Why this matters
A website is never finished. The first version is a best guess, and the only thing that turns a guess into knowledge is real people using it. So the work is not "build it correctly", it is "set up a loop that keeps making it better".
The idea
The PDCA cycle repeats four steps:
| Step | What it means | On a website |
|---|---|---|
| Plan | Set a goal and a plan | "Cut the bounce rate by 10 percent", plan a homepage redesign |
| Do | Carry it out | Change the homepage |
| Check | Measure and evaluate against the goal | Measure the bounce rate; did it drop 10 percent? |
| Act | Decide the next move from that evaluation | Keep it if it worked; if not, try the headings or the images |
The point is not to get it right in one pass. It is to improve gradually by going round again.
Design thinking approaches the same job from the user's side, in five steps:
- Empathize: get into the user's position and understand their difficulties.
- Define: turn what you learned into a clear statement of the problem.
- Ideate: generate many possible solutions.
- Prototype: build something rough, quickly.
- Test: put it in front of users and collect feedback.
Test results often send you back to empathize or define. It moves back and forth, not straight through.
A/B testing decides between two specific options with real data: show version A and version B to actual users and see which performs better, say a red purchase button against a blue one. Two rules make it work. Change only one element at a time, or you cannot tell which change caused the difference. And wait for enough data before deciding.
Picture it
flowchart LR P[Plan] --> D[Do] D --> C[Check] C --> A[Act] A --> P
Worked example
A team wants more people to finish signing up. They change the button colour, rewrite the heading and shorten the form, all at once, and sign-ups rise 20 percent.
They have learned almost nothing useful. Something among the three worked; possibly one helped a lot and another hurt slightly. They cannot repeat the win elsewhere, because they do not know what it was. This is exactly what the one-element rule of A/B testing prevents.
Run properly, it is one change per test. Shorter form against current form, everything else identical. If sign-ups rise, that is knowledge you can reuse on every other form you own. Then test the heading, then the colour.
It also has to run long enough. Call it after an afternoon and you are reading noise: a quiet Tuesday against a busy Saturday will differ by more than most real effects.
Notice this is one turn of PDCA. Plan the hypothesis, do the test, check the result, act by keeping it and moving to the next question. The loop is the method, and a single perfect launch was never available.
Your turn
Try the practice questions for this lesson. Judge statements about PDCA and design thinking, put the five design-thinking steps with their descriptions, and work out why A/B testing changes one thing at a time.
Take it further
Your club runs a web page for its yearly event, but few students sign up through it. Should you redesign everything at once, or run one measured PDCA cycle?
- Collect data (Check first). Choose one number you can measure now, for example how many of ten classmates find the sign-up link within five seconds. Record it as your starting point.
- Plan one change. Name the single biggest obstacle, for a specific group such as a student on a small phone, and plan one change to fix it. State the goal as a number.
- Decide. Make the change, measure the same number again, and say what you would do next and why.
Measuring before you change anything is the step people skip, and without it Check has nothing to compare against.
Recap
- PDCA is Plan, Do, Check, Act, repeated; one pass was never the goal.
- Design thinking runs empathize, define, ideate, prototype, test, and loops back.
- A/B testing compares two options with real users.
- Change one element per test, and collect enough data before deciding.