How Daraz Cut Reporting Time by 60% With a Decision Intelligence Platform

In 2024, Daraz — South Asia's largest e-commerce platform — had an analytics problem that every mid-to-large enterprise recognises. Three analysts, two days a week, building the same weekly operations report. Marketing had its revenue number. Operations had its revenue number. Finance had a third. Monday mornings started with a reconciliation meeting before anyone could actually review the business.

By early 2025, a single decision intelligence platform built on Microsoft Fabric had cut the weekly reporting cycle from two days to under four hours, retired three overlapping analytics tools, and ended the reconciliation meeting. This is how.

The starting point

Daraz's analytics stack in early 2024 was a typical enterprise patchwork:

• A custom-built dashboarding tool for merchandising.
• A separate BI platform used by finance.
• A marketing analytics tool for campaign attribution.
• Excel. A lot of Excel.

Each tool pulled from different source systems — the order management system (OMS), the warehouse platform, and Google Analytics — and each computed its own version of weekly revenue. The three-analyst team spent most of its week building the weekly exec review deck, which involved manually reconciling those numbers before anyone could ask questions about them.

The problem, stated precisely

The actual problem was not the tools. All three tools were fine at what they did. The problem was that no single metric definition was shared across them. Revenue in one tool excluded cancelled orders; revenue in another included them. Cart abandonment rate was measured from cart-created in one place and from checkout-started in another. Any of those choices is defensible — but they can't coexist.

The solution

The whole project ran on Daraz's existing Azure tenant. No data left the environment at any point. All three legacy tools were retired within six weeks of go-live.

The measurable outcome

Three specific changes compounded into the 60% time saving:

1. Single pipeline refresh. One pipeline pull, every four hours, into a shared lakehouse — versus three separate extractions, each with its own schedule.
2. One revenue definition. Encoded once in the semantic model. Every report, Q&A answer, and export used the same definition automatically.
3. Pre-built dashboards. Role-gated Power BI reports shipped with IntelliFabric for e-commerce — revenue intelligence, inventory velocity, cart abandonment funnel, customer lifetime value — were live in week two.

The unexpected benefit

The reporting time saving was the measurable ROI. The unmeasured benefit was cultural. Once everyone trusted the numbers, Monday meetings stopped being about reconciling data and started being about deciding what to do.

The category buyers — who had previously been the biggest consumers of custom-built reports — started building their own views against the governed semantic model. Within three months, 60% of analyst-originated report requests stopped happening, because the people asking for them now answered their own questions.

What we'd do differently

Two honest lessons from the engagement:

• We could have started the semantic model conversation earlier. We spent week one on pipeline work; in hindsight, running the revenue-definition workshop in parallel would have shaved two or three days off the overall timeline.
• Legacy tool sunset was smoother than expected. We'd planned a three-month parallel run; the team was comfortable retiring legacy tools after six weeks. Next time, we'll plan a shorter sunset by default.

Stack and architecture

• Cloud: Microsoft Azure (customer tenant)
• Data platform: Microsoft Fabric (OneLake + Fabric Data Factory + Power BI Premium)
• Accelerator: IntelliFabric for Retail (pre-built connectors, KPIs, semantic model)
• Source systems connected: Order management, warehouse management, Google Analytics
• Refresh cadence: Every 4 hours across all modules
• Users at go-live: ~120 consumers across exec, ops, category, finance

Is your stack a similar fit?

The pattern — multiple overlapping tools, analysts stuck on reconciliation, an executive team tired of arguing about which number is right — is one we see repeatedly. If your weekly reporting cycle still involves hand-reconciling numbers across tools before anyone can act on them, book a demoand we'll walk through a 4-week path to the same outcome.

Or read more about how decision intelligence platforms work — and the pre-built KPI library that made this go-live possible.

About Daraz: Daraz Group is the largest e-commerce marketplace in South Asia, operating across Pakistan, Bangladesh, Sri Lanka, Myanmar, and Nepal. Acquired by Alibaba Group in 2018.