Scoring sites with weights that fit your business, not someone else's

A weighted site score works when its factors are the few things that really drive your sales, each is put on the same 0 to 100 scale, and the weights add up to 100 and reflect what you have learned from your own best and worst sites. Score every candidate the same way, then change the weights a little and see whether the winner changes. If it does, the decision is close and needs a visit; if it does not, you can trust the ranking.

This article is for expansion managers, franchise developers, analysts and owners comparing more than a handful of sites. It shows how to choose factors, how to put them on one scale, how to set weights, a worked example with four sites in Cairo, the mistakes that make scores misleading, and an AI employee that does the arithmetic every time a new site comes in.

Choose four to six factors

More factors do not make a better score. Each extra factor dilutes the ones that matter. Pick the few that separate your best existing sites from your worst. Typical candidates:

• Demand: residents of the right age within a travel time, from Demographics (only in some countries: among them Egypt, India, France, Canada and Australia; not the United Kingdom or the United States).
• Competition: businesses of your kind within the same travel time, from Nearby places. Higher is worse.
• Activity: footfall per day in people-hours around the site, from Mobility Footfall, in the cities it covers, about 10 days behind.
• Anchors: the count of businesses that bring your customers past, such as schools for a tutoring centre or offices for a lunch bar.
• Access: how the roads next to the site move at your busy hour, from Traffic Flow (speed and jam factor, not car counts), or a yes or no for parking.

Put every factor on the same scale

Residents are counted in tens of thousands, competitors in single digits and jam factor from 0 to 10. Before weighting, turn each into a 0 to 100 score across your candidates:

1. For a factor where more is better, give the best candidate 100, the worst 0, and the rest a share in between: (value minus lowest) divided by (highest minus lowest), times 100.
2. For a factor where less is better, such as competitors, turn it round: (highest minus value) divided by (highest minus lowest), times 100.
3. If every candidate has the same value, give them all 50 and leave the factor out of the decision.

A Calculated field block works these out with one line per factor once you know the lowest and highest values, or xMap LLM can apply the same rule across the list.

Set weights you can defend

Weights add up to 100. Give the most to the factor that best separated your good and bad sites. Write each weight down with one sentence on why. A board can argue with a sentence; it cannot argue with a spreadsheet nobody can explain.

A worked example in Cairo

A made-up case. A chain of pharmacies is comparing four corner units in Cairo. It chose four factors and these weights: residents in a 10-minute walk 40, pharmacies in the same walk 30, clinics within 500 m 20, parking 10. The figures and scores below are invented, to show the method.

Made-up scores on a 0 to 100 scale, for illustration only. Total = 0.4 x residents + 0.3 x competition + 0.2 x clinics + 0.1 x parking.

B wins narrowly over A. Now test it. Move 10 points from competition to residents (weights 50, 20, 20, 10): A scores 68 and B 60, so the winner flips. The ranking depends on how much you care about rivals compared with people. That is a real business question, not a data one: the team decides that being the only pharmacy in the walk matters more in this city, keeps the first weights, and visits both A and B before signing. C and D are out under either set of weights.

Mistakes to avoid

• Copying someone else's weights. A coffee chain's score is wrong for a pharmacy. Your weights should come from your own sites.
• Counting the same thing twice. All residents and residents of working age rise and fall together across most sites. Pick one.
• Scaling against the whole city. Score candidates against each other. A 0 to 100 scale across a whole city squeezes all your candidates into a narrow band.
• Mixing measured and missing. If one country has population figures and another does not, score them in separate lists.
• Trusting decimals. A total of 64 against 62 is a tie. Only a clear gap that survives a change of weights is a decision.

Doing it with an AI employee

An AI employee can score every new site the same way the day it is added to your list. The blocks, in order:

1. New rows: starts when a new candidate is added to your sites table in My data or a shared spreadsheet.
2. Travel area: a 10-minute walk or drive around each site.
3. Demographics and Nearby places: residents, competitors and anchors inside each area.
4. Calculated field: each factor on a 0 to 100 scale, then the weighted total, one line each.
5. Sort and limit: highest total first.
6. Email: the new site's figures, its scores and where it ranks against the others.

What arrives within minutes of adding a site: its factor scores, its total and its place in the ranking. Score every neighbourhood for a new branch applies a ready-made score to a whole city, and Where to open next is a simpler starting point you can add weights to.