This article is going to explain a basic concept – how demand and throughput live together. It is going to touch a little bit upon the concept of queuing and suggest ways to control both variables. As usual, we will be working with simple examples that make it easy to understand and apply the techniques that Kanbanize was built to enable.
This time, you will be an artist who owns an online business that doesn’t do so well, even though you send out tons of paintings. Go through the entire post to figure out how you got out of this situation.
Before we start with the interesting stuff, let’s explore the chart, just to get a grasp of what it shows and how it should be interpreted.
What you see is a representation of the total number of created (blue) and completed (brown) tasks per month. Nothing more nothing less. But what does this mean?
Suppose you are an artist and you have a fancy website where people can order portraits. They just upload a photograph and you paint it for them. You offer three different sizes, the bigger the size, the more expensive the painting.
Let’s assume that the chart above shows how your business developed over time.
May was somewhat okay, June was also fine, but you realized that you had loads of unfinished paintings, despite the fact that you spent all weekends working and you achieved your personal all-time best.
Realizing that this cannot go on forever, you invited your friend to work with you. July was a great month because you two were able to finish almost all the paintings and all the clients were happy with your work. August was great because you achieved the amazing 74 paintings, but you also got two angry emails from customers that have been waiting far too long.
You and your friend stepped it up a notch and in September you outperformed a small factory – the unbelievable 86 paintings count! Though, the angry emails from customers increased to a dozen.
Then came October. Almost no requests, poor productivity. What happened?
It turned out that one of your angry customers was a very famous blogger, who happened to be quite influential in the community. He delivered a massive anti-campaign for your service and almost got you out of business. Your friend was pissed off with the constant complaints and decided to quit.
While this story is strongly exaggerated in terms of how fast things happen, the actual events do occur every day in business. Companies go bankrupt just because they cannot deal with scale. Rising demand is good when you are able to keep up with it, but if you get tons of requests that you are not able to serve, then you might be frustrated and your company may be in trouble.
From queuing theory, we know that if we are constantly taking on more requests than we are able to process, then the wait times will also consistently increase until eventually some of them get discarded. If your customers have to wait too long, they will go away. If they have paid you up front, they will be a pain in the back (with full rights to be such) and they will be quite vocal. You don’t want that, do you?
Real vs. Fake Demand
There is no definitive answer to which demand would lead to good ROI. This is especially true when you do not know for sure how much you are going to sell. With the paintings, this is easy – each order means cash, but, in reality, customers will often ask for something that they are not willing to buy. When you estimate your demand, it is a good idea to categorize it as “revenue-generating”, “good to have”, “unknown”, etc. Certainly, the revenue-generating part should be monitored closely.
As simple as it may be, the throughput is represented by the brown bar. Our throughput is the number of paintings we produce per month (week, day, etc.). Increasing throughput almost always means good things for the undertaking. You have to be careful not to create a sub-local optimum, though.
If you are producing thousands of paintings, but nobody is buying them, then you are just wasting time and resources. You could be improving to a million of paintings per second, but then you would be closer to catastrophe than if you don’t do anything at all (the more you spend on a non-buyable product, the more you lose). Produce only what is right, with the right amounts at the right time (JIT).
As a rule of thumb, we would say that your total throughput should always be slightly smaller than your total demand. If your throughput is greater than the demand, you will soon be out of things to do. If the opposite, you will never be able to meet your customer’s needs. If equal, you will be probably doing things that are not so valuable. Only when you have to “descope” some of the demand, you will be spending enough time thinking about what to do and what not. A last tip regarding scope: if a request is very very old, you may want to drop it first.
So, you are in a situation where you have too many things to do, you can’t or don’t want to expand your capacity, but you need to do something. In this case, you should affect your demand negatively. One of the most obvious ways to do so is by making your service or product more expensive. This may sound counter-intuitive, but if you do so, you may be actually making more money than before.
Another approach would be to simply drop some of your demand. For example, you drop the smallest paintings from your offering. Not only that you will be getting more (the biggest ones are most expensive) but you will also have to care about less inventory and hassle. I think that dropping part of your demand is great and I call it “focusing”.
Believe it or not, there are situations where you would want to control your throughput not only on the positive but also on the negative side. As mentioned above, you may be producing a million paintings, but if nobody buys them, then you better shrink your throughput. Get your people working on the five S activities and don’t be afraid to let them “idle”. This is for the good of the company!
Maximizing the throughput is obviously harder to achieve. In the Lean world, this goes through visualization of the work, strict mapping of the workflow, limiting the work in progress and constant measurements. In other words, all that Kanban represents. A good way to proceed towards maximizing your throughput is by reading the cumulative flow and cycle time articles in our blog. They give a pretty neat explanation about what to do and what not to do.
No matter, if you run a small barbershop or a huge multi-billion corporation, you need to measure what your customers require from you and what you are able to deliver. Trying to balance your demand with your throughput is key to sustainable growth and smooth operations for any business and you should find a way to do it for yours. We hope that Kanbanize will help you along the way!