A Modern Approach to Bank Branch Network Optimization Using Location Intelligence and the p-Median Model
Banks are operating in a fundamentally different environment from the one that shaped traditional branch networks. Customers increasingly use mobile applications, online banking, ATMs, and digital payment platforms for routine transactions. At the same time, physical branches continue to play an important role in activities that require trust, advice, complex financial decisions, and personal interaction.
This creates a difficult strategic question for banks: How many branches are actually needed, and where should those branches be located?
Simply closing branches based on low transaction volumes can create gaps in customer coverage. Conversely, maintaining branches in locations with declining demand can increase operating costs without generating sufficient value. A more effective approach is to combine customer, demographic, geographic, competitive, and financial data to determine where physical banking infrastructure creates the greatest impact.
One analytical technique that can support this decision is the p-median model, a location-allocation method designed to identify facility locations that minimize the overall distance between customers and service points.
What Is the p-Median Model?
The p-median model is a classic optimization technique used to determine the best locations for a fixed number of facilities.
The origins of the model can be traced to operations research and network-location studies conducted in the 1960s. S. L. Hakimi's influential 1964 research examined the optimal placement of facilities within network structures, including communication switching centers and police stations. A subsequent 1965 paper extended the work to the distribution of multiple facilities, providing the foundation for what became known as the p-median problem.
At its simplest, the model asks:
If a company can operate a specific number of facilities, where should those facilities be positioned so that customers can reach them with the least overall travel distance or cost?
For banking, the "facilities" can be branches, while customer demand can be represented using households, deposits, account holders, population, income, business activity, or other relevant indicators.
The traditional model minimizes weighted distance. In a modern banking application, however, distance does not have to be the only consideration. Travel time, market potential, deposits, customer density, income, commercial activity, competitor presence, branch capacity, and digital adoption can also influence the analysis.
Why Branch Location Still Matters in a Digital Banking Era
Digital banking has reduced the need for customers to visit branches for everyday activities. Checking balances, transferring money, paying bills, depositing checks, and managing many banking services can now be performed remotely.
However, branches have not become irrelevant.
A physical location can still influence customer acquisition, brand visibility, mortgage and lending conversations, wealth-management relationships, small-business banking, and customer confidence. Branches can also be particularly important in markets where customers have lower digital adoption or where face-to-face financial assistance remains valuable.
The challenge is therefore not simply "branches versus digital."
The more relevant question is:
What combination of physical and digital channels provides the best coverage for a bank's customers and target markets?
This is where location analytics becomes valuable.
Building a Data-Driven Branch Optimization Framework
A bank looking to redesign its branch network can begin by building a geographic dataset containing existing branch locations and potential customer demand.
Several categories of information can be combined.
1. Existing Branch Locations
The first layer identifies where the bank currently operates.
For U.S. banks, the FDIC Summary of Deposits provides an important source of branch-level information. The annual survey reports branch office deposits as of June 30 and provides geographic and institution-level information. The 2025 release covered more than 76,000 domestic offices operated by more than 4,400 FDIC-insured institutions.
This information can help analysts understand the current geographic footprint and deposit concentration.
2. Market Share
Branch counts alone do not reveal whether a network is effective.
A bank may have many branches in a market but hold a relatively small share of deposits. Another bank may operate fewer locations while capturing a significant share of the market.
Market-share analysis therefore provides a useful competitive dimension.
Analysts can compare:
Deposit market share
Branch share
Deposit growth
Competitor density
Geographic coverage
Customer concentration
This can reveal markets where a bank is potentially overextended or underrepresented.
3. Demographic and Economic Variables
Customer demand is rarely distributed evenly across a city.
Population, household count, median or mean household income, age distribution, employment, homeownership, business density, and population growth can all influence potential banking demand.
For example, a rapidly growing suburban area with increasing household income may represent a stronger opportunity for a new branch than a neighborhood with declining population, even if both currently contain similar numbers of residents.
4. Geographic Accessibility
Straight-line distance is not always a good representation of customer accessibility.
A branch located one mile away across a highway or river may be less accessible than another branch located two miles away on a direct road.
A modern model can therefore incorporate road-network travel time, traffic conditions, public transportation, parking availability, and other accessibility factors.
How a Bank Could Apply the Model
Suppose a bank operates 25 branches across a metropolitan area but wants to determine whether 25 remains the appropriate number.
The analytics team could divide the market into geographic demand zones, such as ZIP codes, census tracts, or smaller grid cells.
Each area could then receive a demand score based on factors such as:
Demand Score = Households + Deposits + Income + Growth Potential + Business Activity
The exact formula would depend on the bank's strategic objectives.
The p-median model could then evaluate possible branch locations and assign customer demand points to the nearest selected facilities.
The bank could test several scenarios:
Maintain 25 branches
Reduce the network to 20 branches
Expand to 28 branches
Close low-value locations and relocate them
Open branches in high-growth neighborhoods
Establish smaller-format branches
Combine branches with digital or ATM-based services
Instead of asking which branches are busiest today, management can ask which network configuration is likely to provide the strongest coverage and economic value.
An Illustrative Banking Case Study
Consider a hypothetical bank operating across a large U.S. metropolitan area.
The bank has 30 branches. However, its customer base has shifted significantly over the past decade. Several downtown branches now experience declining foot traffic, while population and household income have increased in outer suburban communities.
A basic branch-performance report might recommend closing several low-transaction branches.
A location-analytics study could produce a different answer.
After combining demographic data, deposit market share, competitor locations, household density, travel times, and projected population growth, the analysis might discover that some downtown branches are still strategically valuable because they serve high-value commercial customers.
Meanwhile, two suburban areas may have substantial customer potential but inadequate physical coverage.
The resulting strategy could involve:
Closing two overlapping branches
Relocating three branches toward growing neighborhoods
Opening two smaller-format branches
Increasing ATM availability in lower-demand areas
Supporting routine transactions through digital channels
The objective is not simply to reduce the branch count. It is to reallocate physical capacity toward locations where it generates greater strategic value.
Real-World Applications Beyond Banking
The p-median concept is not limited to financial services.
Its underlying logic has been applied to numerous facility-location problems.
Emergency Services
Emergency-service planners can use location models to position fire stations, police facilities, ambulances, or other services so that communities can be reached efficiently.
One well-known application of p-median-style modeling examined fire-station location in Barcelona under changing network and demand conditions. The example demonstrates an important lesson for banks: location decisions must account not only for today's demand but also for uncertainty and future changes.
Retail
Retailers can analyze customer concentration, purchasing power, competition, and travel distance to determine where stores should be opened or consolidated.
A supermarket chain, for example, could use a similar model to determine whether five smaller stores provide better market coverage than three larger locations.
Healthcare
Hospitals, clinics, diagnostic centers, and pharmacies can use location-allocation models to improve patient accessibility.
The same methodology can help identify communities that are underserved despite being located relatively close to facilities geographically.
Logistics
Distribution centers and warehouses are another major application.
A logistics company can use facility-location optimization to minimize transportation distances between warehouses and customer demand zones while considering delivery volumes and road networks.
From Traditional p-Median to Modern Location Intelligence
The traditional p-median model remains useful, but today's analytics environment allows organizations to build much richer models.
Modern branch optimization can incorporate:
Geographic Information Systems (GIS)
Machine learning
Customer segmentation
Predictive demand forecasting
Population growth projections
Competitor intelligence
Mobility and traffic data
Digital banking adoption
Branch profitability
Deposit growth
Customer lifetime value
This changes the role of the model from a simple "find the closest branch" exercise into a broader strategic decision system.
For example, a bank could predict which neighborhoods are likely to experience population growth over the next five years and incorporate those projections into its location model.
This helps prevent a common problem in network planning: optimizing the network for yesterday's customers.
Measuring the Success of a Redesigned Branch Network
A branch optimization project should not end when new locations are selected.
Banks should establish measurable performance indicators, including:
Customer acquisition
Deposit growth
Revenue per branch
Cost per customer
Average travel time
Market share
Branch utilization
Cross-selling performance
Customer retention
Digital-to-physical channel migration
These metrics can be monitored after implementation and fed back into the model.
This creates a continuous optimization cycle rather than a one-time branch review.
The Future of Bank Branch Optimization
The future of branch strategy is unlikely to be defined simply by having more branches or fewer branches.
Instead, banks will increasingly focus on having the right branches in the right locations.
The combination of location analytics, demographic intelligence, market-share data, customer behavior, and optimization models provides executives with a more rigorous foundation for these decisions.
The p-median model is particularly valuable because it converts a complex geographic problem into an optimization framework: given a defined number of facilities and a set of customer demand points, how can the network provide the most effective coverage?
For banks, that question has become increasingly important as customer behavior, demographics, technology, and competitive markets continue to change.
The most effective branch network of the future may therefore look very different from the network of the past. Some branches may disappear, others may move, and new locations may emerge in growing markets. The important point is that these decisions should be guided by evidence rather than assumptions.
Data analytics can transform branch planning from a reactive cost-cutting exercise into a proactive strategy for customer coverage, market growth, and long-term network effectiveness.
This article was originally published on Perceptive Analytics. At Perceptive Analytics our mission is "to enable businesses to unlock value in data." For over 20 years, we've partnered with more than 100 clients — from Fortune 500 companies to mid-sized firms — to solve complex data analytics challenges. Our services include AI Consulting Services in Philadelphia and Power BI Implementation Services, turning data into strategic insight. We would love to talk to you. Do reach out to us.
Top comments (0)