Build vs buy site selection software: GIS and spreadsheets or a platform?

For most multi-site occupiers expanding or reshaping a portfolio, the best answer is the third path: buy the portfolio record and a site selection agent, and keep your own weights, cost ceilings and committee format on top. The in-house version (GIS, licensed demographic data and an Excel model) answers the market question well. The portfolio question, how this site compares with the leases, occupancy costs and stores you already have, is where a shared record earns its keep.

Most real estate teams are not choosing between nothing and software. They already have a homegrown model. The question is how much more to invest in it, and which parts are worth owning.

What does "building" site selection actually look like?#

In-house site selection usually has three parts. A GIS tool draws trade areas and maps competitors. A licensed data feed supplies demographics, daytime population and traffic counts. An Excel workbook pulls those numbers together, applies weights, and produces a ranked list for the real estate committee. One analyst, sometimes two, owns all of it.

That setup is a reasonable one, and it is also a build that carries the costs of one. The weights live in cell formulas few people have audited. Trade areas get redrawn by hand for each deal. Existing-store performance and lease terms come from separate systems and get pasted in. When a candidate needs to be compared with the occupancy cost of the three nearest stores you already lease, the analyst opens lease abstracts one by one.

Here is what a team owns when it keeps the build:

Why does the build start to strain?#

The in-house model strains because portfolio questions take far longer than market questions. Pulling demographics for a trade area takes minutes. Comparing a candidate with your own nearby leases, their occupancy cost, their expiration dates and their co-tenancy rights takes days, because that data sits in lease documents and finance systems the GIS never sees.

REAL has modeled this for a complex location-analysis workflow: roughly 40 hours of experienced specialist effort per location, against about 20 minutes of REAL analysis. REAL is explicit that this is a modeled figure, that manual-equivalent effort is not realized savings, and that it is not a price (REAL, Portfolio Strategy). Treat it as a sense of scale. Across 100 locations, the modeled gap is about 4,000 hours versus about 33.

The second strain is key-person risk. When the analyst who built the workbook moves on, the logic often goes too, and the next person rebuilds it rather than inheriting it.

Related reading: what site selection software does.

What does the in-house stack cost to run?#

The in-house site selection stack costs more than its license line suggests, because most of the cost is people and refresh work rather than software. Price it by category, over the same horizon you would use for a subscription. Every input below is yours to fill in; prices vary too much by vendor and contract to quote here.

Annual in-house cost = data licenses + GIS seats + (deal hours + upkeep hours + refresh hours + package hours) × loaded hourly rate. Compare that total with a subscription plus the internal time you would still spend reviewing scores and preparing for committee. If you already pay for the data and a full-time analyst, the question is less about cost and more about throughput and key-person risk.

When does building site selection make sense?#

Keeping or building an in-house model is a reasonable choice in these situations:

1. 01Low deal volume. A handful of openings or relocations a year rarely justifies a platform. A careful analyst with good data will do.
2. 02A stable format and market. One store prototype in a region you know well means the scoring model rarely needs to change.
3. 03A strong, long-tenured analyst. Someone who owns the model, documents it, and plans to stay.
4. 04Market-only decisions. If the committee only needs demographics and competition, without a comparison to existing leases, the GIS stack covers it.
5. 05A one-off study. A single market entry or a single consolidation exercise can be run as a project and archived.

Outside those cases, the model tends to become a bottleneck: deals wait on one person, and the committee sees a ranking without the reasoning behind it.

What is the third path?#

The third path is to buy the portfolio record and the site selection agent, then run your own criteria and committee outputs on top. You keep control of the parts that reflect your strategy: format, weights, cost ceilings and the committee template. The vendor handles reading the leases, holding occupancy cost per location, and keeping scores reproducible.

Ask any vendor who builds those workflows. Some expect your team to build them on their platform, which brings part of the run cost back in house. With REAL, you define the rules and REAL's engineers build and maintain the workflow on the shared record.

REAL's site selection agent scores candidate markets against your criteria (format, trade area, demographics, competition, traffic, cost ceiling) and against the leases you already hold. The action it produces is a ranked shortlist a committee can approve. The evidence is attached: comparable occupancy cost, trade-area overlap with existing stores, and the weights and inputs behind every score. When a deal is signed, LOI terms and dates pass to lease administration without re-keying. Your own scoring logic or committee format runs as custom workflows on the same record.

Build, buy, or buy and build on top: a decision scorecard#

How do you test site selection software on your own stores?#

Test site selection software, or your own model, on stores whose results you already know. A test on your own history tells you far more than a demo on a market you have never entered.

1. 01Give the tool your history. Share store performance, formats and the leases behind each location, so the scores reflect how your stores actually perform.
2. 02Hold back a set of stores. Keep 10 to 20 recent openings out of what the tool sees. Ask it to score those sites as if they were candidates.
3. 03Compare its ranking with what happened. Check whether the stores it ranked highest actually performed best. A model that sorts your own recent openings well is one you can use for your next ones.
4. 04Read the reasoning on a live candidate. Pick one site your team is weighing now. Ask the tool to show the inputs, weights and comparable occupancy cost behind its score. If the committee can follow the reasoning, the score will hold up in the meeting.
5. 05Check the portfolio questions. Ask how the candidate compares with your three nearest stores on rent, pass-throughs, lease term and trade-area overlap. This is where a shared lease record makes the biggest difference.
6. 06Agree measures of success up front. Ranking accuracy on held-back stores, time to a committee-ready package, and whether every score can be explained.

Run the same test on your current in-house model. It is the fairest comparison you can make, and sometimes it shows the existing model is better than people thought.

Questions to ask before you decide#

• How many sites will we evaluate in the next two years, including closures and relocations?
• Can we compare a candidate's occupancy cost with our three nearest stores today, and how long does it take?
• Who can explain the current weights, and are they written down anywhere?
• Can we reproduce the ranking the committee approved last year?
• Does a consolidation question use the same model as an expansion question? See when to consolidate a location.
• Who builds and maintains our custom scoring logic, and what happens to it if that person moves on?

For how the same decision plays out across lease administration, accounting, construction and facilities, see build vs buy AI agents for corporate real estate. For the space side of the portfolio, see build vs buy space management software. To price the in-house version, use the cost model for building AI for real estate operations.

The third path keeps your strategy in your hands and puts the lease reading, occupancy cost and reproducible scoring on a foundation someone else maintains.