HomeCase studies › Social Impact
Case studies

Territorial prefeasibility for new restaurants (MTIE): how we moved a pilot portfolio of 18 gastronomic MSMEs from 71.4% Prime Cost to 61.8% and stopped the default spiral

Diego F. Parra By Diego F. Parra · Updated 2026-08-29· Social Impact
Territorial prefeasibility for new restaurants (MTIE): how we moved a pilot portfolio of 18 gastronomic MSMEs from 71.4% Prime Cost to 61.8% and stopped the default spiral — Masterestaurant
Quick verdict

Territorial prefeasibility for new restaurants (MTIE) will not stop a restaurant from opening in the wrong place. It stops credit from financing that mistake. Across this pilot portfolio of 18 gastronomic MSMEs in a mid-sized city, measuring demand, spending capacity and supplier distance BEFORE disbursement pulled average Prime Cost from 71.4% to 61.8% in seven months and 90-day delinquency from 18.6% to 6.1%. The original error was never operational. It was territorial: the bank was lending against foot traffic nobody had counted.

📈 Case studyA business case broken down: diagnosis, dated decisions and measured results· 18 min read· 2026-08-29

Start with the case file, because none of what follows can be judged without it: a pilot portfolio of 18 gastronomic MSMEs —11 below 500 thousand USD in annual revenue, 5 in the 500 thousand to 1 million band, 2 above a million—, a mid-sized Latin American city of roughly 620 thousand inhabitants, weighted average ticket of 8.40 USD, average age of 14 months at diagnosis, dining room as dominant channel with 27% aggregated delivery, and 214 formal jobs sustained in total. SATE Institute operated the program with a second-tier bank credit line; Masterestaurant S.A.S., technology ally of the model, supplied the MTIE instrument.

What opened the audit was not an income statement. It was delinquency: 18.6% at 90 days in a portfolio the credit committee had approved through conventional financial scoring, against a 6% expectation. When the program officer asked for an explanation, the commercial team said what they always say —«the sector is risky»— and that sentence is exactly where the problem hides. Risk is not uniform within a sector: inside the same city, in this same portfolio, eight locations held margin and ten sank, with identical financial products and comparable teams.

Diego F. Parra put it to the committee in a line that landed badly: the bank was not financing restaurants, it was financing addresses. And nobody had measured the addresses. Not one of the 18 files contained a real pedestrian count, an estimate of gastronomic spending capacity for the polygon, or a supplier map within 40 kilometres. Territorial prefeasibility existed as a form, not as a measurement, and a signed form predicts nothing at all.

It is worth saying what territorial prefeasibility is NOT, since the term has been hollowed out in operating manuals. It is not the 60-page market study commissioned to satisfy a file requirement, nor the competitor list drawn inside an arbitrary radius. It measures absorption capacity: how much real gastronomic demand exists in a polygon, at what spending elasticity, served today by how much installed supply, and at what logistics cost of provisioning. Four variables. Anything the instrument produces beyond those four is expensive noise.

Side-by-side comparison

Side-by-side comparison

BEFORE (baseline, month 0)AFTER (month 7)
Portfolio average Prime Cost71.4% of sales61.8% of sales
Theoretical vs. actual food cost variance9.7 percentage points2.4 percentage points
Labor Cost over sales38.2%31.5%
Portfolio 90-day delinquency18.6%6.1%
Weighted average ticket8.40 USD10.15 USD
Annualized front-of-house turnover142%88%
Food loss and waste (FLW) over purchases11.3%4.9%
Formal jobs sustained by the portfolio214 jobs263 jobs

What did the 18.6% delinquency rate reveal that financial scoring missed?

The 18.6% ninety-day delinquency in this pilot portfolio was not a credit problem but a street-address problem, and that distinction redesigned the entire program.

Eighteen food-service MSMEs in a mid-sized Latin American city of 620,000 residents, 214 formal jobs sustained, weighted average ticket of 8.40 USD, average age of 14 months: the committee had approved using conventional scoring expecting 6% delinquency and got three times that. Eight locations held margin, ten sank, with identical financial products and comparable teams. Latin American food service carries a reputation for risk —in Colombia 95% of the market consists of independent establishments, per Acodrés via Revista La Barra (2024)—, yet that reputation hides the fine reading. Uniform risk does not exist. What exists is one district that absorbs demand and another that does not, separated by fourteen blocks. None of the 18 files contained a pedestrian count, an estimate of the district's food spending, or a supplier map within 40 kilometers: territorial prefeasibility existed as a signed form, never as a measurement.

The bank was financing addresses, not restaurants

Diego F. Parra put it to the committee in a sentence that stung —the bank was not financing restaurants, it was financing addresses—, and the sting came from being verifiable file by file. Worth stating what prefeasibility is NOT, since operations manuals hollowed the term out: it is not the 60-page market study filed to satisfy a requirement, nor the competitor list drawn inside an arbitrary radius. It measures absorption capacity, four variables and nothing else: real food demand in the district, spending elasticity, installed supply already serving it, logistics cost of supply. Anything the instrument produces beyond that is expensive noise. The MTIE —Territorial Economic Impact Model contributed by Masterestaurant S.A.S. as technology partner of the program run by SATE Institute with a second-tier bank credit line— works on four layers measured on the ground, not declared by the applicant.

How the MTIE was applied across the 18 diagnostics?

Pedestrian flow was counted by time band across seven days in each district; available food spending was estimated by crossing housing density with average income of the census tract;

installed supply was inventoried with its seat capacity; suppliers within 40 kilometers were mapped along with their delivery frequency. Out of that comes a territorial revenue ceiling, the number no file carried. Four locations in the portfolio had invested between 62,000 and 118,000 USD in build-out against a territorial ceiling of 340,000 USD per year. At that ratio, no later operating efficiency rescues the loan. CapEx already decided. Thirteen hours open in districts whose real traffic concentrated into four: that was the 38.2% labor cost the committee read as payroll indiscipline. Crossing the pedestrian count by time band against hourly ticket records settled it beyond argument, because staff presence was being paid against nonexistent traffic for nine hours a day.

The 38.2% labor cost was a schedule problem, not a payroll problem

Fitting schedules to the measured curve and reassigning freed staff to peak bands lowered that indicator without firing anyone, which matters when we speak of formal employment in a portfolio holding 214 positions. Food service sustains employment at scale —357 million people live off tourism worldwide, one in ten workers, per UN Tourism (2024), and in Mexico alone tourism generated 2.9 million jobs in 2024, up 3.5%, per INEGI (2024)—, but sustaining it demands that each open hour pay its own payroll. Nine point seven points separated theoretical food cost from actual cost across the portfolio average, and that gap was not born in the kitchen either: it was born in the supplier map. Locations sourcing from beyond 40 kilometers bought on long frequencies, overstocked and lost product, while those with a nearby supplier turned inventory twice a week with marginal waste.

The 9.7-point gap between theoretical and actual food cost

Food waste moves numbers no small owner sizes properly —in the United States food surplus reached 380 billion USD in 2024, of which 85% ended up wasted, per ReFED (2025)—, and inside an MSME with an 8.40 USD ticket every point of deviation eats the month's margin. Measuring distance to the supplier BEFORE disbursement fixes what afterwards can only be administered. Territorial prefeasibility does not stop a restaurant from opening in the wrong place: it stops credit from financing it, and that is the whole thesis of the pilot. Eleven of the 18 companies billed under 500,000 USD a year, five sat between 500,000 and 1 million, two cleared the million; dining-room service dominated with 27% aggregated delivery. Once the MTIE became a step prior to committee, files falling short of a sufficient territorial ceiling stopped being approved at the requested amount and moved to approval with resized CapEx or redirection to another district.

What changed once measurement came before disbursement?

Portfolio Prime Cost gave way and delinquency settled, per the program indicators. A genuine concession fits here:

for years I argued that the MSME food-service problem was operational, and this pilot forced me to accept that much of it gets decided before the stove is lit. Start by measuring your own district this week, with the instrument matching your size. Under 500,000 USD a year: count pedestrian flow in front of your door across three time bands for five days, by hand, and compare it against your hourly tickets; the schedule fix costs nothing and pays inside fifteen days. Between 500,000 and 1 million: map your suppliers by distance and frequency, then calculate what the kilometer costs you in waste. Above 1 million: demand a territorial revenue ceiling in the file before signing any build-out. Above 5 million: audit CapEx per location against the district ceiling, not against the group average.

Transferable lessons by annual revenue band

And above 10 million, the celebrity-chef archetype opening large format on brand reputation —where traffic is assumed guaranteed— is precisely who needs the count most, because fame fills three months and the district pays for the next twenty years. I would not expect these results in three contexts, and saying so matters more than displaying the pretty number. First, in cities above three million residents with dense mass transit: there the pedestrian flow of a single point does not describe its market, because demand travels and the district turns into an elastic concept that the MTIE, as applied here, does not capture well. Second, in formats with more than 60% delivery sales, where the physical address stops being the determinant and the delivery radius rules; aggregated delivery in this portfolio was 27%, half that threshold. Third, in seasonal tourist-destination operations, whose demand curves depend on season rather than residency; the sector added 27.4 million new jobs in a year, per WTTC (2024), and much of that answers to seasonality, not density.

Limits of this case

Measure before assuming your case resembles this one. The symptom read as 18.6% delinquency; the root cause was CapEx sized for revenue the polygon would never produce. Four locations in the portfolio had sunk between 62 thousand and 118 thousand USD into fit-out against a territorial revenue ceiling of 340 thousand USD a year. At that ratio no downstream operational efficiency rescues the loan. The second symptom, Labor Cost at 38.2%, was never a payroll problem either. It was a schedule problem: locations opened 13 hours a day in polygons whose real flow concentrated into four. Staff presence was being paid against traffic that did not exist, and what exposed it was the cross between pedestrian counts by time band and ticket registration by hour. That 9.7-point gap between theoretical and actual food cost came from logistics, not from the kitchen. Twelve operations bought from more than 180 kilometres away, in small orders, with freight prorated at the supplier's discretion.

Where the leak actually was?

Rebuilding provisioning inside the 40-kilometre radius erased 6.2 of those 9.7 points without touching a single recipe. A third finding matters more to a lender than the other two:

territory predicts better than history. In this portfolio the territorial score sorted delinquency cleanly —the six locations scoring low held 74% of overdue balance— while conventional financial scoring had filed them as medium risk. Restaurant credit risk here has a geographic origin the traditional file never captures. The tension of the trade is real, and it has to be resolved rather than dodged: a founder picks a site out of affection, proximity to home, or a lease that happened to appear, and that same impulse sustains the 95% of the Colombian market held by independents (Acodrés, 2024). Killing it with technical bureaucracy kills formalization too. The bridge is a cheap, fast instrument: priced below 1.5% of CapEx and delivered in ten days, founders use it; priced at 6% and delivered in two months, they fake it.

Point by point

Origin error versus the right method, criterion by criterion

How the catchment area was defined
A · BEFORE (baseline, month 0)«Three blocks around», estimated by the founder with no traffic measurement.
B · Masterestaurant8-minute walking isochrone with real counts across three time bands over seven days.
Verdict: The right method wins. In two of the 18 locations the perceived radius tripled the effective one, and the entire CapEx was sized on that illusion.
Source of the revenue ceiling
A · BEFORE (baseline, month 0)Sales projection drafted by the applicant for the credit file.
B · MasterestaurantGastronomic spending capacity of the polygon from official household income series.
Verdict: The right method wins by a brutal margin: four locations had invested up to 118 thousand USD against a territorial ceiling of 340 thousand USD a year.
Treatment of the supply chain
A · BEFORE (baseline, month 0)Absent from the file; supplier chosen through acquaintance or 30-day terms.
B · MasterestaurantA 40-kilometre radius as approval criterion, with aggregated purchasing across operations.
Verdict: Second one is correct. Of the 9.7 points of theoretical-actual variance, 6.2 came from freight and small orders alone.
Frequency of the financial signal
A · BEFORE (baseline, month 0)Quarterly P&L delivered to the bank with more than 40 days of lag.
B · MasterestaurantWeekly Prime Cost and variance on a dashboard, alerting the program officer automatically.
Verdict: A quarter is useless for operating: by the time the number lands, the cash leak has already happened twelve times. Weekly or nothing.
Weight of territory in the credit decision
A · BEFORE (baseline, month 0)Narrative annex to the file, with no effect on the rating.
B · MasterestaurantTerritorial score with its own weighting and the status of a disbursement condition.
Verdict: The right method wins. While territory remains an annex, the committee keeps approving addresses no team can rescue.
Labour schedule and Labor Cost
A · BEFORE (baseline, month 0)Thirteen-hour opening out of sector custom and rent pressure.
B · MasterestaurantSchedule fitted to measured flow by band, with staff shifted into demand hours.
Verdict: Second one is correct. Closing dead shifts moved Labor Cost from 38.2% to 31.5% without a single layoff.
Side-by-side comparison

The method that produced the delinquency: prefeasibility as paperworkWhat failed

  • Catchment radius set by eye —«three blocks around»— with no pedestrian count and no measured traffic window.
  • Market study commissioned at paperwork prices (380 to 900 USD) and delivered with six-year-old census data.
  • Zero supply chain verification: 12 of the 18 locations depended on suppliers more than 180 kilometres away, with freight eating 3.1 points of food cost.
  • Credit scoring built only on the applicant's financial history, blind to the territorial variable.
  • Fit-out CapEx sized against the founder's sales expectation rather than the polygon's spending capacity.
  • P&L reported quarterly to the bank, while the cash leak ran weekly and was already consummated by quarter close.

The right method: MTIE as a gate before disbursementMasterestaurant

  • Polygon defined by an 8-minute walking and 12-minute driving isochrone, with real counts across three time bands over seven days.
  • Gastronomic spending capacity estimated from official household income series and away-from-home consumption penetration.
  • Provisioning map within a 40-kilometre radius: short supply chain as an approval criterion, not an aspiration.
  • Territorial score built into the credit committee matrix with its own weight, not appended as file narrative.
  • CapEx tied to the revenue ceiling the polygon can sustain; if it does not close, the project is resized or relocated before signing.
  • Weekly operating dashboard of Prime Cost and theoretical-actual variance, alerting the program officer when it crosses threshold.
Side-by-side comparison

Side-by-side comparison

BEFORE (baseline, month 0)AFTER (month 7)
Portfolio average Prime Cost71.4% of sales61.8% of sales
Theoretical vs. actual food cost variance9.7 percentage points2.4 percentage points
Labor Cost over sales38.2%31.5%
Portfolio 90-day delinquency18.6%6.1%
Weighted average ticket8.40 USD10.15 USD
Annualized front-of-house turnover142%88%
Food loss and waste (FLW) over purchases11.3%4.9%
Formal jobs sustained by the portfolio214 jobs263 jobs
The numbers that matter

Measured results of the pilot portfolio

9.6pts
drop in portfolio average Prime Cost over 7 months (71.4% → 61.8%)
12.5pts
reduction in 90-day delinquency of the pilot portfolio (18.6% → 6.1%)
49jobs
additional net formal jobs sustained across the 18 MSMEs (214 → 263)
6.4pts
less food loss and waste over purchases (11.3% → 4.9%)
95%
of the Colombian gastronomic market are independent establishments: the universe the instrument must reach
73%
of women-led businesses lack access to the financing they need to grow
Visualization
The numbers, visualized
The numbers, visualized9.6pts drop in portfolio average Prime Cost over 7 months (71.4% → ; 12.5pts reduction in 90-day delinquency of the pilot portfolio (18.6; 49jobs additional net formal jobs sustained across the 18 MSMEs (21; 6.4pts less food loss and waste over purchases (11.3% → 4.9%); 95% of the Colombian gastronomic market are independent establis; 73% of women-led businesses lack access to the financing they ndrop in portfolio average Prime Cost over 7 months (71.4% → 61.8%)9.6ptsreduction in 90-day delinquency of the pilot portfolio (18.6% → 6.1%)12.5ptsadditional net formal jobs sustained across the 18 MSMEs (214 → 263)49JOBSless food loss and waste over purchases (11.3% → 4.9%)6.4ptsof the Colombian gastronomic market are independent establishments: the universe the instrument must re…95%of women-led businesses lack access to the financing they need to grow73%
Sources: Resultados del caso · Acodrés (Revista La Barra) 2024 · UNDP 2024Chart by masterestaurant.com
Real case

“For fourteen months I was convinced my problem was the kitchen, and I hired a new chef twice trying to fix it. The territorial study showed me in one afternoon that my polygon supported 340 thousand dollars a year and I had buried 118 thousand dollars of construction betting on double that. We closed the second shift, cut from eleven to seven operating hours, switched to regional suppliers, and Prime Cost fell nine points in the first quarter. It hurts to admit the site was never right; it hurts less than going under owing the bank.”

— Owner, 16-table casual dining in a mid-sized city, annual revenue under 500 thousand USD
How to apply it in your restaurant

Treatment timeline

Weeks 1-2: territorial diagnosis with MTIE across the 18 files
We mapped each location's real polygon with an 8-minute walking isochrone and counts across three time bands over seven days, then crossed it against spending capacity and installed supply. The finding that reordered the program surfaced here: six locations operated in polygons whose revenue ceiling sat below their own break-even. No commercial strategy corrects that. The first version of the score also weighted traffic above spending capacity and failed badly in two high-flow, low-income zones where it predicted sales that never materialized; we recalibrated toward household income before going further.
Weeks 3-4: Restaurant Model Canvas and resizing of committed CapEx
With the territorial ceiling in hand, each operation rebuilt its model in the Restaurant Model Canvas: value proposition, cost structure and channel, all tied to the spending the polygon truly sustains. Three locations cut seating and sublet the freed square metres; two relocated 1.4 and 2.1 kilometres inside the same city, with program support so licences survived the move. The hard conversation was with the bank: restructuring tenor without penalizing the rating demands a documented territorial-origin criterion, and that document is precisely what MTIE produces.
Months 2-3: short supply chain and Standard Recipe Generator
We rebuilt provisioning inside a 40-kilometre radius and standardized the 40 recipes carrying 80% of each menu's sales through the Standard Recipe Generator, with theoretical plate cost and declared yield loss. Real friction arrived from the supplier side: two local producers could not hold steady weekly volume, so we assembled aggregated purchasing across five pilot operations to reach the producer's minimum. That joint purchase, more than the recipes, is what moved food cost.
Months 4-5: meseros.ai, real scheduling and FLW control
We fitted the labour schedule to pedestrian counts by time band and deployed meseros.ai with its dashboard to track suggestive selling and table time. Labor Cost eased from 38.2% to 31.5% with no layoffs: dead shifts closed and staff moved into the hours that carried demand. Daily waste logging took food loss and waste from 11.3% to 4.9% of purchases, a line the IDB has pushed through #SinDesperdicio under SDG target 12.3 and which, here, was straight cash.
Months 6-7: M&E, consolidation and report back to the credit committee
We closed with formal measurement: weekly Prime Cost per location, theoretical-actual variance, formal jobs sustained and delinquency, all on the program dashboard. Results consolidated at month 7 and held two quarters later, which is the minimum window before you call something a result rather than a bounce. The report back carried what matters for scale: the territorial score became a disbursement condition for every new gastronomic operation on that credit line, with its own weight in the matrix.
✦ AI applied

And with AI?

Apply AI to your restaurant's day-to-day to decide better and faster. Diego F. Parra is an expert in AI applied to restaurants.

Masterestaurant tools & method

Ecosystem instruments used in the program

The twin-ecosystem model splits the roles sharply: SATE Institute sets the local economic development agenda, operates the program and measures impact; Masterestaurant S.A.S., technology ally and owner of the software, supplies the instruments. All of them are closed, off-the-shelf products rather than bespoke builds, and that condition is what allows replication across a portfolio of 18 or of 400 without multiplying implementation cost.

Diego F. Parra

Diego F. Parra — International consultant, expert in creating and scaling restaurants and in AI applied to restaurants, foodtech and HORECA. Methodology applied in 8.400+ restaurants across 43 countries · Expert in Artificial Intelligence applied to restaurants, hospitality and food businesses · 20+ years in restaurants, catering, large events and business growth · Author of 3 ISBN-registered books: «Triunfar o morir en el intento» (2013) and «De esclavo a dueño» (2023) · International keynote speaker for the HORECA sector.

FAQ

Questions from the credit committee

What is territorial prefeasibility for new restaurants (MTIE)?
It is the prior measurement of four variables in the polygon where a site is proposed: real gastronomic demand, household spending capacity, installed supply already absorbing it, and logistics cost of provisioning. It produces a territorial score that enters the credit committee with its own weight. It does not replace financial analysis of the applicant; it corrects it where credit history is blind, which is the address itself.

What is territorial prefeasibility for new restaurants (MTIE)?

It is the prior measurement of four variables in the polygon where a site is proposed: real gastronomic demand, household spending capacity, installed supply already absorbing it, and logistics cost of provisioning. It produces a territorial score that enters the credit committee with its own weight. It does not replace financial analysis of the applicant; it corrects it where credit history is blind, which is the address itself.

Why does conventional financial scoring fail with gastronomic MSMEs?
Because it measures the applicant's repayment capacity and ignores the territory's absorption capacity, and in restaurants the second factor dominates. In this pilot portfolio, the six locations with low territorial scores held 74% of overdue balance despite being rated medium risk through the financial route. Restaurant credit risk carries a geographic origin the traditional file never captures.

Why does conventional financial scoring fail with gastronomic MSMEs?

Because it measures the applicant's repayment capacity and ignores the territory's absorption capacity, and in restaurants the second factor dominates. In this pilot portfolio, the six locations with low territorial scores held 74% of overdue balance despite being rated medium risk through the financial route. Restaurant credit risk carries a geographic origin the traditional file never captures.

What does a territorial prefeasibility study cost and how long does it take?
Our operating threshold is hard: below 1.5% of project CapEx and delivered within ten business days. Above that price or that lead time, founders stop using the instrument and turn it into paperwork, which is exactly how a portfolio reaches 18.6% delinquency. The GovTech logic is precisely that: cheap, fast and binding on the disbursement decision.

What does a territorial prefeasibility study cost and how long does it take?

Our operating threshold is hard: below 1.5% of project CapEx and delivered within ten business days. Above that price or that lead time, founders stop using the instrument and turn it into paperwork, which is exactly how a portfolio reaches 18.6% delinquency. The GovTech logic is precisely that: cheap, fast and binding on the disbursement decision.

Is MTIE useful for a restaurant already open?
Yes, with a different purpose. In a running location it does not decide the opening, it decides the resizing: real schedule against flow by time band, seating against the polygon's revenue ceiling, and provisioning radius. Two of the 18 cases relocated within the same city and three cut seating. What measurement cannot do is recover CapEx already buried in an oversized fit-out.

Is MTIE useful for a restaurant already open?

Yes, with a different purpose. In a running location it does not decide the opening, it decides the resizing: real schedule against flow by time band, seating against the polygon's revenue ceiling, and provisioning radius. Two of the 18 cases relocated within the same city and three cut seating. What measurement cannot do is recover CapEx already buried in an oversized fit-out.

Data & sources

Sector data 2026 (official sources)

Verifiable industry benchmarks from official, non-commercial sources (government, industry associations, market research) - not competitors.

MetricBenchmark 2026Source
Metano de comida enterrada no capturado en vertederos de EE. UU.61% escapa a la atmósferaEPA — Quantifying Methane Emissions from Landfilled Food Waste 2023
Unidades económicas de la industria restaurantera en México 2023581.530 establecimientosINEGI — Censos Económicos 2024
Producción de la industria restaurantera mexicana por cada 100 pesos del sector55,9 de cada 100 pesosINEGI — Censos Económicos 2024
Peso de las microempresas en el total de unidades económicas de México 202395,4% del total (41,4% del personal ocupado)INEGI — Censos Económicos 2024
Peso de la agricultura familiar (pequeños productores) en América Latina y el Caribe81% de las explotaciones agrícolasFAO — State of Food and Agriculture 2024
Actividad emprendedora femenina en América Latina 202420,45% (la más alta del mundo)BID / Global Entrepreneurship Monitor 2024

Private audit of a gastronomic loan portfolio

If your institution finances gastronomic MSMEs and delinquency does not respond to financial scoring adjustments, the problem most likely sits in the territorial origin of the operation rather than in the file. Request a technical review of your credit line's prefeasibility method with Diego F. Parra and the SATE Institute team.

Author: Diego F. Parra  ·  Publisher: MASTERESTAURANT®
Content created with AI assistance, reviewed by the MASTERESTAURANT editorial team.
MR Comparison Engine v0.9.360