Data story 01 · October 12, 2026
Design helps.It doesn’t decide.
KMSWERK gives every product an AI design score across nine criteria. We took the 560 products in our research library that have both a score and a documented market outcome, and asked a plain question: do better-designed products do better?
Pick a phenomenal success and a commercial failure at random. The success has the higher design score 71% of the time. A coin toss gets 50%.
of commercial failures scored at least as high as the median phenomenal success (71.2).
Spearman correlation between design score and outcome. 0 means no relationship; 1 would be a perfect ranking.
The scores lean the right way.
Median design scores rise step by step, from 66.9 for commercial failures to 71.2 for phenomenal successes. Design is not irrelevant.
Show as table
| Outcome | Products | Median | Middle half |
|---|---|---|---|
| Phenomenal success | 145 | 71.2 | 66.5–75.9 |
| Success | 219 | 69.4 | 64.0–74.1 |
| Moderate | 84 | 69.0 | 63.5–73.4 |
| Niche | 61 | 68.5 | 64.1–72.7 |
| Commercial failure | 51 | 66.9 | 61.2–70.6 |
But the groups overlap almost completely.
The middle half of the commercial failures runs from 61.2 to 70.6; the middle half of the phenomenal successes from 66.5 to 75.9. A score of 70 is ordinary in every group. A design score on its own can’t tell you which group a product will end up in.
8 products, fully sourced.
Outcome groups are judgments. For a stricter test we use market scores built only from figures we could open at the source and verify. So far 8 products in the library have one. Google Glass scores 68.8 on design and 10.0 on market; Rubik’s Cube 79.0 and 95.5.
Show as table
| Product | Design | Market |
|---|---|---|
| Google Glass | 68.8 | 10.0 |
| Nintendo Switch | 73.6 | 87.9 |
| Aeron Chair | 77.3 | 74.1 |
| Pebble | 67.5 | 47.0 |
| Rubik's Cube | 79.0 | 95.5 |
| Microsoft Zune | 61.2 | 51.0 |
| Juicero Press | 68.3 | 20.0 |
| Motorola Razr V3 | 76.5 | 86.7 |
Does design still matter?
Design is worth getting right. It just isn’t the whole answer: price, timing, distribution and the category itself sit outside a design score. That is why KMSWERK scores design and market separately, and shows its sources for the second.
Method
- The products
- 560 products from the KMSWERK research library: a curated set of notable designs, not a random sample of the market. Because it leans toward celebrated designs, 89% of them score 60 or higher.
- Design scores
- AI-evaluated on nine criteria, from photos where we have them and from product descriptions otherwise.
- Outcome groups
- Editorial labels (phenomenal success, success, moderate, niche, commercial failure) compiled from public records for research, each with a confidence rating. They are used only to check our scores, never as an input to the market score.
- A possible leak
- Some descriptions mention a product’s reputation. If the AI picked that up, design scores and outcomes would look more related than they are. The real link may be weaker.
- The numbers
- Spearman correlation on the five groups in order. The 71% figure compares every phenomenal success with every commercial failure, ties counting half. Medians and quartiles are per group.
- Sourced market scores
- Built from figures whose quoted sentence appears on the source page and whose number appears in the quote; each case page lists them. Computed on October 12, 2026.