Skip to main content

E‑commerce P&L Reporting Dashboard: four marketplaces on one screen

Four marketplaces, four report formats, and hours every week spent stitching data together by hand to answer one question that decides everything: which marketplace is actually making money? This is how I used AI to delete that manual step entirely.

Role Framing the problem
& designing the logic
Status In daily use
Tools Google Antigravity · Claude Code · Codex
Core result ~6 hrs/week saved
P&L managed in near real time
The marketplace reporting dashboard interface
01 · Problem

The data was all there, just fragmented and inconsistent

  • A matrix of reports: Shopee, TikTok Shop, Lazada and Tiki each have their own reporting system, with different structures, different names for the same metric and completely different fee schedules
  • Decisions arrived late: downloading, cleaning and joining files by hand in Excel took hours and invited mistakes at the reconciliation step. By the time the final number appeared, the window to act on the campaign had closed
  • Profit was a blur: costs could not be cleanly separated per marketplace — platform fees, ad spend, hidden charges — so ad budget was being allocated with no financial basis
02 · Solution

Automate with AI, break the P&L out accurately

My role was to set the brief and review the processing logic, using AI as the developer to generate the code that automates the whole data flow.

  • Extract: pull data from the marketplaces automatically via API, or read the raw report files
  • Normalise: bring the metric vocabulary of all four marketplaces onto one standard so they can be compared side by side
  • Calculate: deduct platform fees, shipping and ad spend automatically to reach net profit for each sales point
  • Visualise: push the numbers into dashboard charts so the financial picture reads in seconds
03 · How it runs

From raw data to chart

  1. Raw sources: pull three report types from each marketplace at once — ads, orders, and income/settlement
  2. Pre-processing and merge (ETL): map the three sources on order ID, remove duplicated costs on multi-SKU orders, then allocate ad spend down to the day and the individual order
  3. P&L engine: compute cost of goods sold (default 40% of list price), deduct platform fees, payment fees and both shop and platform vouchers to reach net profit, and flag any order whose costs exceed 60% of its selling price
  4. Output: write to a cleaned internal database, then publish in parallel to a visual profit-and-loss dashboard and a detailed Excel report
The four-layer data flow: raw sources, ETL pre-processing, the P&L engine and the dashboard output
The four layers of the data flow, from raw marketplace reports to the P&L dashboard and the Excel file.
04 · Results

Decide on profit, not on surface revenue

~6 hours
Saved every week
4 marketplaces
Merged onto one screen
~10 minutes
From raw data to a finished report
MeasureBefore the dashboardAfter automation
Time to a report~6 hours a week~10 minutes from raw data
Scope of oversightEach marketplace on its ownAll four marketplaces on a single screen
Basis for decisionsInstinct, reading gross revenuePrecise, based on net profit

The real value: the payoff here is not the six hours freed up each week. It is the change in decision quality. Moving budget between marketplaces is now directed by the profit actually received, which ends the habit of burning money chasing revenue that was never really there.

Next project
AI Content Production Pipeline
→