At InSource, we surprise you with how quickly we can build custom products from scratch. Life is too short for endless planning cycles and schedule adjustments. We wait until we know what you need and still deliver your product in time to keep your lines humming.
In 2014, we went looking for a method that would reduce lead times in a complex environment. Our product mix is too complex for Kanban. Finite Capacity Scheduling is too inflexible and requires almost perpetual updating. We discovered Quick Response Manufacturing (QRM) online, and it fits our needs perfectly. We unlock the power of time, empower teams, provide them with spare capacity, extend the concept to the whole company, and watch your jaws drop.
But, how can we afford the 20% spare labor capacity that QRM calls for? We serve your markets that have both seasonal and multi-year demand changes. We build products where labor capacity is the primary constraint. Unlike machines, people cost just as much idle as they do busy. People also become demotivated if not challenged. Spare capacity is essential to cutting response time, but does it come at an unsustainable cost?
We thought we could afford spare capacity with higher revenue and better efficiency. At the 2017 QRM world conference, we learned to calculate how much spare capacity we can actually afford. The results were disappointing: we could only afford 6 to 8%. This translates to 2 or 3 hours per week and is enough to reinvest in cross-training and process improvement. But we need a full 20% spare capacity to truly drive down lead times.
Frustrated with how to afford spare labor capacity, we kept ourselves busy on other QRM ideas. The 2017 conference introduced us to the power of hands-on simulations. We created a LEGO simulation building semi-trucks that focuses on lead time. Through this experience, we discovered how intensely competitive our cell members were. This led to weekly production targets and tracking of output and lead time by cell.
Early in 2016, we launched a website to manage jobs in Kanban-like stages. This dramatically dropped work-in-process and total manufacturing time. It also allowed us to coordinate documentation and job time tracking. The addition of on-demand web reporting further increased output and dropped manufacturing lead time. Yet every spike in demand left us late and frustrated. We tried having idle time for each employee but could not make it pay.
We knew that overtime could provide a burst of excess capacity. But we saw employees shifting into a slower gear when we used overtime for long periods of time. And we needed to eliminate the cost of overtime to afford the spare capacity in the first place. Two questions still troubled us: did other companies experience a large increase in labor efficiency by using QRM and how did they compensate employees for the loss of overtime income?
We were asking the wrong questions. We should have asked: how can we build spare capacity without idling people? It starts with 6 to 8% spare capacity that we usually invest in training, cross-training, and process improvement but that we can convert to production when demand spikes occur. Add a better use of overtime. Previously we used overtime for months at a time to prevent hiring. Now we plan to use overtime a week at a time when we need to catch up.
The breakthrough concept is using our labor capacity to get a few days ahead of schedule. Our goal is to be one week ahead of schedule, not working overtime, and spending two to three hours per employee on training and improvement. That adds up to more than 25% spare capacity relative to the 10 weeks it takes us to find and train new employees. We combine overstaffing, overtime, and overachieving to build spare capacity.
Once we had a vision, we created a workable system for creating spare capacity. A study of our job release system indicated that we could be one week ahead of schedule without increasing planned lead time. We routinely schedule components to arrive one week before a job must start to prevent run-outs. As we pull ahead of the schedule, the gap between component arrival and start of production shortens. We do not want to create a response-time spiral by inserting extra lead time or getting sloppy. A few rules are necessary to prevent this. We build actual orders you need: we don’t speculate. We keep WIP low and do not build up sub-assembly stocks. We just execute the normal schedule and finish custom products a few days early.
We borrowed a system that forecasts next week’s overtime requirements with a colored flag. The flag sets production targets and a broad range of acceptable work hours. The cells are responsible for hitting the targets regardless of the hours required, except that overtime caps prevent runaway costs. Often, the cells hit it hard and finish early on Fridays. We decide which color of flag to fly based on how many days we are ahead of schedule. We change the expected investment in training and process improvement by flag color as well.
The results in our wire harness shop have been dramatic: 46% improvement in jobs completed on-time. Manufacturing lead time dropping from 15.5 to 8.8 days. Dramatic short-term spikes in demand no longer make the shop late. All of this without a significant change in labor cost. Also, employees have more flexibility in setting their own work schedule. Some people want maximum overtime, some want minimum.
Overall, the system works very well. Our employees get notice of the upcoming workload, we throttle up or down based on our current situation, and you reap the benefit of products delivered quickly.
One big challenge is handling multi-month increases in labor demand. We experience this every few years. Our most recent surge in 2019/2020 went far better than one in 2017/2018, but we know we can do better. Stay tuned as we face this challenge head-on in the next few years.