Commerce analytics · Experiment design · 2026
Southeastics
A merchant decision lab for separating sales a campaign likely caused from sales it merely received credit for.
Stop rewarding
sales that
would happen
anyway
Southeastics turns merchant campaign data into traceable decisions.
Build your merchant baseline
These inputs set your economics, they do not yet tell us what is incremental
Methodology-led browser prototype. It does not connect to marketplace accounts or claim production causal inference.
The commercial question
Which sales would disappear if the campaign had never run?
Commerce dashboards assign orders and revenue to ads, affiliates, creators, and live commerce. But some of those customers would have bought anyway through repeat behavior, organic discovery, seasonality, or existing brand demand.
Southeastics reframes the workflow around the counterfactual. It is designed for regional growth and commercial managers deciding where the next growth dollar should go—and needing to defend that decision to marketing, finance, and leadership.
Evidence authority
The claim changes when the evidence changes.
Every output keeps its evidence label. The interface does not let an illustrative scenario masquerade as a measured commercial result.
Verified estimate
Campaign and credible comparison cohorts can estimate conversion lift, incremental orders, revenue, and contribution.
Can estimate—not claim certainty.Directional model
Real campaign performance without a counterfactual can diagnose efficiency using disclosed assumptions.
Cannot present modeled lift as measured causality.Illustrative example
Deterministic generated data demonstrates the workflow and calculation mechanics.
Export and real budget actions remain blocked.Product journey
Evidence enters once. Authority follows it through every decision.
- 01Merchant baseline
Market, category, spend, AOV, orders, and gross margin.
- 02Evidence standard
Choose what the available data can legitimately support.
- 03Evidence input
Load campaign and comparison cohorts or explore an example.
- 04Attribution gap
Compare credited revenue with demand above the baseline.
- 05Channel diagnosis
Inspect contribution, lift, repeat behavior, and evidence strength.
- 06Holdout design
Lock the outcome, threshold, guardrail, and confidence target.
- 07Decision brief
Translate the analysis into an explainable commercial artifact.
Analytical core
Make the commercial logic inspectable.
The product exposes the calculation record behind each finding instead of hiding the decision inside a composite score.
(exposed orders / exposed users) − (control orders / control users)conversion lift × exposed usersincremental orders × realized AOV(incremental revenue × gross margin) − campaign spend1 − (incremental revenue / attributed revenue)and positive contribution
without clearing Scale
investigate before spending
These thresholds are transparent prototype heuristics, not universally validated causal standards.
Pre-registered experiment
Decide what success means before seeing the answer.
A matched cohort without the selected channel intervention.
A comparable cohort receiving the campaign treatment.
Incremental contribution per user, with repeat rate as a guardrail.
Regional product design
One decision framework across eleven Southeast Asian markets.
Currency display localizes with the selected country. The prototype deliberately does not imply that category behavior or economic assumptions are identical across markets.
Brunei · Cambodia · Indonesia · Laos · Malaysia · Myanmar · Philippines · Singapore · Thailand · Timor-Leste · Vietnam
Technical design and safeguards
Browser-local by design. Honest about what is not production-ready.
Commercial logic
Browser-side TypeScript functions calculate lift, contribution, quality, and the attribution gap.
Local data handling
CSV files remain in the browser; workspace state is stored in localStorage with no application database.
Decision governance
Evidence labels persist, formulas remain visible, and illustrative mode blocks export and allocation actions.
Current boundary
No marketplace API, authentication, statistical power, confidence intervals, or merchant validation yet.
Never give a result more authority than the evidence used to produce it.