TANGLE- Evidence First Market Expansion Intelligence

Inspiration

Starting a business or expanding into a new market involves uncertainty. Entrepreneurs need to understand the local market, identify relevant businesses, assess geographic suitability, estimate operating costs, and validate whether their concept can attract customers.

However, cultural insights, business listings, and financial projections can easily be misunderstood as proof that a business will succeed. In reality, a popular place does not necessarily indicate demand for a new business, and a promising concept does not guarantee profitability.

This inspired us to build TANGLE — Evidence-First Market Expansion Intelligence, a decision-support platform that helps entrepreneurs investigate a potential market before committing capital.

Our central idea is simple: business expansion decisions should be based on traceable evidence, transparent calculations, and clearly identified uncertainties—not unsupported predictions.

What it does

TANGLE analyzes a proposed business expansion through geographic evidence, cultural and business signals, concept hypotheses, risk identification, and financial scenario calculations.

A. Target Geography and Data Telemetry

Users define the target city and choose the geographic scope of their analysis.

The platform supports: City-wide market: evaluates evidence at the broader city level. Neighborhood-specific market: supports a more localized submarket investigation. Demo Fixtures: provides a reproducible demonstration using predefined synthetic or fixture data. Live Qloo API: retrieves live cultural and place-related data from Qloo.

The analysis records the selected data mode and source provenance so users can distinguish live-source results from demonstration data.

B. Brand and Concept Hypothesis

Users can describe their proposed business through configurable fields, including: Business or brand name Business category Core concept and value proposition Target audience Menu or product direction Customer experience strategy

These inputs define the hypothesis being investigated. TANGLE uses the supplied category and available evidence to support preliminary screening and generate validation recommendations.

The platform does not assume that a concept is commercially viable merely because related businesses exist.

C. Geographic Evidence Matching

TANGLE evaluates the geographic metadata associated with returned entities and classifies the evidence by geographic relevance.

The dashboard distinguishes: Verified city match: the entity’s city metadata exactly matches the requested city. Partial match: the evidence has a partial geographic relationship. Global or non-local evidence: the record does not establish the requested local fit. Unknown geography: the available metadata is insufficient. Geographic mismatch: the record explicitly identifies a different geographic location.

Geographic Specificity Calculation

TANGLE measures the proportion of returned evidence records whose city metadata exactly matches the target city.

Formula:

Geographic Specificity Ratio = (Verified City Matches ÷ Total Evidence Records) × 100

This metric indicates how much of the evidence is verified at the city level. It does not establish customer demand, market size, or commercial success.

These classifications prevent nearby or unrelated locations from being silently treated as local evidence.

D. Category-Relevance Screening

The system screens returned entities against the proposed business category using a rule-based lexical matching process.

It distinguishes direct category matches from related matches where applicable. This helps organize potentially relevant records for further investigation.

However, a lexical match is not the same as a verified competitor, customer preference, or market-demand measurement. The screening process is a preliminary evidence filter rather than a complete competitor census.

E. Evidence Quality and Grounding Confidence

TANGLE presents an evidence-quality score and a breakdown of the evidence supporting the analysis.

The dashboard evaluates five dimensions:

Source quality: whether the information comes from the selected data source.

Geographic specificity: how clearly the evidence matches the target geography.

Entity relevance: how closely the returned records relate to the proposed business category.

Data freshness: how recently the evidence was observed.

Completeness: how much relevant information and category coverage is available.

The dashboard also displays supporting details such as verified-city counts, category coverage, timestamps, and unavailable affinity values.

The score communicates the grounding of the available evidence. It is not a probability of business success, a demand forecast, or a profitability prediction.

F. Key Findings and Concept Hypotheses

TANGLE synthesizes the retrieved evidence into concise findings, concept hypotheses, supporting records, and identified evidence gaps.

The findings separate observable facts—such as the number of records returned and the number with exact city metadata—from interpretations that still need validation.

Concept hypotheses can help users decide what to investigate next. Missing information about customer preferences, target audiences, menu choices, pricing, or experience strategy is identified rather than silently assumed.

G. Expansion Risk Radar

The platform identifies risks and presents them with severity levels, explanations, missing information, and recommended mitigations.

The dashboard organizes risks into categories such as: Demand validation Geographic evidence coverage Competitive context Operating economics Menu and product fit Customer acquisition Operational feasibility Data quality and source reliability Users can filter risks by severity to prioritize due diligence.

Each risk explains why the issue matters and what information is missing. Recommendations focus on practical next steps, including customer interviews, competitor research, site visits, price testing, supplier checks, and operational validation.

Risk severity represents the platform’s assessment of an identified uncertainty. It does not guarantee that a risk will occur or that mitigation will eliminate it.

H. Evidence Table and Provenance

TANGLE provides a detailed evidence table so users can inspect the records behind the analysis.

Available fields include: Entity name and evidence identifier Entity category Category-fit classification Geographic-match classification Source engine Affinity score, when available Popularity score, when available Provenance notes and limitations

The table supports searching and filtering evidence by category and geographic status.

This makes the analysis more transparent: users can trace findings back to individual records and understand why some records should not be used to justify local-market conclusions.

I. Financial Assumptions and Scenario Analysis

TANGLE allows users to enter their own financial assumptions, including: Initial capital budget Monthly rent Expected daily paying customers Average transaction value Monthly staffing costs Monthly ingredient costs

Monthly Revenue Scenario

TANGLE uses user-entered assumptions to calculate hypothetical revenue.

Formula: Monthly Revenue = Daily Footfall × Average Transaction Value × Operating Days per Month

Daily footfall represents the expected number of paying customers, while average transaction value represents the expected spending per customer.

Monthly Operating-Cost Calculation

The financial scenario combines the monthly expenses entered by the user.

Formula: Listed Monthly Operating Costs = Monthly Rent + Monthly Staffing Costs + Monthly Ingredient Costs

This subtotal includes only the listed cost categories. Additional expenses must be included separately if they are not already covered by the model.

Indicative Operating Surplus

TANGLE can use scenario revenue and the listed monthly operating costs to illustrate the financial effect of the user's assumptions.

Formula:

Indicative Operating Surplus = Monthly Revenue − Listed Monthly Operating Costs

This value is not necessarily net profit. Other expenses, including utilities, maintenance, marketing, taxes, depreciation, and financing costs, may affect actual profitability.

The result is a scenario estimate based on user inputs, not a verified demand forecast. The financial module uses these inputs to examine hypothetical business economics and sensitivity to changing assumptions.

It is important to distinguish user-provided assumptions from verified market data. The platform does not automatically establish local rent, customer footfall, menu pricing, or supplier costs simply by changing the target city.

J. Recommended Field Validation

The platform converts evidence gaps and risks into actionable research steps.

Depending on the identified gaps, these may include: Interviewing prospective customers Testing willingness to pay Mapping relevant competitors in the intended catchment Comparing competitor prices and offerings Visiting candidate locations Obtaining current rental and supplier quotations Testing customer-acquisition channels Checking licensing, staffing, and supply-chain requirements Refreshing evidence and verifying source reliability

The goal is to help entrepreneurs determine what they should verify before investing—not to replace primary market research.

K. Reporting and Export

The results dashboard provides an organized analysis report that can be printed or exported as JSON.

The report brings together the evidence summary, findings, confidence breakdown, risk assessment, concept hypotheses, detailed evidence, financial assumptions, and validation recommendations.

This gives users a record of the analysis that can be reviewed, shared, or used to plan further research.

L. QLOO Integration and How it Helps

TANGLE integrates the Qloo Insights API to retrieve live cultural intelligence and place-related data for a selected target market. This provides an external evidence source for investigating local businesses, places, and cultural signals relevant to a proposed expansion.

When a user selects Live Qloo API mode, TANGLE retrieves available data and processes the returned entities through its analysis pipeline. It evaluates geographic metadata to distinguish verified city matches from partial matches, unknown locations, and geographic mismatches. A rule-based category screen then checks the relevance of entity names, categories, and tags to the proposed business concept.

The resulting evidence supports TANGLE’s findings, geographic coverage analysis, evidence-quality assessment, and risk identification. When information is missing or unreliable, the platform highlights these limitations and recommends further validation instead of making unsupported assumptions.

How Qloo helps: Qloo supplies the external cultural and place-related intelligence, while TANGLE transforms the retrieved data into a structured, transparent market-expansion analysis. Qloo signals are used as supporting evidence—not as direct proof of customer demand, profitability, or business success.

How we built it

TANGLE combines a user-configurable analysis workflow, external data retrieval, evidence-processing rules, risk assessment, and a dashboard for presenting results.

Step 1: Capture the user's hypothesis

The interface collects the target geography, geographic scope, data mode, proposed business concept, and financial assumptions.

These inputs establish what the system should investigate and which assumptions should be treated as user-supplied.

Step 2: Retrieve and organize evidence

In live mode, the platform connects to the Qloo API to retrieve available place and cultural-related information.

In demo mode, it uses predefined fixtures to make the demonstration reproducible.

The system retains source information so that live telemetry and demo evidence can be distinguished.

Step 3: Normalize and classify geographic evidence

The returned records are processed to evaluate their geographic metadata against the requested location.

The resulting geographic classifications are retained separately rather than treating every returned record as a verified local match.

This is important when an API returns nearby places or records whose location metadata does not precisely match the requested city.

Step 4: Screen category relevance

The system applies a rule-based lexical screen to compare the proposed business category with available entity names, categories, and tags.

The results are classified for evidence review. The screen helps organize records but does not independently establish demand or market share.

Step 5: Assess evidence grounding

The evidence-quality process considers source quality, geographic specificity, entity relevance, freshness, and completeness.

The dashboard displays the resulting score alongside its component evidence, making it possible to inspect the reasons behind the score instead of relying on a number alone.

Step 6: Generate findings and hypotheses

The system summarizes observable evidence and identifies gaps that prevent stronger conclusions.

Where the information is incomplete, the platform surfaces the uncertainty and recommends research rather than treating missing data as a positive signal.

Step 7: Identify and prioritize risks

The risk-assessment layer organizes identified business uncertainties into categories and assigns severity levels.

Each risk is presented with its rationale, critical information gaps, and mitigation recommendations.

The resulting risk radar helps users prioritize the questions that should be answered before making a financial commitment.

Step 8: Calculate financial scenarios

The financial module uses the assumptions entered by the user to evaluate hypothetical revenue and cost relationships.

The results depend on the inputs. They should be interpreted as scenario arithmetic, not as forecasts based on independently verified local market conditions.

Step 9: Present an auditable report

The frontend organizes the analysis into findings, evidence-quality metrics, geographic coverage, risk cards, evidence details, and financial scenario information.

Users can inspect individual records, review the limitations, print the report, export JSON, or begin another analysis.

The system's design principle is to make the evidence and assumptions visible, allowing users to challenge the analysis rather than blindly accept its conclusions.

Challenges we ran into

A. Separating cultural information from commercial demand

Cultural and place-related data can provide useful context, but it does not directly prove that customers will purchase a particular product.

We addressed this limitation by distinguishing cultural or place evidence from observed demand and by recommending direct customer validation.

B. Geographic ambiguity

A result associated with a nearby town or another location should not automatically be treated as evidence for the target city.

TANGLE therefore classifies geographic matches explicitly and retains mismatched records as mismatched evidence instead of silently counting them as local.

C. Missing affinity scores

Some returned records may not contain affinity scores. Popularity values, when available, are a separate metric.

The platform exposes unavailable scores and documents their limitations rather than substituting popularity for affinity or inventing missing values.

D. Measuring evidence quality without overstating confidence

A high evidence-quality score can be misunderstood as proof that a business will succeed.

We addressed this by explaining the dimensions of evidence quality and clearly separating evidentiary grounding from commercial success.

E. Identifying competitors accurately

A business that matches a category keyword is not necessarily a direct competitor. It may have a different offering, price range, target audience, or operating model.

The category screen therefore acts as an initial filter. Detailed competitor mapping and verification remain necessary.

F. Building meaningful financial scenarios

Financial calculations depend on assumptions about customers, average spending, rent, staffing, and ingredients. Incorrect assumptions can make a scenario misleading even when the arithmetic is correct.

TANGLE keeps these inputs explicit and identifies the need to validate costs and demand independently.

G. Turning analysis into useful action

A dashboard that only presents scores is not enough to support a decision.

TANGLE connects its findings to risk categories, information gaps, and field-validation steps so that users can move from reviewing evidence to collecting missing information.

Accomplishments that we're proud of

A. Evidence-first design: The platform prioritizes traceable evidence and explicit limitations instead of presenting business success as a certainty.

B. Live data integration: The application supports live Qloo data retrieval as well as a reproducible demo-data mode.

C. Geographic transparency: Local matches, partial matches, non-local evidence, unknown locations, and explicit mismatches can be distinguished.

D. Inspectable evidence: The detailed evidence table links the analysis to individual records, categories, sources, scores when available, and limitations.

E. Multi-dimensional evidence assessment: The dashboard communicates source quality, geographic specificity, relevance, freshness, and completeness.

F. Structured risk assessment: Business uncertainties are organized into risk categories with severity, missing information, and recommended mitigations.

G. Actionable recommendations: The report turns evidence gaps into concrete validation activities, including customer interviews, competitor research, and site checks.

H. User-controlled financial assumptions: Entrepreneurs can test different economic assumptions without mistaking them for verified local estimates.

I. Reusable reporting: Users can print reports and export structured JSON for review and further analysis.

J. Transparent limitations: The platform explicitly communicates that category matching, place popularity, and cultural signals do not establish customer demand or commercial viability.

Our main accomplishment is creating a structured decision-support workflow that helps entrepreneurs understand both what the available evidence says and what it cannot yet prove.

What we learned

A. Evidence is only useful when its limitations are understood

B. Geographic precision changes the quality of a decision

C. Cultural signals are not demand measurements

D. Transparent uncertainty is more useful than false precision

E.Financial calculations are only as reliable as their assumptions

F. Risk identification should lead to validation

G. Decision-support systems should support human judgment

What's next for TANGLE- Evidence First Market Expansion Intelligence

A. Stronger local-market intelligence

Improve neighborhood-level evidence matching and add more precise catchment analysis so users can investigate individual commercial areas rather than relying primarily on city-level metadata.

B. Verified competitor analysis

Expand the competitor workflow to verify business locations, offerings, pricing, customer segments, and differentiation through reliable sources and field research.

C. Primary customer research

Introduce a structured workflow for recording interviews, customer surveys, willingness-to-pay responses, and pilot-test results. This would help distinguish actual demand evidence from indirect cultural signals.

D. More robust financial modeling

Extend scenario analysis to include additional operating expenses, taxes, utilities, maintenance, marketing, working capital, and financing costs where appropriate.

E. Stronger explainability

Provide more detailed explanations of how evidence scores, geographic classifications, category matching, and risk severities are determined, with traceable reasons for each result.

F. Decision tracking

Allow entrepreneurs to record validation results, update assumptions, compare analysis versions, and document why they proceeded with, modified, or rejected an expansion idea.

Built With

• css
• express.js
• html
• javascript
• jwt
• node.js
• python
• qloo
• react
• restapi
• typescript
• vite