Architecture swarm · Territory-analysis + market-scoring agent · Cannibalization-risk-scoring skill · Build pillar · Published July 11, 2026

How to build on-demand cannibalization risk scoring at multi- location scale

Multi-location franchise systems evaluating a candidate site need to know how much volume the new unit will pull from existing franchisees within commuting distance. Buxton + SiteZeus + Tango Analytics + Mosaiq + Placer.ai Forecast + Esri Business Analyst + Pitney Bowes MapInfo ship quarterly trade-area studies that the development team commissions per site. By the time the study lands, the candidate has often already progressed through letter of intent or lease negotiation. The Model + Score + Gate + Audit skill bundle on the territory-analysis-market-scoring agent returns a tiered cannibalization risk score with named affected existing locations, revenue-impact estimate, time-to-recovery, and an FDD Item 12 territorial-protection attestation in minutes. Real regulatory anchors preserved in every audit record: FDD Item 12 + 15-state franchise registration + ECOA Regulation B disparate-impact + Fair Housing + FTC v X-Mode / Mobilewalla / Kochava + Massachusetts AG v Copley Advertising 2017 + EU AI Act Annex III + Article 22 + 50 + NIST AI RMF.

Start with a Tier 1 AI Readiness Assessment Tier 3 Fractional CMO with AI Swarm Take the 3-question quiz

The 4-skill bundle on the territory-analysis + market-scoring agent

Model

Per-candidate trade-area mathematics. Radius bands (half-mile, 1, 2, 3, 5, 10, 15, 20 mile); drive-time isochrone (5, 10, 15, 20, 30 minute) via Mapbox Directions + Google Distance Matrix + HERE Routing + TomTom Routing + Esri Network Analyst; walk- time and transit-time isochrone where urban; Huff gravity model with operator-specific distance-decay parameter (the canonical retail trade-area math); MCI multiplicative competitive interaction; Reilly-Converse spatial interaction; Voronoi tessellation + Thiessen polygon for tessellated market-area allocation; LBC location-based clustering + DMA / CBSA overlay. Geospatial substrate on PostGIS + H3 + S2 + Uber Movement + ArcGIS + QGIS.

Score

Three coordinated subsystems compose into a risk tier (no-risk + low-risk + medium-risk + high-risk + very-high-imminent). Overlap estimation (Jaccard + Sorensen-Dice + Tversky + Huff expected-share + MCI expected-share + Voronoi + Thiessen with confidence intervals via bootstrap + Bayesian posterior + Monte Carlo simulation + sensitivity analysis). Demographic cohort weighting (Census ACS one-year + five-year + Esri Tapestry + Claritas PRIZM + Experian Mosaic + Nielsen PRIZM + operator internal segment with cross-cohort overlap). Foot- traffic co-visitation (Placer.ai + SafeGraph + Foursquare + Veraset + Cuebiq with per-day-part + per-day-of-week time windows). Per-candidate confidence interval + per-candidate explainability (SHAP + LIME + feature importance + counterfactual + attribution graph + causal DAG + multi-LLM narrative under per-vendor zero-retention).

Gate

Five anchors before delivery. FDD Item 12 territorial-protection per FTC Franchise Rule 16 CFR 436 + NASAA Commentary + 15- state franchise registration + 7-additional-state disclosure + 20-state franchise relationship laws + FDD Item 17 + 19. ECOA Regulation B + Fair Housing Act disparate-impact + FCRA (algorithmic site-selection cannot use or proxy for protected class). FTC v X-Mode Social / Outlogic January 2024 + FTC v Mobilewalla December 2024 + FTC v Kochava + Massachusetts AG v Copley Advertising April 2017 + Illinois BIPA + Texas CUBI + WA My Health My Data Act 2024 + Texas SCOPE Act 2024 (foot- traffic data input). EU AI Act Annex III high-risk classification (capital deployment + lease commitment) + Article 9 + 13 + 14 + 15 + 22 + 26 + 50; CCPA + CPRA + 17- state + GDPR Article 22. FTC Section 5 substantiation when referenced in franchisee-facing FPR / FDD Item 19 + AICPA SSAE attestation + audit committee defensibility + NIST AI RMF + ISO 42001 + per-vendor LLM zero-retention. Policy-as- code via OPA Rego + AWS Cedar + Casbin + Cerbos + Oso + Styra DAS + Permit.io.

Audit

Per-candidate WORM record: candidate ID + banner pointer + franchise-system pointer + canonical trade-area model snapshot + per-existing-location overlap record + demographic cohort weighting record + foot-traffic co-visitation record + FDD Item 12 attestation (affected franchisee list + protection type) + multi-LLM classification snapshot + per-candidate confidence interval + per-candidate explainability + per- anchor Gate decision with evidence + sibling-handoff pointers + FBC feedback loop (realized vs predicted feedback + trade- area model recalibration + overlap recalibration + demographic weighting recalibration + co-visit recalibration + Item 12 attestation recalibration + confidence recalibration). Storage on AWS S3 Object Lock + Azure Blob immutable + Google Cloud Storage Bucket Lock + Wasabi WORM. Retention stacks: 7-year FTC + 7-year IRS + 7-year FDD record + per-state franchise registration retention + 3-year Illinois BIPA biometric + GDPR Article 30 + EU AI Act Article 12 + SOC 2 CC7 / CC8.

The real vendor ecosystem this sits above

Site-selection + geospatial

Buxton, SiteZeus, Tango Analytics, Mosaiq, Placer.ai Forecast, RetailNext, Esri Business Analyst, Pitney Bowes MapInfo, GeoBlue, RegradesIQ, Synergos, Kalibrate, Geomatic AI site- selection platforms remain the per-engagement substrate. ArcGIS + QGIS + PostGIS + H3 + S2 + Uber Movement geospatial + Mapbox Directions + Google Distance Matrix + HERE Routing + TomTom Routing + Esri Network Analyst routing for isochrone generation.

Foot-traffic + demographic + LLM ensemble

Placer.ai + SafeGraph + Foursquare + Veraset + Cuebiq foot- traffic; Census Bureau ACS + Esri Tapestry + Claritas PRIZM + Experian Mosaic + Nielsen PRIZM demographic; OpenAI + Anthropic + Google + Mistral + Cohere LLM providers under per-vendor zero-retention back Score multi-LLM narrative explanation; LangSmith + Weights & Biases + Arize + WhyLabs + Helicone + Langfuse + PromptLayer + Galileo observability.

Policy-as-code + WORM + sibling skills

OPA Rego + AWS Cedar + Casbin + Cerbos + Oso + Styra DAS + Permit.io policy-as-code expresses every Gate rule. AWS S3 Object Lock + Azure Blob immutable + Google Cloud Storage Bucket Lock + Wasabi compliance WORM holds the per-candidate audit substrate. Sibling skills on the same agent: cannibalization-analysis (parent commercial pillar); site- selection software; market-opportunity analysis; competitor mapping; franchise-territory mapping; franchise-registration states; per-prospect franchise-territory analysis pipeline (sibling build-pillar); foot-traffic integration (sibling build-pillar); per-location demographics; demographic data ingestion; commercial real estate API; cross-location cannibalization detection (sibling build-pillar).

The 6-workstream reporting cycle

Numeric uplift commitments are not made up-front. The engagement ships a pre-engagement baseline across six workstreams; the cycle tracks delta against that baseline. Reporting is the substrate, not the promise.

Model coverage. Per-trade-area family coverage across radius bands + drive-time / walk-time / transit-time isochrones + Huff + MCI + Reilly-Converse + Voronoi + Thiessen + LBC + DMA / CBSA overlay; per-routing-provider availability + per-geospatial-substrate adherence.
Score quality. Per-candidate confidence-interval coverage (bootstrap + Bayesian + Monte Carlo + sensitivity); demographic cohort weighting completeness; foot-traffic co- visitation provider availability; multi-LLM agreement rate; risk-tier distribution across the candidate pipeline.
Gate quality. Per-anchor evaluation completeness (FDD Item 12 + 15-state franchise registration + ECOA + Fair Housing + FTC location-data precedent + EU AI Act Annex III + FTC Section 5 substantiation); per-anchor pass / fail / route- to-counsel distribution; FDD Item 12 attestation cadence; ECOA / Fair Housing disparate-impact audit cadence.
Audit quality. Per-candidate WORM record completeness; retention-window coverage (longest of 7-year FTC + 7-year IRS + 7-year FDD + per-state franchise registration + 3-year Illinois BIPA + GDPR Article 30 + EU AI Act Article 12 + SOC 2 CC7 / CC8); end-to-end replay success rate.
Compliance posture. FDD Item 12 attestation coverage across the candidate pipeline; 15-state franchise registration posture; foot-traffic data-provider posture against FTC v X-Mode / Mobilewalla / Kochava precedent; ECOA / Fair Housing disparate-impact audit posture; EU AI Act Annex III + Article 50 disclosure coverage when AI-driven scoring drives capital deployment.
Audit-trail completeness. Per-anchor regulatory citation completeness; FBC feedback-loop recalibration cadence across realized vs predicted volume transfer; sibling-handoff pointer completeness into the territory-analysis-market- scoring bundle (cannibalization-analysis commercial pillar + site-selection software + market-opportunity analysis + competitor mapping + franchise-territory mapping + franchise- registration states + per-prospect franchise-territory analysis pipeline + foot-traffic integration + per-location demographics + demographic data ingestion + commercial real estate API + cross-location cannibalization detection).

Frequently asked questions

What is on-demand cannibalization risk scoring at multi-location scale — and why is quarterly Buxton not enough?

Multi-location franchise systems evaluating a candidate site face the question of how much volume the new unit will pull from existing franchisees within commuting distance. Buxton, SiteZeus, Tango Analytics, Mosaiq, Placer.ai Forecast, RetailNext, Esri Business Analyst, Pitney Bowes MapInfo, GeoBlue, RegradesIQ, Synergos, Kalibrate, and Geomatic AI ship quarterly trade-area studies that the development team commissions on a per-site basis. By the time the study lands, the candidate has often progressed through letter of intent, lease negotiation, or has already lost optionality to a competing developer. On-demand cannibalization risk scoring closes the gap: a candidate site enters the territory-analysis-market-scoring agent and the Model + Score + Gate + Audit four-skill bundle returns a tiered risk score (no-risk + low-risk + medium-risk + high-risk + very-high-imminent) with named affected existing locations, revenue-impact estimate, time-to-recovery estimate, and an FDD Item 12 territorial-protection attestation in minutes — with the audit substrate retained for franchisee-facing defense if a territorial dispute arises.

Why do Buxton + SiteZeus + Tango Analytics + Mosaiq + Placer.ai Forecast + Esri Business Analyst break at multi-location franchise scale?

Each site-selection platform ships a per-account per-quarterly-trade-area-study primitive — a paid engagement that lands a study deck per candidate site. None coordinates the FDD Item 12 territorial-protection attestation that franchise systems require before signing a new franchisee within a protected territory. None composes the per-existing-location overlap estimation across the standing Huff gravity model + multiplicative competitive interaction (MCI) + Reilly-Converse spatial interaction + Voronoi tessellation + Thiessen polygon families. None integrates Placer.ai + SafeGraph + Foursquare + Veraset + Cuebiq foot-traffic co-visitation data with the trade-area model. None enforces ECOA Regulation B + Fair Housing disparate-impact analysis when site-selection scoring proxies for protected class via ZIP code or demographic. None enforces the EU AI Act Annex III high-risk classification that applies when AI-driven site-selection drives a major investment decision. The four-skill bundle Model + Score + Gate + Audit sits above the vendor primitives — it does not replace them. Model runs the trade-area mathematics. Score composes overlap + demographic weighting + foot-traffic co-visitation into a risk tier. Gate enforces FDD Item 12 + state franchise registration + ECOA / Fair Housing + FTC location-data precedent + EU AI Act Annex III. Audit writes a per-candidate WORM record.

What does Model do — trade-area mathematics (Huff + MCI + Reilly-Converse + Voronoi + isochrones)?

Model computes per-candidate trade-area boundaries across multiple families. Radius bands (half-mile, 1, 2, 3, 5, 10, 15, 20 mile) for coarse comparison. Drive-time isochrone (5, 10, 15, 20, 30 minute) computed against current traffic via Mapbox Directions API + Google Distance Matrix + HERE Routing + TomTom Routing + Esri Network Analyst. Walk-time and transit-time isochrone where the trade area is urban. Huff gravity model with operator-specific distance-decay parameter (the canonical retail trade-area math). Multiplicative Competitive Interaction (MCI) model when multiple competitors compete for the same trade area. Reilly-Converse spatial interaction when the trade area spans population centers. Voronoi tessellation and Thiessen polygon for tessellated market-area allocation across the existing portfolio. Location-based clustering (LBC) and trade-area DMA / CBSA overlay. Each trade-area model carries a confidence tier and an explainability trace written into Audit. Geospatial substrate runs on PostGIS + H3 + S2 + Uber Movement + ArcGIS + QGIS for hex / polygon arithmetic.

What does Score do — overlap estimation + demographic cohort weighting + foot-traffic co-visitation + risk tier?

Score composes three coordinated subsystems into a per-candidate risk tier. Overlap estimation against existing locations runs Jaccard + Sorensen-Dice + Tversky set-overlap metrics on the trade-area polygons; Huff expected-share and MCI expected-share to estimate volume transfer; Voronoi tessellation and Thiessen polygon for tessellated overlap; with confidence intervals via bootstrap + Bayesian posterior + Monte Carlo simulation + sensitivity analysis. Demographic cohort weighting pulls per-block-group counts from Census Bureau ACS (one-year + five-year vintages) and overlays Esri Tapestry segmentation + Claritas PRIZM + Experian Mosaic + Nielsen PRIZM + the operator’s internal segment definitions; cross-cohort overlap quantifies whether the candidate draws the same customer profile as the affected existing locations. Foot-traffic co-visitation pulls Placer.ai + SafeGraph + Foursquare + Veraset + Cuebiq co-visit data within configured time windows (per-day-part + per-day-of-week) to estimate the share of visitors at the candidate site who also visit existing locations. The three subsystems compose into a risk tier (no-risk + low-risk + medium-risk + high-risk + very-high-imminent) with affected existing locations enumerated, revenue-impact estimate, time-to-recovery estimate, per-candidate confidence interval (bootstrap + Bayesian + Monte Carlo + sensitivity), and explainability (SHAP + LIME + feature importance + counterfactual explanation + attribution graph + causal DAG + multi-LLM narrative explanation under per-vendor zero-retention).

What does Gate do — FDD Item 12 + 15-state franchise registration + ECOA / Fair Housing + FTC location-data precedent + EU AI Act Annex III?

Gate evaluates five operationally distinctive regulatory anchors before any cannibalization score is delivered to a franchise-development decision. Anchor 1 (the most operationally distinctive): FDD Item 12 territorial-protection per FTC Franchise Rule 16 CFR 436 + NASAA Commentary on Item 12 — when the franchise system grants protected territory, any new candidate site whose trade area overlaps the protected territory of an existing franchisee triggers an Item 12 attestation; the scoring output must explicitly enumerate every overlapping franchisee plus the protection type (exclusive / non-exclusive). Plus 15-state franchise registration (New York, California, Illinois, Maryland, Michigan, Minnesota, Nebraska, North Dakota, Rhode Island, South Dakota, Virginia, Washington, Wisconsin, Hawaii, Indiana) + 7-additional-state franchise disclosure + state franchise relationship laws (Arkansas + California + Connecticut + Delaware + Florida + Hawaii + Illinois + Indiana + Iowa + Michigan + Minnesota + Mississippi + Missouri + Nebraska + New Jersey + Rhode Island + South Dakota + Virginia + Washington + Wisconsin) + FDD Item 17 termination provisions + FDD Item 19 financial performance representations. Anchor 2 (anti-discrimination in site-selection): ECOA Regulation B disparate-impact + Fair Housing Act disparate-impact + FCRA when site-selection scoring affects credit decisioning — algorithmic site-selection cannot use or proxy for protected class via ZIP code + surname + demographic correlate. Anchor 3 (FTC location-data precedent — applies because foot-traffic co-visitation is a primary Score input): FTC v X-Mode Social / Outlogic consent order (January 2024) + FTC v Mobilewalla consent order (December 2024) + FTC v Kochava + Massachusetts Attorney General v Copley Advertising (April 2017) precluding precise-location targeting tied to sensitive categories; Illinois BIPA + Texas CUBI when biometric proxies sneak into the foot-traffic pipeline; Washington My Health My Data Act 2024 + Texas SCOPE Act 2024 when the trade area touches healthcare or under-18 demographics. Anchor 4 (EU AI Act high-risk): EU AI Act Annex III classification when AI-driven cannibalization scoring drives a major investment decision (capital deployment + lease commitment + multi-year franchise-grant); Article 9 risk-management system + Article 13 transparency to deployers + Article 14 human oversight + Article 15 accuracy + robustness + cybersecurity + Article 22 transparency of automated decisions + Article 26 deployer obligations + Article 50 transparency for AI-generated content. Anchor 5: FTC Section 5 substantiation (Pfizer 1972 reasonable-basis) when cannibalization scoring is referenced in franchisee-facing FPR / FDD Item 19; AICPA SSAE attestation + audit committee defensibility; CCPA + CPRA right to opt out of automated decision-making + 17-state comprehensive privacy + GDPR Article 22 automated decisions; NIST AI Risk Management Framework Govern + Map + Measure + Manage; ISO 42001 AI Management System; per-vendor LLM zero-retention verified per call. Policy-as-code expression via OPA Rego + AWS Cedar + Casbin + Cerbos + Oso + Styra DAS + Permit.io.

What does Audit do — per-candidate WORM record + FBC feedback loop + end-to-end replay?

Audit writes a per-candidate WORM record at every Score: per-candidate ID + per-banner pointer + per-franchise-system pointer + canonical trade-area model snapshot (radius + drive-time + walk-time + transit-time + Huff + MCI + Reilly-Converse + Voronoi + Thiessen + LBC + DMA / CBSA overlay) + per-existing-location overlap record (Jaccard + Sorensen-Dice + Tversky + Huff expected-share + MCI expected-share + Voronoi + Thiessen + confidence interval + multi-LLM explainability) + demographic cohort weighting record (Census ACS + Esri Tapestry + Claritas PRIZM + Experian Mosaic + Nielsen PRIZM + operator internal segment) + foot-traffic co-visitation record (Placer.ai + SafeGraph + Foursquare + Veraset + Cuebiq + time-window + confidence interval) + FDD Item 12 attestation (protected-territory check + affected franchisee list + protection type + encroachment-detection signal) + multi-LLM classification (per-model version + system prompt + temperature + zero-retention verification + LLM cannibalization-risk-tier extraction + LLM affected-existing-locations extraction + LLM revenue-impact estimation + LLM time-to-recovery estimation + LLM FDD Item 12 violation extraction + chain-of-thought trace) + per-candidate confidence interval (Bayesian posterior + frequentist + bootstrap + Monte Carlo + sensitivity analysis) + per-candidate explainability (SHAP + LIME + feature importance + counterfactual + attribution graph + causal DAG) + per-anchor Gate decision with evidence + sibling-handoff pointers + FBC feedback loop record (realized vs predicted feedback + trade-area model recalibration + overlap estimation recalibration + demographic weighting recalibration + foot-traffic co-visit recalibration + Item 12 attestation recalibration + confidence recalibration). Storage on AWS S3 Object Lock + Azure Blob immutable + Google Cloud Storage Bucket Lock + Wasabi compliance WORM. Retention stacks (longest applicable wins): 7-year FTC substantiation + 7-year IRS tax + 7-year FDD record + per-state franchise registration retention + 3-year Illinois BIPA biometric retention + GDPR Article 30 records of processing + EU AI Act Article 12 record-keeping + SOC 2 CC7 / CC8. End-to-end replay rewinds Model + Score + Gate + sibling handoffs + FBC feedback with confidence tier and explainability at every stage. Sibling handoffs flow into the cannibalization-analysis commercial pillar (parent), site-selection software, market-opportunity analysis, competitor mapping, franchise-territory mapping, franchise-registration states, per-prospect franchise-territory analysis pipeline, foot-traffic integration sibling build-pillar, per-location demographics, demographic data ingestion, commercial real estate API, and cross-location cannibalization detection sibling build-pillar.

Engage Completions on the territory-analysis bundle

The Model + Score + Gate + Audit four-skill bundle ships as the orchestration layer above your existing site-selection + geospatial + foot-traffic + demographic surface. FDD Item 12 territorial-protection + 15-state franchise registration + ECOA / Fair Housing disparate-impact + FTC location-data precedent + EU AI Act Annex III + Article 22 + 50 + NIST AI RMF anchors are preserved in every per-candidate audit record. Tier 1 AI Readiness Assessment scopes the bundle in two to three weeks; Tier 3 Fractional CMO with AI Swarm operates the bundle end-to-end.

Tier 1 AI Readiness Assessment Tier 3 Fractional CMO with AI Swarm Take the 3-question quiz