Principles of Productive Design Work

Key ideas#

Getting better, not perfect#

1. Improve rather than perfect. Nothing is ever finished; things are declared done because they have to be. Don’t polish what is too early to polish, and don’t chase new features while known problems remain. Golovach treats a day spent only adding good things, without fixing bad ones, as a day lost.

2. Be pessimistic about your own work. If everything can be improved, then everything you have done is still bad. Admitting that is what drives improvement.

3. Set up the work so you never wonder what you forgot. Write lists before you start (interface objects, checklists, target qualities) instead of relying on memory when you check.

4. Rework rather than overthink. A quick bad version, quickly reworked, gets the same result as long planning. But the work can be stopped at any point, and colleagues and clients can review it from the first iteration. This needs a fast prototyping method.

5. Start with the most painful part. Users judge a product by its worst part, not by its average, so fix the worst first. Find out what hurts most, by asking the client or by testing, before you start.

Knowledge and its limits#

6. Don’t act like an expert. Knowledge is useless until it becomes a design action. Assuming you know nothing makes you check the interface against a list rather than trust that it “will be fine”.

7. Don’t philosophize. There are very few universal answers; Fitts’s and Hick’s laws are among them. Only testing proves that a solution works. Don’t look for a general answer until you have solved the specific problem at least three times.

Optimizing what matters#

8. Do only the optimization that improves the product. Golovach’s formula:

improvement = share of users who use the fragment × importance weight of that user group × noticeability of the improvement

A full redesign of a rarely used dialog can score far lower than a small change to a screen everyone uses. He also explains why small improvements get blocked: deadlines, managers who can’t see them, and marketing that needs improvements explainable in a few words. Start with “will this help the product sell?”, not “do I like it?”

Professional conduct#

9. Act professionally. Use correct terminology. Don’t promise the client large improvements; Golovach suggests no more than 10–15%. If a client insists on their own solution, state the consequences clearly and make it explicit that you are not responsible for advice that was not followed.

10. Don’t do what you can’t do. Golovach lists five prerequisites for the job: listening and empathy, taste, writing, logic, and curiosity. All of them can be developed, but through years of deliberate work.

11. Accept that you will be wrong. He estimates that at least a third of interface decisions work worse than expected or not at all, because the user’s half of the interaction can’t be fully predicted. The answer is testing, not a grand theory.