How to present a site recommendation so the board can decide

A good site recommendation fits on one page before any appendix: the site you recommend and why in two sentences, the alternatives you rejected in one table measured the same way, the risks and what would make you change your mind, and the date and source of every figure. Decision makers rarely reject a site because of the data; they reject the paper because they cannot tell which numbers are solid and which are hopes.

This article is for expansion managers, franchise developers, analysts and consultants who have done the work and now have to win the decision. It sets out a structure that works, shows it with an example, lists the habits that lose meetings, and explains how to have an AI employee produce the figures and the first draft for you.

The one-page structure

1. The answer. Name the site and give the two reasons that matter most. If you need a third reason, it is probably not a reason.
2. The choice. One table with every shortlisted site as a column and the same rows for each: people, competition, access, cost. Mark the best value in each row.
3. The trade-off. One paragraph on what the recommended site gives up. Every site gives something up; saying so earns trust.
4. The risks. Two or three things that could go wrong, each with what you would watch for and when.
5. What would change my mind. A specific fact that would flip the answer, such as a rival signing the unit opposite.
6. The sources. For every figure, where it came from and when. Mark anything modelled or estimated.

Everything else, the long lists, the maps of every rival, the methods, goes behind the page. The people deciding will read the first page carefully and the appendix only if they doubt the first page.

Making the table carry the argument

The table is where most papers go wrong. Rows get added because the data was available, not because they matter, and the reader is left counting ticks. A few rules keep it honest:

• Same method, same moment. Every site measured from its own door, with the same distance and the same kind of competitor, on the same day.
• Say which way is better. More people is better; more rivals is usually worse; a lower rival rating can mean weaker competition. Write it in a column.
• Group the rows. People, competition, access, risk. Readers weigh groups, not rows.
• Keep empty cells empty. If a figure does not exist for a site, such as population in a country without population data, leave the cell blank and say why in the caption rather than filling it with a guess.

An example, written out

A made-up example. A children's activity centre is choosing between three units in Vancouver. The page might open like this:

We recommend the Kitsilano unit. It has the most families with young children within a 10-minute drive and only two similar centres within 2 km, both rated below 4. It is the most expensive of the three and has the least parking; we expect most visitors to arrive on foot or by bus, and the unit has four bus stops within 500 m.

Illustrative figures only. Population from a residents grid where available; competition from a places database; both read on the same day.

Notice that Burnaby wins on children and parking. The paper says so, and says why it still loses: four competitors, well rated, and almost no bus service for parents without a car. A reader who spots Burnaby's advantage on their own and finds it unmentioned stops trusting the rest.

Habits that lose the meeting

• Opening with method. Nobody needs to know how the radius was drawn before they know which site you want.
• Precise numbers from rough data. Writing 14,213 children suggests a census of the street. Round to what the data can bear.
• Hiding the losers. Show at least two alternatives. A single site is a request, not a recommendation.
• Unlabelled estimates. If an area was drawn as a circle because the road area could not be drawn, or a figure is a sample, say so beside it.
• A verdict the table does not support. If the words say the site is the busiest, a row must show it.

Letting an AI employee draft it

The figures and the first draft can be produced for you and refreshed whenever the shortlist changes.

1. My data: the shortlisted sites, one row each with its position and name.
2. Travel area: a 10-minute drive around each.
3. Demographics: residents and age groups inside each area, where population figures exist.
4. Nearby places: similar businesses within 2 km of each site, passed on as each site with counts and the average rating.
5. xMap LLM, set to work on the whole list: told to write the answer paragraph and the trade-off paragraph, using only the figures it was given and naming the site it recommends.
6. Make a report: a PDF or 16:9 slides with your logo, the summary, the figures and a map, emailed to you or saved to a Google Drive folder.

Treat the draft as a draft: check that every sentence matches a row, then add the risks and what would change your mind, which only you know. For the side-by-side table itself, Compare sites measures two or three spots the same way, marks the best value in each row, writes a verdict that may only quote figures in the table, and gives you a PDF and a link to share.