How Liv Data turned a multi-brand fashion group’s disconnected systems and a library of hand-refreshed spreadsheets into a single intelligence platform — covering sales, supply chain, marketing, and finance.
Client: North American fashion group operating multiple brands
Sector: Apparel Wholesale & D2C
31 Live dashboards across five departments, behind one sign-in
3 Source systems read live — no exports, no overnight copies
70+ Workbooks replaced, including two finance files of 74 sheets
19 Automatic checks run on every refresh, not once a quarter
250k+ Journal lines, one click from any figure on any statement
0 Systems replaced — no migration, no parallel run, nothing new to maintain
The Problem Was Not Missing Data. It Was Scattered Data.
The client sells a portfolio of brands through department stores, specialty retailers, off-price, export, and direct to consumers, across several operating entities and currencies.
Nothing was missing. Orders, shipments, and inventory sat in the system that runs the business day to day. The ledger, receivables, and payables sat in the accounting system. Campaign results sat in the marketing platform, in a separate account per brand.
Each system answered its own questions perfectly well. None of them could answer a question that crossed two — and almost every question worth asking crosses two. Which styles are aging, and what did we pay for them. Which customers are late, and what does that do to cash. Which brand is genuinely profitable once you count the cost of carrying its stock.
So people bridged the gap by hand, in spreadsheets. Dozens of them, rebuilt on a schedule, each one correct on the day it was saved and drifting from that moment on.
Data Still Lives in Spreadsheets
Before: More than seventy workbooks doing work the systems could not — including two finance files holding 74 sheets between them, rebuilt by hand every cycle. The tools had evolved. The workflow had not.
After: Replaced by a platform reading the source systems live. The logic inside the workbooks was kept, since it was the valuable part, and rebuilt as something the client owns and can correct.
Tools That Don’t Talk to Each Other
Before: The order and inventory system, the accounting system, and the marketing platform — three systems, each good at its own questions, none able to cross two.
After: All three read into one platform, live. Stock aging sits next to what it cost. Late customers sit next to what that does to cash.
Reports That Are Always Out of Date
Before: Operational reports ran off exports already stale by the time they were saved. Consolidation existed only in the moment someone was rebuilding the workbook.
After: Nothing is exported and nothing is copied. Multiple entities and currencies consolidate continuously.
No Shared Definition of KPIs
Before: The sharpest problem. Margin on the order book and recognized profit in the ledger came from different systems with different definitions, and had never been shown side by side — so leadership reconciled them in meetings, from memory.
After: Both sit on the platform next to each other, each labeled with what it measures and which system it came from. The meeting starts from the numbers instead of arguing about them.
What We Built, Department by Department
We did not replace a single system. That was the first decision and the most important one — the client had already invested in their platform, so we built something that reads it live rather than copying it. No migration, no parallel run, no new system for anyone to maintain.
Supply Chain & Logistics:
An exception list, not a dashboard. It does not show what is going well. It shows only what needs acting on today — stock that has aged, purchase orders that are late, customer orders that cannot ship in their window — each with the document to act on, who it sits with, and what to do next. Alongside it: stock aging and risk, availability, inbound orders and receipts, and a supplier scorecard measuring whether each vendor delivered when promised and in the quantity ordered, scored across tens of thousands of order lines going back years.
Sales & Commercial:
Order book, bookings against shipped, performance by account, territory, salesman, and market, returns, and a forecast grounded in committed orders rather than projections. Margin on the open book and margin on what has actually shipped, shown side by side — two numbers that had never been in the same place.
Finance:
Income statement, balance sheet, cash flow, receivables, payables, and a rolling forecast, built straight from the ledger. Multiple entities and currencies consolidated continuously. The part that matters most is not the statements — it is that every figure opens the entries behind it, and those entries add back up.
Marketing:
Channel and product performance, and campaign results read directly from the marketing platform — email, SMS, and automated flows, brand by brand, next to the sales they actually produced.
Executive:
A weekly view, brand performance, and an executive summary sitting on top of the departments beneath them — so leadership reads the same numbers the teams do, rather than a separate version assembled for the meeting.
Ask in Plain English:
An assistant that answers questions from the same governed data, deliberately scoped to commercial data, with the general ledger kept out of its reach by design.
How We Worked
Started with the people doing the manual work. Before writing code, we mapped the data landscape and sat with the teams, so the platform was designed around the decisions they actually make rather than the charts that are easiest to build.
Built in phases, each in use before the next began. Nothing was built speculatively, and nothing sat unused waiting for the rest.
We encoded their knowledge instead of replacing it. The logic that lived in workbooks and in people’s heads became a versioned, reviewable layer the client owns and can still correct.
Said no when the data would not support an answer. Several things we could have shipped quickly, we deliberately did not, because the underlying data could not carry them honestly. Where a figure could not be proven, the platform says so on the page rather than showing a confident-looking number that isn’t.
We handed it over. Documented, tested, access-controlled, running in the client’s own cloud, with the calculation logic written down in plain language.
What Changed
One set of numbers. Departments stopped arriving at meetings with different answers, because they stopped starting from different files.
Numbers became checkable. Not “trust the dashboard,” but “here are the records, add them up yourself.”
Operations got a to-do list. The supply chain view surfaces only exceptions, each with the document and the next action, so a morning is spent acting rather than hunting.
Reporting stopped depending on individuals. Work that lived with specific people now runs whether or not they are at their desk.
Teams moved from rebuilding to reviewing. The forecast reproduces the client’s own model, but assumptions are changed in the system and every save keeps a version, so the question moved from “is the model right?” to “is the assumption right?”
“Each system answered its own questions well. None of them could answer a question that crossed two — and every question worth asking crosses two.”
“We didn’t replace their systems. We made the ones they already had talk to each other.”
“Every number opens the records behind it — and those records add back up. That’s the difference between a report people look at and a report people rely on.”
“It doesn’t show you what’s going well. It shows you what needs doing today.”
“Where the data couldn’t support an answer, we said so on the page. A confident number that isn’t true is worse than no number.”
