Research
Tanqory Research
Design and position papers on how commerce software should work, from personalization and checkout to payments, fulfilment and selling across regions. Each states its method and its limits; the designs are argued from the literature, not yet tested in production.
Personalization and decision support
Personalization across regions and languages, kept fresh, safe and running in production, and agents that support decisions in small firms.
Event-driven decision making in commerce: a design for interpreting signals as they arrive
We describe an event-driven design that turns commerce signals into merchandising, pricing, payment and delivery decisions under explicit constraints.
Agent-based decision support in commerce platforms for small and medium-sized firms: a conceptual framework
We propose a conceptual framework for agent-based decision support built into commerce platforms used by small and medium-sized firms.
Detecting and responding to distribution shift in a deployed personalization system
We describe a design for detecting covariate, label and concept shift per cohort in a personalization system, and for containing it with fallbacks.
Checkout and conversion
How checkout speed, attribution and the system as a whole shape whether a shopper buys.
Conversion as a system outcome: a design for behavioural optimisation under constraints
We propose treating conversion as an outcome of the whole commerce system and describe a design that optimises it under fairness and latency limits.
Cross-channel attribution with learned credit models, calibrated by experiments
We describe an attribution design that combines Shapley, graph and sequence credit models and calibrates them against randomised experiments.
Latency, perceived performance and checkout: a design argued from response-time research
We describe a checkout latency design built on published response-time research, with a worked latency budget and a slowdown experiment to test it.
Measurement and forecasting
How to measure, compare and forecast commerce across regions, and turn the numbers into governed decisions.
An analytics-to-action pipeline: from event telemetry to governed decisions
We describe a design that turns commerce event telemetry into governed decisions, using causal estimation, constrained bandits and staged rollout.
Behavioural cohorts for customer lifetime value prediction: a design
We describe a design that predicts customer lifetime value from behavioural cohorts, probabilistic purchase models and survival models.
Real-time analytics for high-velocity commerce: an event-time streaming design
We describe a streaming analytics design for commerce surges: event-time windows, online change detection, latency budgets and overload handling.
Selling across regions
How shoppers pay, choose and receive orders in different markets, and how a merchant decides where to sell next.
Continuous market sensing from commerce signals: separating demand from noise
We describe a design for continuous market sensing that combines behavioural, operational and external signals and checks them causally before acting.
Delivery promises and carrier selection in cross-region commerce: a design for logistics intelligence
We describe a design that sets delivery promises from forecast delivery-time distributions and chooses carriers and service levels per lane.
Measuring consumer decisions mediated by recommender systems and AI assistants: a design
We propose a measurement design for how recommenders, AI summaries and shopping assistants change consumer search, reliance, trust and purchase.
Marketing automation
Deciding whether, when and where to send a message, within fatigue limits, time zones and consent.
Marketing automation as a constrained decision system: an event-driven design
We describe a design that treats each marketing send as a decision: event intake, causal effect estimation and policies bound by explicit constraints.
Consent-aware marketing automation: purpose and consent as inputs to every decision
We describe a design that makes consent and purpose inputs to data access, feature computation, model training and sending, not a flag checked at send time.
Send-time selection across time zones: a constrained, experimental design
We describe a design that picks a send window per customer and message from local routines, under quiet hours and contact limits, and learns by randomising.
Payment routing, settlement and risk
Routing, settling, securing and checking payments across acquirers, rails and jurisdictions.
A decision architecture for payment routing, retries and risk controls
We describe a design that treats payment routing, retries and step-up checks as constrained decisions, argued from work on bandits and causal inference.
Compliance by design in payment systems: a policy engine for onboarding, payments and payouts
We describe a design that expresses payment compliance rules as versioned, testable policy and calls it at onboarding, payment and payout.
Risk and trust decisions in payments: a design for calibrated, policy-constrained fraud controls
We describe a design for payment risk decisions that combines supervised fraud scores, anomaly detection, trust histories and a cost-based action policy.
Fulfilment and inventory
Where an order ships from, how stock is replenished, and how a fulfilment network holds up under disruption and law.
Legal constraints and risk penalties inside fulfilment planning: a design
We describe a design that treats trade and legal rules as hard constraints and supply risk as a priced penalty inside routing and network planning.
Event-driven visibility for orders, shipments and stock: a design
We describe a design that turns carrier, warehouse and order events into one state model for shipments and stock, with delay detection and ETA updates.
Periodic, multi-objective fulfilment network design for cross-border commerce: a design
We describe a design that re-evaluates fulfilment node locations and service areas periodically, with cost, delivery time and reliability as objectives.