UCIP unifies real orders, market data and social signal on one platform. The system identifies which lever actually drives sales, proposes actions inside safe limits, then checks every result against real orders.
Observe reality and collect data — social market, e-commerce market, brand activity and first-party commerce — then detect and standardize the signals.
Interpret what the signals mean for the brand, category and commerce, and turn them into actionable suggestions. No decision is made here.
The operator decides which suggestion to run, then activates it — campaign, creator, commerce or promotion — to change market behavior.
Validate the result against real commerce with a two-tier check: market movement, then the first-party business outcome — sales, orders, payout.
Turn the result into knowledge: update signal confidence and feed the next iteration with what actually worked.
The system points the direction — it makes no promise of an exact forecast.
The social-driven uplift is separated from seasonality and the campaign calendar via a partial rank correlation.
Social is a 3rd-party signal; the truth is always the brand’s 1st-party orders.
Below the confidence threshold the system makes no proposal — instead of a number it cannot stand behind.
Decompose sales into three measurable parts — see what lifted and what dragged.
What people think, say & share
SocialHeat · 3rd
Conversation, sentiment and share of voice across social — a leading signal that runs about a week ahead of orders.
Conversations · Sentiment · Share of Voice · Creators
What people buy & how the market moves
EcomHeat · Creator Commerce
Share, competitors and category pricing from an ecom panel. Labelled "3rd · estimate", never mixed into first-party KPIs.
Market share · Category growth · Pricing · Product ranking
What the brand does
Activation · holdout · spend
Campaign calendars, budget, promotion mechanics and holdout groups. First-party context for deciding what to activate and how to measure it.
Media spend · Creator activation · Promotion · CRM trigger
What customers do with us
Shopee · TTS · Lazada
Real orders from Shopee, TikTok Shop, Lazada and POS. This is the truth every recommendation is checked against.
Revenue · Orders · GMV/NMV · Retention · CLV
Ingest 3 marketplaces + POS, normalize schema, materialize daily KPIs.
GMV, NMV, payout cross-checked XLSX · silver · ledger. Deltas surfaced.
Each brand is an isolated partition. RLS enforced, per-brand credentials.
Briefings and anomalies in Vietnamese. Provider chosen per organization.
Yesterday's GMV reached ₫4.28B, 12% above plan. Shopee led growth on the 9–11 flash sale. SKU SP-238 has 14 days of stock left — reorder suggested.
Every action passes a human approver, runs inside safe limits, with a stop button.
Every recommendation is re-measured on 1st-party orders — the system never grades itself.
RLS enforced on every org/brand table.
Per-brand encrypted creds, versioned keys.
Every change logged immutably per org.
Integer — no floats, no rounding drift.
