# Movement: from a national market to a local club

National screen v2, calculated 9 September 2026. Website edition: 11 September 2026.

## What this study answers

Where does metropolitan France contain concentrated populations that merit investigation for a club? The screen identifies demographic candidates. It does not establish paying demand, available properties, a revenue TAM or a number of viable clubs.

| Scale | Selected areas | Outside Île-de-France | Île-de-France |
|---|---:|---:|---:|
| Density concentrations | 14 | 11 | 3 |
| 5 km zones | 79 | 53 | 26 |
| 3 km sectors | 63 | 31 | 32 |
| 1.5 km bassins | 165 | 76 | 89 |

Each layer is selected separately. They are not a hierarchy and their populations must not be added. Small bassins outside the 14 density concentrations remain in the inventory. Names and regional labels describe the nearest source-cell commune; they do not establish official boundaries. Searches also use the communes represented in each national catchment.

## Evidence and assumptions

The national source is the INSEE Filosofi 2021 metropolitan 200 m grid, released in February 2026. The input contains 2,298,582 populated cells and 62,592,659 fiscal individuals. Source population and age bands are aggregated without inventing an age distribution. Some source attributes are imputed; unknown ages are excluded from the target age bands.

Income is an annual living standard in reference-year euros per consumption unit (UC), adjusted for household size. It is not salary. The published cell mean is capped at departmental P5/P95. The study does not treat this as an uncapped mean or a median, and does not project it to 2026 prices.

The base scenario assumes a lognormal income distribution in each cell, with dispersion 0.35 and an annual threshold of €30,000/UC. With cell mean m, dispersion σ and threshold T, μ = ln(m) − σ²/2; the estimated share above the threshold is 1 − Φ((ln(T) − μ)/σ). This share is applied separately to source counts aged 25–54 and 25–64. That assumes age and income are independent within the cell. INSEE does not supply the joint local distribution required to validate this assumption.

An income scenario estimate is therefore modelled, not an observed affluent customer count. It establishes neither interest in the club nor willingness to pay.

## How the national areas are selected

Source coordinates are southwest corners of cells in EPSG:3035. The calculation adds 100 m on each axis for cell centres. A circle includes the full population of cells whose centres are within its radius, without partial-cell apportionment. A 1.5 km circle has 177 possible lattice positions, representing 7.08 km² of full grid cells; its displayed geometric area is πr².

The algorithm considers grid centres across metropolitan France, including unpopulated centres. It selects the highest estimated age-25–64 income score, then excludes centres within twice that radius. Thus national circles in one layer share no source-cell centres. Minimum estimates are 4,000 at 1.5 km and 12,000 at 3 km and 5 km. These are explicit screening settings, not commercial viability thresholds. Greedy selection is not an optimal packing or an exhaustive list of every possible club address.

Density concentrations start from cells with estimated age-25–64 density of at least 1,500/km². A grid operation connects nearby cells, then retains connected areas with an estimate of at least 10,000. The nominal 600 m bridge parameter uses a 300 m-radius kernel; at 200 m resolution its effective footprint is 3×3. It does not bridge every 600 m gap. Counts and area include connecting cells. These shapes are exploratory concentrations, not official agglomerations.

Displayed circles are constructed in EPSG:3035 with 128 edges and transformed to WGS84. Density shapes use a 10 m topology-preserving simplification. Display geometry does not change source counts. The provenance download records source and output checksums.

## Marseille Sud: the first local case study

The local search evaluates 705 centres within the 6th, 7th and 8th arrondissement IRIS boundaries. Twelve comparison centres are selected using the full-circle 1.5 km base score, with centres more than 1 km apart. Their circles overlap: they are alternative search areas, not independent markets or available properties.

Each centre has two radii (1.5 and 3 km) and nine income scenarios: thresholds of €25,000, €30,000 and €35,000/UC crossed with dispersions of 0.25, 0.35 and 0.50. The 216 rows represent 12 centres × 2 radii × 9 scenarios, not 216 locations. These scenarios are sensitivity assumptions, not confidence intervals. National results retain the base scenario.

Full-circle counts can extend outside Marseille Sud. The local view also reports counts restricted to cell centres inside the study boundary. A population share inside the boundary measures people, not land area. The strict Marseille Sud membership ratio uses the inside-boundary denominator.

The separate neighbourhood layer covers 75 IRIS statistical areas. Census counts use RP2022 with 2024 geography; income medians use Filosofi 2021 with 2022 geography. Median income is published for 66 areas and unavailable for nine. Missing values remain null. Medians are not averaged or pooled, and census population is not substituted for fiscal population.

The IGN/INSEE boundary extract is from 1 January 2026 under Open Licence 2.0. Matching IRIS codes does not prove unchanged boundaries between vintages. The map is descriptive; historical geometric equivalence remains unverified.

## Relating a catchment to the financial model

The website reads the membership target from the shared Business Model assumptions. Required share = membership target ÷ the explicitly labelled audience. Changing the financial target changes this arithmetic ratio; it does not change demographics. The workbook has its own editable target cell and does not synchronize with browser settings.

Required ratios are neither measured capture nor predicted conversion. A broad concentration is not a single-club service area. The reported 1,500 La Montgolfière Canal members relate to December 2019; member origins are unavailable and no local capture benchmark follows. Lamarck membership remains unreported in the sources checked.

Next evidence: address-specific travel times; a concrete offer and price tested through reservations or deposits; comparable competitors and reasons to choose the club; and usable premises, rent, fit-out, capacity and cash needs entered into the financial model. The fiscal grid excludes some populations, includes some students at their parents’ address, and does not add nonresident daytime workers.

## Sources and files

- [INSEE Filosofi 2021 grid](https://www.insee.fr/fr/statistiques/8735162?sommaire=8735243)
- [INSEE grid methodology](https://www.insee.fr/fr/statistiques/fichier/8735106/documentation_donnees-carroyees_filosofi2021.pdf)
- [INSEE geographic codes 2021](https://www.insee.fr/fr/information/5057840)
- [INSEE RP2022 population by IRIS](https://www.insee.fr/fr/statistiques/8647014)
- [INSEE Filosofi 2021 income by IRIS](https://www.insee.fr/fr/statistiques/8229323)
- [IGN/INSEE Contours IRIS](https://cartes.gouv.fr/rechercher-une-donnee/dataset/IGNF_CONTOURS-IRIS)
- [Open Licence 2.0](https://www.etalab.gouv.fr/licence-ouverte-open-licence/)

The webpage provides the complete workbook, all 321 national records, the 216 Marseille scenarios, 75 IRIS areas and the source register. provenance.json records export hashes and checks. The underlying calculation and reproducible export scripts are maintained in the project repository; this note and its public downloads do not include the approximately 468 MB raw input.