Start with a situation you will recognize if you have run a business for more than a few years. Demand is fine, the phone rings, the quotes go out, the calendar is full, and yet the business does not grow. This year's good month looks a great deal like last year's good month. When owners describe this to me, they nearly always propose the same remedy, which is "do more". More advertising, more leads, another person on the payroll. Today I want to show you why that remedy so often fails, and what to look for instead.
The idea we are going to work with is old, well tested, and simple enough to state in one sentence. Every system that does work has one step that limits all the others, and the output of the whole system is set by that step. In operations research this is called a constraint. In everyday language it is a bottleneck: the narrowest step in how work moves through your business, the one that sets your real capacity. In a small business, more often than not, it is a manual step, something only one person can do, done by hand, that everything else waits on.
Why does one bottleneck decide how much work you can take?
Because your capacity is not the sum of everyone's effort. It is set by the narrowest step, and improving any other step does nothing for output until the narrow one moves. That is the central claim of the Theory of Constraints, which Eliyahu Goldratt introduced in his 1984 book The Goal: "the rate of goal achievement by a goal-oriented system (i.e., the system's throughput) is limited by at least one constraint."
Imagine a four-lane highway feeding a one-lane bridge. You can add lanes to the highway all day. The bridge still decides how many cars get across, and every lane you add upstream only lengthens the line waiting to cross.
Now we can explain why "more" fails. A second result from queueing theory, proved by John Little in 1961 and known since as Little's Law, states that the average number of items waiting in a system equals the rate they arrive multiplied by the average time each one spends there. It is remarkably general: the relationship "is not influenced by the arrival process distribution, the service distribution, the service order, or practically anything else." The practical consequence follows once you apply it to a business whose bottleneck is the person who writes every quote by hand. If you double the leads arriving and the bottleneck's pace does not change, the excess has to go somewhere, and it goes into the queue. What grows is the number of jobs waiting and the time each one waits. You have paid to grow a line.
Consider what that looks like from the customer's side. I called on Tuesday and did not hear back until Friday, so I went with whoever answered first. The customer never sees a bottleneck. They experience it as slowness, and the business never sees that customer at all. Nothing in the monthly numbers explains where they went. This is also why a business can feel maxed out while the owner insists there is plenty of demand. Both statements are true. The demand is real, and the bridge is one lane wide.
How do you find your bottleneck?
You look for where work sits and waits, not for where people are busy, because everyone is busy. The bottleneck is the point where a job stops moving and sits until one specific person gets to it.
Goldratt's first instruction is simply to identify the constraint, and in a small business a handful of questions will usually do it. What do you do after hours because there is no room for it during the day? What has to route through one person, and what happens to everything else when that person takes a week off? Where does the same information get typed a second time, into a spreadsheet, into an invoice, into a system that does not talk to the first one? Pay particular attention to that last one... Retyping is the clearest signal there is, because it means two systems are being held together by a human being, and that human being is the bridge.
Then ask the question owners tend to answer quickly once they allow themselves to think about it. What do you turn down, delay, or avoid selling because of what it would take to fulfill? That answer usually points straight at the step.
Notice that the step is almost always boring. It is not the craft, and it is not the part of the business you are proud of. It is the paperwork in the middle. That is precisely why it survives for years without anyone naming it.
What if it is not a software problem?
Some bottlenecks are process. The step exists because of a rule nobody remembers making, and the fix is to stop doing it. Some are staffing, where the honest answer is that the work is real work and needs another set of hands. And a good many are already solved by software you are paying for and not using properly. If a tool already on your bill can do the job, use the tool already on your bill. That is a shorter conversation and it costs you nothing.
Custom software earns its place in a narrower case. The step is repetitive, it happens constantly, the rules behind it are specific to how you work, and no off-the-shelf product fits those rules without you bending your business to match. That is the moment building something makes sense, and it is a much smaller moment than the software industry would like you to believe.
What does removing a bottleneck actually look like?
Smaller than people expect. When we build at TKBS, it starts at the bottleneck and stops there, whether that turns out to be a plugin added to a store that already works, a widget built into an existing site, or a quiet internal tool nobody outside the building ever sees.
For a retail business running a Shopify store, that has meant plugins that handle the way that particular store sells: the special order that gets taken differently, the bundle the platform does not support, the connection to a tool the shop already uses. The store keeps working the way it does. The missing piece gets added.
For a business whose website should be doing more than sitting there, it has meant web apps and custom widgets built into the site, so the page does a job for the visitor instead of describing one. If your staff answers the same question over and over, that question is a tool waiting to be built.
How does TKBS remove a bottleneck?
We map the constraint, clear it with the smallest fix that works, and then go looking for the next one. Goldratt laid out five steps for working on a constraint: identify it, get everything you can out of it as it stands, subordinate the rest of the system to it, then elevate it, and finally go back and find the next one. The correspondence is deliberate.
We map before we build. The planning is its own piece of work: a written roadmap and proposal covering what should get built, in what order, and what it takes. That is the identification step, done properly, and the roadmap is yours to keep whether you build with us or hand it to another developer. You should be able to walk away from the first step with something useful in your hands.
Then we keep the build small on purpose. Ship the smallest thing that clears the constraint, watch it work, and grow it from there. If a bottleneck can be cleared with a focused tool in a few weeks, we are not going to propose a platform that takes a year.
The last of Goldratt's steps is the one people skip. Once a constraint is broken, another step becomes the constraint. You fix one, the business grows until it stalls somewhere new, and you go again.
Frequently asked questions
What is a business bottleneck?
It is the narrowest step in how work moves through your business, the one that sets your real capacity. Everything ahead of it can only move as fast as it does, so adding more work upstream creates a line instead of more revenue.
How do I tell a bottleneck apart from just being busy?
Being busy is not the same as being the constraint, because everyone is busy. Look for the point where a job actually stops moving and sits until one specific person gets to it. That waiting point, not the busiest desk, is the bottleneck.
How do I know if my bottleneck needs custom software?
Look at how often it happens and how specific the rules are. A step that runs many times a day, follows rules particular to your business, and cannot be handled by software you already own is the case where building pays. Anything else is usually cheaper to fix another way.
Will more marketing help if I have a bottleneck?
It will bring you more leads, and those leads will wait longer. Marketing is worth doing, but if the middle of your process is the constraint, fix the constraint first or you are paying to grow a line.
Should I just hire someone instead?
Sometimes yes. If the work needs judgment, hire. Software is the better answer when the work is repetitive and rule-based, the kind of thing a person does the same way every time and resents doing at all.
How long does it take to remove one?
It depends on the step, which is why we scope it before quoting it. A focused tool that clears one bottleneck is a matter of weeks, not the year-long project people picture when they hear the words custom software.
What if I have more than one bottleneck?
Almost everyone does. Only one of them is the current constraint, though, so you fix that one, see where the business stalls next, and go again. Trying to fix all of them at once is how these projects turn into the expensive story everyone has heard.
Start with the step, not the software
If there is a step in your business that everything waits on, that is where the next chunk of growth is sitting. You do not need a software project to find it. You need an honest hour with how the work actually moves.
Sources
Goldratt, E. M. (1984). The Goal. The Theory of Constraints and its five focusing steps are summarized at Wikipedia: Theory of constraints.
Little, J. D. C. (1961). "A Proof for the Queuing Formula: L = λW." Operations Research. Summarized at Wikipedia: Little's law.