the structure is corrected and pages genuinely differ, four to eight weeks. If only the copy changes while intent stays the same, it does not resolve at all.
Is a domain per city worth it?
Almost never. It splits authority, multiplies maintenance and usually creates exactly the problem it was meant to avoid.
Check whether your branches compete with each other
Positions by query and by URL, search data and indexing verification, free.
Open the Semalt dashboard