06 / 08Systems

Find the bottleneck

A weekday morning has a fixed shape. Wake up, shower, coffee, lunches, out the door. When it runs late, the lateness usually traces back to one step: one bathroom for three people, one kettle, one signature on a permission slip. Speeding up any other step changes nothing. A weekly report at work runs the same way. Five people send numbers, one analyst builds the charts, one manager signs off. If the analyst can build two sections a day and eight arrive on Thursday, the report is late on Friday however fast everyone else works. The bottleneck is the step that sets the pace for the whole chain. Finding it shows where an hour of effort will show up in the result, and where it will vanish.

In five minutesWrite the steps of your weekday morning in order. Beside each, mark where you usually end up waiting. That step is the bottleneck. Tomorrow, change one thing about that step and nothing else.

Method · Theory of constraints  /  Business · Engineering

↺From last time · Then what?

Answer these from memory before you start. Tap to check.

Why ask "then what?" more than once?

The first effect is the one everyone already sees. The effects that cost or help the most often come second or third, later and somewhere else.

What is the unseen cost of a choice?

What stops happening because of it. Bastiat's shopkeeper pays the glazier six francs, and the shoes he would have bought never appear.

When should the chain stop?

At about four or five links, or wherever you can no longer say why one effect follows from the last. Past that point, guessing outruns knowing.

01Where it comes from

Eliyahu M. Goldratt, an Israeli physicist, developed the theory of constraints in the late 1970s [2]. He spread it through a novel, The Goal, written with Jeff Cox in 1984, about a plant manager given three months to turn his factory around [1]. The method is usually taught as five focusing steps: identify the constraint, get the most from it, pace everything else to it, add capacity there if needed, and look again once it moves. A second tool comes from queueing theory. In 1961 John Little proved that the average number of items in a system equals the rate they arrive times the average time each spends inside [3]. Little's law now runs through operations and computer design [4]. When a reading pile holds steady at 30 articles and 10 get read a week, each waits about three weeks. The evidence for the theory of constraints comes mostly from case reports [5], and a 2008 review found it largely untested by controlled study [6].

Also called Bottleneck, constraint, five focusing steps, Little's law, The Goal.

02The method

Theory of constraints

  1. Write the steps of the process in order, from the start to the finished result.
  2. Beside each step, write how much it can handle in a day, and how much is waiting in front of it.
  3. Find the bottleneck: the step with the least capacity, usually the one with the biggest pile waiting.
  4. Get the most from it. Keep it from sitting idle, and move off it any work another step could do.
  5. Pace every other step to it. Producing more upstream only grows the pile.
  6. If it is still too slow, add capacity there. Then look again, because the bottleneck may have moved.

03An example

What it looks like done

Worked on A weekly sales report that goes out late most Fridays

The process, start to finish

  • Five team members send their numbers, all 8 sections in by Wednesday noon.
  • The analyst cleans the numbers and builds the charts: 2 sections a day.
  • The manager reviews all 8 sections in about an hour.
  • Due Friday at 3 p.m.

Waiting before it

  • Sections pile up in front of the analyst. All 8 arrive at once, and 2 go out a day.
  • 8 waiting at 2 a day is about 4 days. Wednesday noon to Friday 3 p.m. is about 2.5 working days. Late.
  • The manager's hour never has a queue. Speeding up the review would change nothing.

The bottleneck, and one change to try

  • The bottleneck is the analyst's cleanup.
  • Each sender cleans their own numbers before sending. The analyst now builds 3 sections a day, so 8 take under 3 days.
  • Senders send by Tuesday noon, so the queue starts a day earlier.
  • A late report meant a Friday rush, errors, and rework next week. Moving the cleanup off the analyst's desk breaks that loop.

04Try it · 20 minutes

For a class

Write out how one problem set or lab report goes from assigned to submitted: reading, notes, problems, checking, writing up, submitting. Beside each step write how long it takes and what tends to wait in front of it. Circle the slowest step. Write one change that protects its time, such as doing it first while fresh. Time the next assignment and compare. About 20 minutes.

For work and life

Map a weekly process with handoffs: a report, a hiring loop, the weekday morning at home. List each step, who does it, and how much it can handle per day. Count what is waiting before each step. Divide what is waiting by what gets finished per day to get days of waiting. Pick one change at the bottleneck only, and measure the whole process again a week later.

The blank page for this is in the Systems workbook.

05Where it goes wrong

  • Speeding up the wrong step. An hour saved anywhere but the bottleneck leaves the finish time where it was. Find the pile first.
  • Keeping every step busy. Steps running flat out ahead of the bottleneck only build a bigger queue in front of it.
  • The bottleneck doing extra work. When the slowest step also handles tasks others could do, its time leaks away. Move those tasks elsewhere.
  • Never looking again. Fix one bottleneck and another step becomes the slowest. Repeat the check after each change.

06Research

What research has measured

What the case reports show

One review of more than 80 published case reports found large gains after applying the theory of constraints, and no reported failures, which points to selective reporting.

The numbers and sources
Effect
A meta-analysis of over 80 successful applications reported in the literature found significant improvements in both operating and financial performance. Despite extensive searches, it found no reports of failure [5]. A later review judged that the theory meets the criteria of a good theory, but for the most part has not been tested empirically [6].
Population
Organizations, mainly manufacturers, whose theory of constraints projects were written up and published [5].
Caveats
Case reports, mostly by people who chose to publish a success. Finding no failures suggests what got published, not a perfect record. No controlled trials. The abstract gives no pooled figures [5].

07Check yourself

Answer from memory first. The next lesson asks these again.

Why does speeding up a step that is not the bottleneck change nothing?

The chain finishes only as fast as its slowest step. Faster work before the bottleneck waits in its queue, and steps after it wait for its output.

How can you spot the bottleneck without timing every step?

Look for the biggest pile of work waiting. Work stacks up in front of the slowest step.

Why look again after fixing it?

Once the slowest step speeds up, another step becomes the slowest. The bottleneck moves, so the check repeats.

Sources
  1. Goldratt EM, Cox J. The Goal: A Process of Ongoing Improvement. North River Press; 1984. first edition
  2. Rahman S. Theory of constraints: A review of the philosophy and its applications. International Journal of Operations & Production Management. 1998;18(4):336-355. https://doi.org/10.1108/01443579810199720
  3. Little JDC. A proof for the queuing formula: L = λW. Operations Research. 1961;9(3):383-387. https://doi.org/10.1287/opre.9.3.383
  4. Little JDC. Little's law as viewed on its 50th anniversary. Operations Research. 2011;59(3):536-549. https://doi.org/10.1287/opre.1110.0940
  5. Mabin VJ, Balderstone SJ. The performance of the theory of constraints methodology: Analysis and discussion of successful TOC applications. International Journal of Operations & Production Management. 2003;23(6):568-595. https://doi.org/10.1108/01443570310476636
  6. Gupta MC, Boyd LH. Theory of constraints: A theory for operations management. International Journal of Operations & Production Management. 2008;28(10):991-1012. https://doi.org/10.1108/01443570810903122

MVIII puts the day in front of you each morning. Get MVIII