Skip to main content
Locatalyze
Sign in
Start Free Location Score

Data and Methodology

Know what is observed, estimated and modelled.

Locatalyze separates external evidence, customer assumptions and modelled outputs. A rules-based engine calculates the score and recommendation; AI helps explain what the result means and what to verify next.

Start a free location analysisView sample report
Methodology
Engine 6.0.0
Last reviewed
30 July 2026
Typical report
typically 2–4 minutes

Method at a glance

A traceable path from address to action.

The report separates evidence gathering, deterministic calculation and written interpretation. Each stage has a different job, and no narrative step is allowed to silently replace the scoring rules.

  1. 01

    Define the decision

    You provide the address, business type, proposed rent and any commercial assumptions you know.

  2. 02

    Resolve the location

    The address is geocoded and nearby providers are queried around the confirmed pin.

  3. 03

    Classify the evidence

    Each input is treated as observed, user-provided, estimated, proxy-based or unavailable.

  4. 04

    Score five dimensions

    A rules-based engine calculates rent, competition, demand, profitability and location-quality scores.

  5. 05

    Apply decision gates

    Economics, coverage, contradictions and missing commercial inputs can make the recommendation more conservative.

  6. 06

    Explain the result

    AI turns the structured output into readable risks, conditions and verification actions; it does not set the score.

  7. 07

    Support due diligence

    The report shows what to validate before signing rather than presenting the output as professional advice.

Evidence language

One label for every kind of input.

These terms distinguish where a value came from. A source label describes provenance; confidence describes how much the report can rely on it.

User-provided input
Entered by the customer and accepted as an assumption; not independently verified by Locatalyze.
Observed external data
Returned by a named provider or public dataset for the location or surrounding area.
Benchmark / estimate
A category, area or listing-based estimate used when a site-specific value is unavailable.
Proxy-based
An indirect signal used in place of a direct measurement, such as nearby activity standing in for footfall.
Modelled scenario
Calculated from inputs, benchmarks and deterministic formulas; it is not an observed trading result.
AI-written
Narrative generated from structured results, risks and validation checks.
Missing / unavailable
The system could not resolve the data reliably enough to present it as evidence.

What goes into a report

Every input has a source and a limit.

Site-specific does not always mean site-measured. This matrix shows the normal primary source, fallback, geographic precision and decision effect for each major input category.

InputSource and fallbackPrecision and limitationReport effect

Address and coordinates

User-provided inputObserved external data

Typed address plus latitude and longitude returned after confirmation.

PrimaryLocationIQ geocoding and autocomplete; Mapbox renders the map.

FallbackExisting submitted coordinates or pipeline geocoding where available.

Typical precisionStreet-level when the geocoder resolves a valid pin; otherwise area-level context is disclosed.

Main limitationA geocoded pin is not a cadastral survey and can be wrong for suites, arcades or new developments.

All radius searches, the map and location-quality signals.

Competition

Observed external dataProxy-based

Nearby businesses, category match, rating, review volume, distance and listing status where returned.

PrimaryGoogle Places, with Geoapify/OpenStreetMap support; Foursquare is used for fitness categories.

FallbackCategory-specific provider fallbacks, including OpenStreetMap coverage for some fitness searches.

Typical precisionThe scored count is anchored to 500 m; provider discovery can search a wider area.

Main limitationListings can be missing, duplicated, miscategorised, closed or not yet published.

Competition score, saturation, opportunity notes and the competitor map.

Demographics

Observed external dataBenchmark / estimateMissing / unavailable

Population, household and income context when the relevant Australian public-data lookup resolves.

PrimaryABS Census 2021 data and stored Australian geographic mappings.

FallbackBroader area context or an explicit unavailable state.

Typical precisionSA1, postcode or suburb context depending on what resolves for the report.

Main limitationArea demographics do not describe the exact frontage, daypart or customer catchment.

Demand context, customer fit and confidence.

Demand and street activity

Observed external dataProxy-basedBenchmark / estimate

Demand trend, nearby anchors, transit, local activity and category-fit signals.

PrimaryPipeline market signals plus pin-anchored place, anchor and transit data.

FallbackCompetitor-density and category proxies; the demand weight is reduced when the signal is proxied.

Typical precisionDirectional trade-area context, not a verified pedestrian count at the door.

Main limitationMost reports do not contain a measured, continuous foot-traffic series.

Demand score, location quality, opportunity notes and confidence.

Rent pressure

User-provided inputBenchmark / estimateModelled scenario

Submitted monthly rent, estimated comparison context and rent as a share of modelled revenue.

PrimaryThe customer’s rent quote plus commercial-listing and area/category signals returned by the pipeline.

FallbackCategory cost-stack benchmarks when current comparable evidence is thin.

Typical precisionThe quote is site-specific; the comparison context is usually suburb or corridor level.

Main limitationPublic listings are not a complete rent database and rarely capture incentives, outgoings or lease condition.

Rent score, rent-burden status, recommendation gates and financial scenarios.

Revenue and costs

User-provided inputBenchmark / estimateModelled scenario

Ticket size, customers, COGS, labour, rent, other operating costs, setup budget and working-capital assumptions.

PrimaryUser inputs plus location-agent estimates checked against Australian category benchmarks.

FallbackCategory defaults, clearly labelled and subject to more conservative recommendation rules.

Typical precisionMonthly model outputs; low-confidence values are displayed as wider bands.

Main limitationThese are commercial estimates, not audited accounts, quotes or guaranteed performance.

Profitability score, break-even, payback, scenarios and recommendation.

Data confidence

Modelled scenario

Feed presence, provider coverage, benchmark reliance and material corrections made by the engine.

PrimaryDeterministic confidence and validation logic.

FallbackA lower confidence tier, wider display range or missing-data state.

Typical precisionA report-level percentage plus topic-level confidence labels.

Main limitationConfidence describes evidence coverage; it is not an accuracy guarantee.

How figures are shown and whether a positive recommendation is allowed.

Location Score

Five dimensions. One rules-based composite.

Each dimension is scored from 0–100, multiplied by its current weight and added. The base weights total 100%; proxy demand uses a deliberately smaller contribution.

20%Rent affordability
25%Competition
20%Market demand
25%Profitability
10%Location quality
20%base weight

Rent affordability

User rent; modelled revenue; category cost stack.

Increases with
Lower sustainable rent burden against realistic revenue.
Decreases with
Rent consumes an increasing share of modelled revenue.
25%base weight

Competition

Provider listings, category match, reviews and distance.

Increases with
Fewer relevant, high-threat competitors with verified coverage.
Decreases with
More relevant competitors, stronger incumbents and closer proximity.
20%base weight

Market demand

Market-agent signals, public context, anchors and proxies.

Increases with
Stronger resolved demand, demographic fit and supporting location activity.
Decreases with
Weak or declining signals; a neutral internal assumption is used when the display value is unavailable.
25%base weight

Profitability

Modelled revenue less COGS, labour, rent and other costs.

Increases with
A stronger positive monthly net result after the modelled cost stack.
Decreases with
Thin profit, break-even or losses; unverified inputs cap the axis.
10%base weight

Location quality

Pin-anchored access, transit, anchor and activity context.

Increases with
Better access, transit, relevant anchors and supporting activity.
Decreases with
Poor access or weak local activity signals.

Worked example

A transparent composite, not a hidden grade.

Illustrative measured-demand case. It is not a customer report and does not imply a recommendation.

Rent affordability70× 20%
Competition65× 25%
Market demand60× 20%
Profitability70× 25%
Location quality80× 10%
70 × 20% + 65 × 25% + 60 × 20% + 70 × 25% + 80 × 10%= 68/100

The engine rounds the final weighted sum to a whole number. Two nearby sites can differ because their confirmed pin, competitor set, rent, access, user assumptions or fallback state differs.

Score versus recommendation

Three outputs answer three questions.

The score describes the weighted location read. The recommendation asks whether the current commercial case supports action. Confidence describes the evidence behind both.

01

Location Score

How the five dimensions combine.

A weighted 0–100 composite. It is rounded after the axis contributions are added.

02

Recommendation

What the current case supports.

A rules-based action call after score bands, economics and trust gates are applied.

03

Data confidence

How much evidence backs the read.

A separate coverage and model-quality signal. It is not added to the Location Score.

PROCEED60–100 base score

The score, economics and evidence gates support progressing, subject to normal lease and professional due diligence.

VERIFY40–59 base score

The site may work, but one or more commercial, evidence or competition checks must be resolved first.

AVOID0–39 base score

The current inputs show a material mismatch in economics, competition, location or the operating model.

INSUFFICIENT DATANo decision-grade base

Core commercial inputs or critical report evidence are missing, so the product withholds a normal recommendation.

Why the recommendation can be harsher

A good composite cannot hide a decision-critical weakness.

The score band is a ceiling on how favourable the recommendation can be. The confirmed rules below only keep or lower that result; they do not turn a weak score into PROCEED.

  • Benchmark-only revenue or missing ticket/setup inputs block a PROCEED recommendation.
  • If both location/rent and demand feeds are missing, the result cannot read PROCEED.
  • Data completeness below 35% cannot support a positive display verdict; the engine’s stricter hard gate requires more than 45% for PROCEED.
  • Unknown or zero-result competitor coverage is treated as unverified, not as evidence of a blue ocean.
  • A location-grounded loss cannot be turned into PROCEED solely because a category benchmark is more optimistic.
  • A user budget below 75% of a credible recommended setup blocks PROCEED; below 50% forces AVOID.
  • Negative base-case profit, serious rent pressure, fatal competition/profitability reads or validation contradictions can make the final recommendation harsher than the score band.

Financial methodology

A modelled commercial estimate—not a promise.

The paid financial view combines customer assumptions, category benchmarks and validated pipeline estimates. It calculates revenue, costs, break-even, sensitivity and payback, but it is not a full set of accounts.

Base revenue

daily customers × hours multiplier × access multiplier × average ticket × 30

User ticket and calibration inputs take priority; category values fill gaps. Location-agent estimates are plausibility-checked and may be blended with benchmarks.

Gross margin and COGS

revenue × category gross-margin / COGS rate

Category rates are defaults, not supplier quotes. Implausible upstream values are corrected or rejected.

Monthly net

modelled revenue − COGS − labour − rent − other operating costs

This is a commercial estimate rather than a full accounting P&L. Tax, finance structure and owner circumstances are not fully modelled.

Break-even customers per day

fixed monthly costs ÷ contribution per transaction ÷ 30

Contribution per transaction is average ticket after variable COGS. The value is withheld if the required inputs are not usable.

Payback period

setup cost ÷ positive monthly net profit

Only shown when setup cost is greater than zero and monthly net profit is positive.

Scenario range

worst 65% revenue · base 100% · best 140%

Variable costs move with revenue. Fixed costs stay fixed, with a small stress/efficiency adjustment. These scenarios are sensitivities, not forecasts.

Workable screening band8–18%

Engine estimate from the category cost stack—not a universal lease rule.

Healthy
Up to 14%
Watch
14–18%
Risky
Above 22%

Free location signal

What you receive before an unlock

  • PROCEED / VERIFY / AVOID recommendation
  • Competitor map (500 m radius)
  • Headline location score (0–100)
  • Data confidence and risk/demand insights
  • Basic research tools (guides & calculators)

Paid report · $29

What the financial unlock adds

  • Modelled revenue and cost breakdown
  • Break-even estimate (customers per day)
  • Best, base & worst case scenarios
  • Professional PDF decision report
  • Deeper financial analysis & simulators

What AI does

AI explains the model. It does not own the decision.

The language layer is useful for turning structured outputs into plain-English risks and actions. Keeping its boundary visible prevents readable prose from being mistaken for numeric evidence.

AI may

Explain structured evidence.

  • Turns structured scores, risks and scenarios into readable narrative.
  • Summarises competitive positioning and SWOT-style interpretation where supported.
  • Drafts conditions, failure modes and actions to verify next.
  • May contribute upstream estimates that are labelled and checked against deterministic bounds before use.

AI does not

Control the decision rules.

  • Set the five score weights or the score-band thresholds.
  • Independently calculate the final Location Score.
  • Override deterministic recommendation and validation gates.
  • Convert an estimate into a measured fact or guarantee a trading outcome.

Confidence and fallbacks

Uncertainty changes the output.

Missing data is not silently treated as a good result. Depending on the gap, the engine uses a labelled fallback, reduces a weight, widens a range, lowers confidence or withholds the normal recommendation.

A

Data completeness

Data completeness measures how much usable external and commercial evidence resolved, then subtracts for rejected values, operational corrections and fallback-heavy inputs.

B

Model confidence

Model confidence is a High / Medium / Low-style classification of how much the commercial model depends on observed feeds versus estimates and corrections.

C

Location Score

A score describes the modelled location read. Confidence describes how much reliable evidence supports that read. They answer different questions and are never interchangeable.

Limits and due diligence

What you should still verify yourself.

Locatalyze is an early decision-support layer. A responsible lease decision still requires physical observation, current quotes, professional advice and inspection of the property and lease.

Known limitations

What the report cannot know reliably.

  • Public and third-party data can be incomplete, delayed or unavailable.
  • Competitor listings can be missing, duplicated, miscategorised or out of date.
  • Street activity is often proxy-based; it must not be read as a measured pedestrian count.
  • Commercial rent depends on incentives, outgoings, lease terms, fit-out and property condition.
  • Demographic context may describe a wider area than the site’s true customer catchment.
  • Financial results are highly sensitive to the rent, ticket, volume, staffing and setup assumptions.
  • The model cannot reliably predict future competitors, roadworks, developments or demand shocks.
  • A report does not inspect the property, lease, licences, planning position or physical condition.

Before signing a lease

Turn the report into a verification plan.

  1. 01Count foot traffic and stopping behaviour at the actual frontage across weekday, weekend and relevant dayparts.
  2. 02Verify base rent, incentives, outgoings, review clauses, permitted use and make-good obligations.
  3. 03Walk the catchment and confirm which competitors are open, relevant and trading strongly.
  4. 04Check planning, zoning, food, liquor, signage, accessibility and other licence requirements.
  5. 05Replace category labour, supplier, fit-out and utility assumptions with current quotes.
  6. 06Stress-test revenue using your capacity, conversion rate, ticket and realistic ramp-up.
  7. 07Ask an accountant to review the commercial model and working-capital allowance.
  8. 08Ask a commercial leasing lawyer to review the lease before signing.
  9. 09Inspect the property and check the local development, vacancy and construction pipeline.

Useful next steps: opening checklist, break-even foot-traffic tool and Expert Business Review.

Sources and reference

Named providers, update policy and plain language.

“Current” means the provider was queried for the report, not that every record is complete or real-time. Source responses may be cached, and model versions change only when the decision logic is deliberately updated.

Provider and categoryReference and refreshCoverage and limitation

LocationIQ opens in a new tab

Address autocomplete and geocoding

ReferenceQueried when the address is confirmed

RefreshPer confirmation, with application caching

CoverageAddress or area, depending on match quality

LimitationNot survey-grade; complex premises can resolve imperfectly

Google Places opens in a new tab

Competitor and place listings

ReferenceCurrent provider response

RefreshQueried for a report; responses may be cached

CoveragePin-centred; competition scored within 500 m

LimitationA listing service is not a complete register of active businesses

Geoapify / OpenStreetMap opens in a new tab

Places, anchors, transport and fallback coverage

ReferenceCurrent provider/community dataset

RefreshQueried for a report; responses may be cached

CoveragePin-centred radii

LimitationCommunity-edited records may be incomplete or stale

Foursquare opens in a new tab

Fitness-category place discovery

ReferenceCurrent provider response

RefreshUsed for applicable categories

CoveragePin-centred radius

LimitationCategory coverage varies by market

Reference2021 Census reference data

RefreshUpdated when the underlying public-data integration is revised

CoverageSA1, postcode or suburb context where resolved

LimitationHistorical area data is not an exact-site customer survey

Commercial-listing and category signals

Rent comparison context

ReferenceSignals available to the report pipeline

RefreshPer report where available

CoverageUsually suburb or corridor level

LimitationNot a complete market database or a broker valuation

Locatalyze compute engine

Score, financial model, confidence and recommendation

ReferenceEngine 6.0.0

RefreshVersioned when decision logic changes

CoverageDeterministic calculation from the report inputs

LimitationModel quality remains constrained by input quality

Glossary

Terms used across the product.

Location Score

A weighted 0–100 composite of five location and commercial dimensions.

Recommendation

The action label PROCEED, VERIFY, AVOID or INSUFFICIENT DATA. It can be more conservative than the score band.

Data completeness

The share and quality of usable evidence after fallback and correction penalties.

Model confidence

A qualitative read of how strongly the financial and location model is supported.

Rent burden

Monthly rent divided by modelled monthly revenue.

Workable rent band

The engine’s 8–18% rent-to-revenue screening range; it is a model benchmark, not a universal rule.

Competition intensity

Pressure from relevant nearby competitors, considering count, strength and proximity.

Demand signal

A directional read from market, demographic, anchor and activity evidence.

Location quality

Access, transit, nearby anchors and local activity around the pin.

Break-even

The transaction volume required for contribution margin to cover fixed modelled costs.

Contribution margin

Average ticket less variable cost of goods for that transaction.

Modelled estimate

A calculated value produced from inputs and assumptions rather than observed accounts.

Observed external data

Information returned by a named third-party or public source.

Proxy

An indirect signal used when the direct measurement is unavailable.

Failure condition

A defined change or threshold that would make the current case no longer viable.

Common questions

The shortest version of the method.

Does AI calculate the Location Score?

No. A deterministic compute engine applies the scoring weights and recommendation gates. AI writes parts of the explanation and may contribute labelled upstream estimates, but it cannot independently set or override the final score.

Why can a 60+ score still receive VERIFY?

The score is a weighted composite, while the recommendation also checks commercial inputs, economics, competition coverage, confidence and contradictions. A positive-looking composite cannot override a missing or serious decision risk.

Is foot traffic measured at the shopfront?

Usually not. Most reports use nearby activity, anchors, transit, provider signals or other proxies. The report should label these as proxy-based and users should conduct physical counts before signing.

Are the financial outputs forecasts?

They are modelled scenarios built from user inputs, provider signals and category benchmarks. They are decision-support estimates, not audited forecasts, guaranteed revenue or professional financial advice.

Apply the method to an address

Start with the evidence. Decide what to verify.

Your first location signal includes the score, recommendation, data confidence and competitor map. No card is required to begin.

Start a free location analysisView the full sample report