Leveling to Build Capacity and Flexibility
Dear Gemba Coach,
We manufacture a seasonal product and are working on leveling, but are having a tough time. Our peak sales occur in a four-month time frame, but we need to use year-round production for capacity. Any advice on how to level what feels to us an impossible cycle?
Your challenge is difficult. I once visited a diary factory. We started at the beginning of the process as I always do, and we walked through a huge, half-empty warehouse. To be more specific, half of the warehouse was full, and half of the racking was empty. I asked them whether they should not have, at this stage, half-empty racks all over the place and they just looked at me blankly. They had nine months to prepare for the diary selling season, so they’d release one type from design, produce it, store it and then move on to the next type of diary – which is why the half of the warehouse was full, and half empty.
What were their main problems? I asked. Well, they told me, we are late on own release schedule, which was a real worry because they sold through retail chains, and so if they lost the time slot, they lost the shelf positioning. And of course, they had forecasting problems because, for inexplicable reasons, the same types of dairies did not sell every year, and they were always left with large quantities of diaries to pulp. Hmm.
This is indeed a difficult question and one where I’d dearly wish to have a magic wand – clearly the leveling concept doesn’t apply naturally to a seasonal business. Or does it? It’s hard to be specific without knowing more about what you manufacture or your industry but we can take a step back and try to reason out how leveling might work in case of a seasonal activity.
To start with, why level in the first place? You do so to build capacity and flexibility. Say you’re making 100 units a day. Demand increases 10% to 110 per day. You’ll struggle a bit, but you’ll cope without even considering changing the capacity of the process. If demand suddenly increases to 130 a day or more, though, you have a real problem. You’re going to have to work weekends, hire temps, train them, order more from suppliers, deal with parts shortages until the extra parts arrive, strain your equipment because there’s no more time to do routine maintenance, and so on. All of this will create exceptional costs that will eat significantly into your margins.
But how come production increased suddenly by 30%? Did customers decide from one day to the next to buy that much more than they did? Unlikely. Demand probably did increase but not in one day. Say we’re producing 100 units daily in June and decide that July’s plan should be 130. We’d know about it since the very beginning of June – sales must have picked up progressively. If we had been flexible enough to change the plan in, say, mid June, we could have moved to 115 parts a day on June 15 and 130 on June 30. More flexible still, we could have increased from 100 to 110 on the second week of June, to 120 on the third and to 130 on the forth, ready for July.
How would that make a significant difference? Ask yourself how many new temps a factory can hire and train to standard all at once. Or how many new parts you can get out of a supplier once they’ve shipped all that is in their stocks. Or what is the lead-time to get an additional machine on line. In production, lead-time is of the essence as we have a limited capability to do things all at once, but an infinite capability to do so one at a time.
Making Coats and Havoc
Most companies work to forecast, and the forecasting cycle itself adds to the delay in responding to real market changes. Forecasts tend to be reasonably accurate in terms of total demand – we know historically what the market is like. But forecasts are very poor at predicting what customer preferences are in terms of mix – the weather is sunnier, people feel more cheerful, and start preferring wearing bright upbeat colors. Snow comes a bit early this year, and instead of buying elegant Indian summer coats, the rage is for the heavy duffle coat. The total number of coats is likely to be the same, but not the actual product type. This has a huge impact for production because although the total volume forecasted might be practically equal, on one given line or equipment wanting more As than Bs this can create havoc.
How does this relate to seasonality? In my experience, seasonal businesses rely more than others on their forecasts. The forecast cycle is usually long enough to encompass the seasonality: six-monthly, or yearly. Once the yearly plan is set, it’s very tempting to adapt production accordingly and optimize to plan. If the plan is wrong, which it always is, you’re in deep trouble with no easy way to get out.
For argument’s sake, say that you produce twelve variants of a product (A, B, C… L) and that, as you say, you mostly sell them over four months – like the summer season. January, February, March, April, May are off-season, then June, July, August, September are the peak sales, and then October, November, December are low sales again. As you need to produce all year long for capacity reasons, the temptation is to ignore the sales pace, and produce to forecast to fill the warehouse, and hope for the best when sales orders come raining in.
Typically, optimizing to plan would lead you to produce As in January, Bs in February and so on until Ls are made in December. In doing so you’re keeping your changeovers to a minimum and can focus on one product at a time. You’re optimizing the use of your production equipment compared to forecast.
From a lean perspective, however, you’re producing each type once a year, which means a lead-time of a full year. If come June, you find out that customers are, for whatever reasons, buying a lot more As than Bs, you’re stuffed, because you’ve already produced As and Bs, and should you decide to start producing As again to fulfill the sudden demand, you’ll have to postpone producing Fs scheduled in June. But in June, people are asking for some Fs as well – you’ve got some in stock from last year, but will they be enough? and so on.
On the Level
What’s the alternative? What happens if we produce all twelve types every month? Well, by June, when the season is about to start, we’ll have produced 70% of the total volume. We can now focus more on the models that are selling best this season. By producing every product every month, we reduce lead-time to a month and are that much closer to real demand.
Clearly, this example is childishly oversimplified, but I hope it highlights the reasoning. By leveling the volume forecast and trying to produce all products all the time, we can use the off-season to increment inventory by small quantities regularly. When the season starts, we have a cushion of products in inventory – if we’re experienced, this cushion is the baseline that sells every year, the expected minimum for every product type, and then we have the flexibility to produce the models that actually sell when the peak season arrives.
Back to the first point, if one type starts selling like hotcakes, we can progressively increase the demand on the supply chain without changing the forecast at the last moment and going to the factory and suppliers saying, “you know what, fashion experts tell us that red is going to sell like crazy this year, so please double the volume of the red ones, for next month.” By (1) building all types stocks progressively through the year and (2) increasing our flexibility accordingly, we can (1) have a mattress of each type that should sell no matter what and (2) the flexibility of following real demand during the peak season.
Which brings us, not surprisingly, to the real bottleneck of such a strategy: our changeover capability. As long as changing batches in production is not studied, reduced and under control, building inventory by producing a little bit of each type every day is in jeopardy. One rule of thumb is that you invest 10% of your capacity in flexibility so batch sizes are calculated at ten times the changeover time. If you want to make every product every day, which is kind of the lean first goal, you need to reduce changeover time accordingly.
A further advantage of building a little bit of each type frequently rather than once a cycle is that you can work on standardizing operations for each type simultaneously. If you only build one model at a time, people will solve all the issues while they’re building As, but will forget about all they’ve learned about As when they start on Bs, and so on. By the time they’re doing As again, it’s like starting from a blank sheet. Producing A, B, … L frequently will keep them focused on keeping up standards for all products at all times, which will come in extremely handy when real demand kicks in, and we finally find out which model we need to focus on.
Human nature being what it is, your planning and production colleagues will probably mock you for suggesting that although your building up inventory for eight months in a row, you should produce as if sales were regular week by week. You’d be using your warehouses as a perfect customer. If you were in the toy business, for instance, your warehouse would “purchase” from the production plan as if parents averaged out their Christmas spree for their kids over the year on a monthly, even daily basis. It’s laughable.
Yet, what you’re doing in practice is build up the production sites and supply chain’s ability to change. By changing frequently and building a little of each one every week, the manufacturing system learns flexibility by practice. Consequently, when that flexibility is really needed because, in the sales season, customers never behave according to forecast but do their own thing – and surprise you – your processes are ready and able to change. This ability to change will then protect your profitability by keeping the exceptional costs of flexibility down to a minimum, and your production installations running close as can be to nominal capacity.
Lean Lessons from Cobra Kai(zen) and the Karate Kid
The unexpected wake-up call of the modest perfection of the original Karate Kid movie was that we need to move beyond defending this or that method of work and look to highlight opportunities of improving things beyond monetization, says Michael Balle in this reflection on the meaning of this classic movie.
How Using Kanban Builds Trust
Kanban functions as a trust machine because everyone using it must understand what they have to do and why, says Michael Balle: "Our purpose here is to share our ideas on what we believe is important in lean thinking."
The Sanity of Just-in-Time
Path dependence is the worst enemy of smart resolution, argue the authors, who suggest greater "frame control" with enabling tools such as just-in-time to respect people on the frontline and respect the facts they share about what is happening to them. "Mastering the path as opposed to being led by it, means looking up frequently to reevaluate both destination and way as new information comes to light."