D2C FMCG — Ready-to-Eat
Reduction in weekly planning time
OTIF achieved, up from 83%
About
The brand, a fast-growing D2C ready-to-eat brand, wanted to reclaim Fridays by letting AI build the weekly production schedule. Promo spikes the spreadsheet couldn't see forced 8-hour planning marathons and kept OTIF stuck at 83%. We paired a reinforcement-learning agent with a constraint-programming core that proposes and explains multiple feasible plans, then pushes the winner into the TMS each night.
Industry
D2C FMCG — Ready-to-Eat
Company size
200 – 500 employees
Founded
2019
The Company
A D2C brand whose growth outran its spreadsheet
The brand produces and ships ready-to-eat meals direct to consumers, with a promotional calendar that moves fast and a production schedule that has to move with it. The planning team built the weekly schedule by hand in a spreadsheet that had no way of anticipating a promo spike before it hit.
Every promotional surge became an emergency replan, absorbing the better part of a working day and leaving little room to consider trade-offs properly before committing.
The challenge
Promo spikes turned planning into firefighting
The spreadsheet model could only work from historical averages, so a scheduled promotion routinely blindsided the plan the week it launched. Planners spent eight hours reworking the schedule by hand, under time pressure, with little ability to compare more than one or two alternative plans.
OTIF sat stuck at 83% as a direct consequence — not because the team lacked the skill to plan well, but because they never had enough time to properly evaluate trade-offs before a schedule had to be locked.
The Solution
A reinforcement-learning agent that proposes and explains trade-offs
We paired a reinforcement-learning agent with a constraint-programming core: the model proposes several feasible production plans, explains the trade-off behind each one — cost, changeover time, promo coverage — rather than returning a single black-box answer, and pushes the selected plan directly into the TMS overnight.
Planners retained the final call on which plan to run, but went from building a plan from scratch to reviewing and adjusting a fully-formed one, cutting the weekly cycle from a full day to under an hour.
The Results
Planning time down 90%, OTIF up to 96%
Weekly planning time fell roughly 90%, and OTIF rose from 83% to 96% as the schedule could finally account for promotional demand ahead of time rather than reacting to it after the fact. Freight and re-handling costs tied to last-minute replanning fell by an estimated Rs. 1.6 crore a year.
The planning team redirected the time they recovered toward supplier coordination and new-product ramp planning — work that had been permanently deferred while Fridays were consumed by the manual schedule build.
KEEP READING
Explore more studies on AI, analytics, and enterprise intelligence.
