Queuing Theory: Go Slow to Go Fast

An interesting article in the June 11 Wall Street Journal talks about one company’s real-life realization that to maximize efficiency you have to not keep people working at full capacity - and this is especially true if you are looking for innovation. Furthermore, you need to carefully filter which innovations you invest in rather than trying to jam too much into the pipeline.

Avery Dennison Corp. loves to innovate. In recent years, the adhesive-label maker has expanded into areas such as stick-on automotive trim and heat-transfer inks to label clothing. But Avery executives grew vexed a few years ago at how long it took to turn ideas into products. Schedules were slipping. Customers were chafing.

Hoping to find the culprit, Avery hired George Group Consulting LP of Dallas to examine its practices. The surprising conclusion: Avery was jamming too many new ideas into its product pipeline, without enough slack time to ensure that critical tasks stayed on schedule. The remedy: Shrink the number of rollouts.

This fundamentally comes down to queuing theory, and how many things you can stack up without having dependencies which throw off schedules unpredictably. As the article notes, this type of thinking has been standard practice in manufacturing for decades, but “the same notion can seem like heresy when applied to scientists, designers or other creative types who launch new products.” In fact, this is exactly the approach that Donald Reinertsen advocated in his outstanding book Managing the Design Factory, which I have in my recommended books section.

Briefly his notion is this: innovative product ideas are like inventory sitting on a company’s books - until you get them out in the market they should show up as a debit in the accounting sheet (though company’s rarely track these things well, so the costs are often invisible). So you want to have as many inventory turns as possible to get ideas out quickly. But you don’t want to put out bad ideas, you need to balance time to market with “time to right”. This requires focusing your efforts and managing the innovation pipeline carefully. (I’ve written before about Google also dealing with this “innovation surplus” problem.)

The consultants mentioned in the WSJ article apply a mathematical theory on queuing, called the Pollaczek-Khintchine theory, to also show how  you need to have slack in the system to be most efficient overall. In other words, don’t keep people loaded up to 100% all the time if there are possible variances in how long tasks will take, as inevitably bottlenecks will occur during peaks. Since most real innovation efforts have multiple variables like this, to be most effective with a new innovation program you need to “under-utilize” people.