SFOS + PRAT AI for FMCG · 2021, 2024

An Excel-to-AI digital transformation inside the world’s largest FMCG.

Outcome Enterprise B2B SaaS shipped end-to-end. SFOS (Sales Forecasting) + PRAT (Promotion Real-time Analytics), built on bank-grade enterprise UX. 4-year white-label partnership, live across multiple regions of the FMCG’s business.
SFOS + PRAT. Excel-to-AI platform for FMCG sales forecasting and promotion analytics

Enterprise B2B SaaS doesn’t win on the data. It wins on the UX that lets a Category Manager run their whole day on the new screen without slowing down.

A global AI & data-science specialist had won the contract to replace decades of Excel inside the world’s largest FMCG, an enterprise digital-transformation program covering sales forecasting, demand planning, and the entire promotion lifecycle. They owned the data engineering, the analysis, the ML, and the backend. They needed a white-label enterprise B2B SaaS design + frontend partner who could sit in their team and ship the customer-facing half. We embedded a pod, one product designer plus two React engineers, and shipped two flagship enterprise products together: SFOS and PRAT.

01

White-label, in the partner’s team. Our pod shipped inside the prime contractor’s repo, in their standups, with the FMCG’s end users in the loop, they stayed the named vendor, we stayed the design + frontend layer.

02

Excel-to-platform without breaking the workflow. Category Managers had spent decades inside spreadsheets, promotions, sales, demand planning. We had to ship interfaces fast enough not to slow them down on day one.

03

Enterprise UX on top of AI. Multi-stage approval chains, region rollouts, audit-grade reporting, five-plus reviewer workflows, the surface had to handle real enterprise complexity, not show off the model underneath.

★★★★★
We work best with Denovers when it comes to product designs and front-end development. Across multiple enterprise AI products under one ongoing partnership, their pod has consistently translated complex data workflows into interfaces that ship at bank-grade. Their passion, commitment, and collaborative teamwork focus on delivering outstanding results.
Ghena Ahsan Project Manager · AI & data partner · SFOS + PRAT
Impact
70%+ less Excel
Manual spreadsheet work reduced, per the partner’s public outcome metric. Category Managers stopped exporting, pivoting, and reconciling tabs.
Rollout
Multi-region
Live across multiple regions of the FMCG’s business. Hardened on one market, then rolled out to new categories and territories.
Investment
~$100K
Single engagement, per the verified Clutch review. 1 product designer + 2 React engineers from our side.
Class
Enterprise B2B
Bank-grade UX, multi-stage approval chains, audit reporting, role-based workflows, the enterprise stack, not a startup MVP.
The story

Enterprise B2B SaaS, white-label, in front of the world’s largest FMCG.

The partner is a global AI & data-science specialist. Their end customer is the world’s largest FMCG, Category Managers spending their days inside Excel for promotions, sales, and demand planning. The partner owned the data engineering, the ML, and the backend; they needed us as the white-label enterprise B2B SaaS design + frontend half of the product. We shipped under their name into the FMCG’s estate: SFOS (Sales Forecasting Operating System) for analytics + demand planning, PRAT (Promotion Real-time Analytics Tool) for the entire promotion lifecycle.

Embedded in the partner’s team, not over the wall. Same Slack, same Jira, same repo, same standups with the end users. Designs got reviewed in the same pull request as the React code that implemented them. Two products shipped under one shell (Dashboard / Demand Planner / Reports nav, scope filters at the top of every surface), one component vocabulary across SFOS and PRAT. Color carries semantic load (Amber / Green / Red OPSO bands, 5-status calendar tiles) so an operator scans the year ahead like a heatmap. Every Activity carries its own approval-history side panel: Acom → Cdf → Mgr Fi → regional → master, time-stamped, role-stamped, one click from the detail page.

ClientAI & data-science specialist · end customer: world’s largest FMCG
RegionGlobal · enterprise FMCG
EngagementWhite-label design + frontend partner · ~$100K · 2.5yr SFOS+PRAT · 4yr total
StatusManual Excel down 70%+ · live across multiple regions
01 · Design

One product designer, embedded in the partner’s repo.

CEO Muzammil ran requirements engineering and initial concept design with the partner’s PM in the early months; the embedded product designer carried both products through 2.5 years of delivery.

02 · Engineering

Two React engineers shipping in the same PR as design.

React.js frontend integrated with the partner’s AI / ML / data backend. Designs landed reviewed and built in the same pull request, no spec hand-off across the wall.

Chapter 01

SFOS: sales forecasting, finally legible.

SFOS. Sales Forecasting Operating System, replaces the analyst’s yearly-volumes Excel workbook with a Primary Dashboard. Seven scope filters at the top (Category / Brand / Sub Brand / SKU / Region / Vol vs Val / Demand Driver) carry through every chart. A YOY Volume matrix leads, color-banded so an analyst spots the worst row in two seconds. AWS table underneath, then Trend Line + Quarter Analysis + H1-vs-H2 charts pivot the same data three ways. AI-driven demand forecasting feeds the Demand Planner module on the same shell.

app.sfos.fmcg/primary-dashboard
SFOS Primary Dashboard, top nav (Dashboard · Demand Planner · Reports), seven scope filters (Category, Brand, Sub Brand, SKU, Region, Vol/Val, Demand Driver), a YOY Volume year-by-year matrix from 2015–2020 with monthly + quarterly + half-yearly + total columns, a YOY AWS table beneath, plus three small charts (Trend Line, Quarter Analysis, H1 vs H2) and KPI cards (YOY, CotC vs ROB, H1 vs H2 Growth, VG for Year 2020)
SFOS · Primary YOY matrix · AWS table · trend / quarter / H1-H2 charts Excel, replaced
Chapter 02

PRAT: a 12-month plan you can read at a glance.

PRAT. Promotion Real-time Analytics Tool, plans, runs, and audits every promotion across categories, brands, regions, and channels. The home is a Dashboard with the year’s Top 10 Activities, TTS / DTTS charts, and an Activation Metrics panel. The Activity Plan moves the work into a 12-month calendar. SKU rows down the side, months across, week-coded tiles plotted in. Tiles carry their type (TPR, EDLP, SBB, A&P, MT, Coupon) and a status pill so the calendar reads like a heatmap of the year. Hover surfaces Activity Code, Project Code, ROI, and categories without leaving the calendar.

app.sfos.fmcg/prat/dashboard
PRAT Dashboard. PRAT wordmark and top nav (Dashboard · Demand Planner · Reports), four scope filters (Division / Category / Week From / Week To), Top 10 Activities table (Brand & SKU, Activity Description, Activity Cost in MLN, Pre OPSO percentages) with multi-select Type of Promo / Brand filter dropdowns; below: TTS 2015–2020 line chart by category and DTTS 2016–2020 bar chart with Type of Promo / Month / Year filters, plus an Activation Metrics panel beneath
PRAT · Dashboard Top 10 Activities · TTS / DTTS charts · activation metrics Home
app.sfos.fmcg/prat/activity-plan
PRAT Activity Plan, eight scope filters (Promo Level, Division, Category, Brand, SKU, Master SKU, Year, Status), 12-month calendar grid with SKU rows and Jan–Dec columns; week-coded activity tiles in TPR / EDLP / SBB / A&P Gifts / Mtfts / Cp colors, status legend (Draft · Submit · Send Back · Approved · Reject); an open tooltip shows Activity type EDLP, Activity Code AC000001-TPALKNR20B, Project Code T00230-15168536…, Activity ROI 5%, and Category tags (Fabric Sensations · Oral Care · Deodorants & Fragrances · Skin Cleansing · Non Corporate Personal Care)
PRAT · Activity Plan 12-month calendar · status pills · activity tooltip The year, on one page
Chapter 03

Activity detail: every promotion has a paper trail.

A promotion crosses Account Management → Customer Development Finance → Manager Finance → regional → master before any budget moves. We made that chain visible: every Activity has its own page (Promotion / Geographic / Channel detail cards + Timeline + Project Code & Total Budget). The headline move is the Approval History side panel, opened on demand, time-stamped, role-stamped, with status pills on every entry. The Promotion List view adds a Human Intervened vs Machine Generated Values band so the operator always sees where AI ended and human judgment took over.

app.sfos.fmcg/prat/activities/80482-0q820934
PRAT Activity detail · Activity Code 80482-0q820934-8029348, three detail cards: Promotion Detail (Activity Code 2337, Category Oral Care · Spreads, Brand Knorr · Rafhan, SKU Rin 70 · Rin 1K · Rin 2K, Promotion Type TPR, Promotion Level Brand), Geographic Detail (Region Karachi · Lahore + 3 More, Area Nawabshah · Sukkur · Bahawalpur, Town Raiwind · Karachi + 3 More, Distributor Habib Traders · Pet Pump/Minimart), Channel Detail (cropped); Timeline (April 2019 · Start Week 1918 · End Week 1919); Project Code TPALKNR20BFJ · Detail of Offer · Total Budget 2520000 M. Right-side Approval History panel with timestamped entries (25th April 2020) tagged Submit By Acom, Rejected By Cdf, Send Back By Cdf, Rejected By Mgr Fi
Activity detail Promotion + geo + channel cards · approval history side panel Audit, on demand
app.sfos.fmcg/prat/promotion-list
PRAT Promotion List, three detail cards (Promotion / Geographic / Channel) for Activity Code 2337 with brand, SKU, region, area, town, master channel, channel, element, sub-element, plus a Timeline row (Year & Month, Start Week, End Week, Total Activity Weeks, Weeks in Current Month). Below: a Total Activity · Pre OPSO band tracking GP/Ton, AW Baseline, Total Cost, Inc Vol, Activity Duration, ROI %, Activated AWS, Tot Activated Vol, Post AWS, DNDS Vol, and Total Vol, with a legend at top distinguishing Human Intervened Values from Machine Generated Values
Promotion List Human Intervened vs. Machine Generated values Where AI hands off
Chapter 04

Reports: the auditor’s view, in one screen.

Two power-user surfaces sit on top of everything PRAT + SFOS know: a Pre vs Post OPSO Report (Amber / Green / Red bands for one-glance performance) and an Activity Network cross-tabbing SKU × promotion type. Both share an eleven-filter scope, XLS export, and a Lock SOP Version control so finance can freeze a snapshot without losing the live view. The conversation moves from "send me the workbook" to "send me the link."

app.sfos.fmcg/prat/reports/pre-vs-post-opso
PRAT Pre VS Post OPSO Report, eleven scope filters (Division, Category, Brand, SKU, Format, Year, Month, Promotion Type, Region, Channel, Customer, Week From / Week To), Absolute Spends matrix with color-coded column groups (Amber · Green · Red · Grand Total) split into Pre / Post columns, with Volume figures in millions and percentage shares; row stack covers Fabric Sensations Category, Lipton, Supreme, Pearl Dust, Foods, F&D, HHC, Laundry, HC; XLS export at the top-right
Reports · Pre vs Post OPSO Amber · Green · Red column groups · XLS export One-glance performance
app.sfos.fmcg/prat/activity-network
PRAT Activity Network. Lock SOP Version control top-right, eight filters (Category, Brand, SKU, Promotion Type, Year, SOP Version, Sum, Month), and a SKU × Promotion Type cross-tab matrix split into Volume and Value column groups (EDLP · SBB · TPR · A&P Gifts · MT · Total) with rows for Laundry, Home Care, Skin Cleansing, Skin Care, Hair Care, HPC, Savoury, Rafhan, Drinks, Foods, Beverages, every promotion in the entire estate on a single screen
Reports · Activity Network SKU × promotion-type matrix · volume + value · SOP version lock The estate, in one matrix
Outcome · 70%+ Excel reduction · multi-region rollout

Excel chaos in. Intelligent FMCG platform out.

One designer, two React engineers, embedded in the partner’s repo for 2.5 years. Manual Excel work down 70%+. Live across multiple regions of the FMCG’s business.

Book a free sprint
White-label sprint · on us

AI / data-science shop with an enterprise customer? Start with a white-label sprint.

One scoped surface, embedded in your team, in your repo. The same way we started with the SFOS + PRAT partner. If it doesn’t click, keep every file we ship.

  • Capped at 4 sprints / month
  • No card · NDA on request
Next case study → Case 01 · HoneyBricks · Multifamily real estate

How design + engineering helped a real-estate SaaS get acquired.

HoneyBricks came to us looking for an in-house Product Designer. Two years later they were acquired by EquityMultiple. The embedded-pod model the studio runs on.

Acq. EquityMultiple · Apr 24
24 mo Embedded pod
5.0 Andrew Crebar, CEO
Real-estate · SaaS · AI Read →