// description
An iterative four-stage improvement cycle: Plan (identify the problem and design a solution), Do (implement on a small scale), Check (measure results against expectations), Act (standardise if successful, or restart the cycle with new information).
// history
Statistician Walter Shewhart introduced the concept in the 1930s as part of statistical process control. W. Edwards Deming popularised it in post-war Japan, where it became foundational to Japanese manufacturing quality. It was often called the "Deming Cycle" in Japan (though Deming himself credited Shewhart). PDCA underpins ISO quality standards, Lean, Kaizen, and Agile development cycles.
// example
Plan: you think a new cover design style will improve your KDP click-through rate. Do: update three books with the new design. Check: compare CTR and sales over 30 days versus unchanged books. Act: if CTR improved, roll out new design to the whole catalogue; if not, analyse what didn't work and try a different variable.
// katharyne's take
PDCA is basically the scientific method for your business. And honestly? Most creators skip the "Check" step entirely — they do stuff and then just do more stuff without measuring. The magic is in the measurement. I run PDCA mini-experiments on everything from email subject lines to Midjourney prompt structures. One change at a time, one cycle at a time. That's how you actually improve instead of just staying busy.
// creative uses
- Use PDCA to systematically test Etsy listing variables: run one PDCA cycle on cover photos (Plan: new mockup style, Do: update 5 listings, Check: click rate after 30 days, Act: roll out or revert). One variable per cycle. This is how you compound improvements rather than guessing.
- Apply PDCA to your Midjourney prompt development: Plan (hypothesis — a specific prompting technique will produce more consistent KDP interior quality), Do (generate 20 pages using the new technique), Check (compare against your previous 20 pages on defined quality criteria), Act (adopt or discard the technique). Build a personal prompt library from the Act step of each cycle.
- Run a PDCA cycle on your course completion rate: Plan (a welcome video with a "commitment ritual" will improve module 1 completion), Do (add it to one course), Check (compare 14-day completion rate before and after), Act (roll out to all courses or test a different intervention). Most course creators never measure completion rate at all — start there.
// quick actions
- Identify one specific variable in your KDP or Etsy business you've never properly tested (cover style, price point, listing description format, main photo background). Write a Plan for a 30-day test this week and set a calendar reminder to Check on day 30.
- Create a simple tracking spreadsheet with columns: Hypothesis, Change Made, Date, Metric Before, Metric After, Decision. Run every experiment through this sheet. The goal is a growing evidence base for your business, not one-off improvements.
- Look at the last change you made to your business that you "just did" without measuring. Go back and measure it now — find the before metric in your historical data and compare to current. You may have made an improvement you never credited, or a mistake you never caught.
// prompt ideas
Design a PDCA experiment for me to test [the variable you want to test — e.g. "a new cover design style", "a different pricing tier", "a revised email subject line formula"] in my [KDP / Etsy / course] business. Write out the full cycle: the specific Plan hypothesis, what I should Do and for how long, exactly what metric I should Check and where to find it, and the decision criteria for the Act step — including what "roll it out" and "try something else" each look like.
Help me build a simple PDCA experiment tracker I can maintain in Notion or a spreadsheet. My business involves [brief description — platforms, product types]. I want to run one experiment per month. Design the tracker columns, write the template row I'd fill in for each experiment, and give me a list of ten high-value variables I should test in the first year — ordered by likely impact on revenue or traffic.
I made a change to my [Etsy shop / KDP listing / email sequence / course] about [time period] ago: [describe the change]. I never measured the impact at the time. Help me run a retroactive Check step: tell me what data I should look at now, where to find the before-period baseline in my historical analytics, and how to interpret what I find. Then give me an Act decision based on what a likely result would look like.