
Consumer · Food & Beverage
Better Forecasting Took $6M Out of Delivery Costs
- About $234M revenue (2018)·
- 90 locations across U.S. & Canada·
- PE-backed
Ninety locations. Demand that shifts with every season. One way of running them.
A PE-backed packaged-ice producer had aggressive growth targets, but poor demand predictability and inconsistent operations left its delivery network underused. Between July 2019 and September 2020, Synergetics built an AI demand forecasting model and one standard operating model for a packaged-ice producer’s 90-location distribution network, cutting distribution cost per ton 10%, or more than $6M a year.
- 80% to 90% forecast accuracy from an AI demand model the client still runs
- 3 points reduction in no-sale stops across the client’s 90 locations
- 2 points reduction in stockouts nationwide, from better forecasting rather than more inventory
The challenge
The producer had grown into a network of 90 locations across the U.S. and Canada, and its owners had set aggressive top- and bottom-line growth targets. But demand for ice swings with the season and differs from region to region, and each location ran its operations its own way.
Leadership couldn’t predict demand closely enough to set how often each outlet should get a delivery, so trucks ran stops that ended in no sale while other customers ran out. Without a standard way of running a location, there was nothing to scale as the business grew.
Our approach
A pilot from July to November 2019, then a rollout to 65 locations that finished in September 2020, with on-site work paused from late March to May 2020 because of COVID-19.
What we actually did
- Built an AI demand forecasting model in Azure Machine Learning that tracks 44 variables by region and location, lifting forecast accuracy from 80% to 90%
- Used the forecast to set delivery frequency for each outlet, cutting no-sale stops 3 percentage points across all 90 locations
- Built a standard operating model in a pilot—performance metrics, staffing models, standard processes and flow control—as the playbook for every location
- Installed the management operating system in 65 locations between November 2019 and September 2020
- Designed a tablet-based ride-along tool and location audit checklist so the client’s own team could measure each driver against standard
What the plan had to account for
- Demand moved with the season and differed by region, so the model had to weigh 44 variables location by location.
- Each location ran its own way, so one playbook had to work across 65 different operations.
- COVID-19 stopped on-site work from late March to May 2020, in the middle of the rollout.
- The client had to run the system without us, so the work ended with its own continuous improvement team trained on our audit playbook.
The results
- The AI forecasting model, built in Azure Machine Learning and tracking 44 seasonal variables by region and location
- A four-person continuous improvement team the client hired to audit its regions with our playbook
- Driver ride-alongs at every location, captured on a tablet tool and measured against standard
- A full operations audit checklist for every location
- One standard operating model—metrics, staffing models, processes and flow control—installed in 65 locations, with the client’s continuous improvement team assigned to roll it out to the rest
The client hired its own four-person continuous improvement team to run the audits, using the ride-along tool and checklist built during the engagement, and kept running the forecasting model after Synergetics left.
Related work
Private Equity
127% of the committed EBITDA target
A PE-backed refrigeration and HVAC service business had 60+ branches and no common way of running them. Synergetics moved every branch onto one operating model its own leaders run, adding $7.6M in annualized EBITDA.
Read the case study
Private Equity
$14.7M in annualized labor savings
A PE-backed healthcare navigation platform needed a leaner cost structure without breaking a service 3M+ members rely on. A 12-week organizational design diagnostic turned into 19 months of execution and $14.7M in annualized labor savings.
Read the case study
Private Equity
$5.7M in EBITDA savings in year one
A PE-backed foodservice supplier had grown revenue 20-fold with flat margins and no common way of running its depots. Synergetics built one model depot, then rolled one management operating system across the network: $5.7M in EBITDA savings.
Read the case study
1/3Case study 1 of 3
Let us help you figure out where to start.
You know the number that has to move but not which solution moves it? Let’s start with that number.
Talk to an advisor