Kansei engineering

Description

Maps users' feelings about products, captured as image words on semantic differentials, onto coded design elements (shape, colour, material) so design choices can target an intended emotional image.

Kansei engineering translates feelings into product properties. Its formal version has five parts: collect representative product samples, decompose them into design elements, gather the semantic space of image words, have users rate samples on those words, and map ratings to elements, often with statistical models. Zhu et al. extend it with brand-culture elements and 'mainstream' elements counted across a competitor library, then verify the final design with the same semantic differentials. A lightweight qualitative variant (Pavliscak) harvests emotion words from transcripts, diaries, reviews and social posts, clusters them, and takes the clusters back to participants in follow-up interviews or participatory sessions, because words like 'love' or 'distracted' mean little without context. Rigour comes from agreeing target words with experts, managers and users, using a balanced rating sample, and verifying the outcome on the same scales.

Attributes

Loss type
Siloed
Data
Mixed
Effort
Weeks

Steps

  1. Collect domain texts and stakeholder input (brand materials, manager interviews, reviews, transcripts) and extract candidate image or emotion words.
  2. Screen to a short set of target image words agreed by experts, managers and users; for the lightweight form, cluster emotion words into themes instead.
  3. Gather competitor or candidate products (for example 87 reduced to 60 after removing near-duplicates), decompose into shape, colour and material codes and count element frequencies.
  4. Select about 10 highly varied representative samples and have a balanced user sample rate each on 7-point semantic differentials (-3 to +3) for every target word.
  5. Average scores; treat means above 0 as matching the image, and map samples' coded elements to the words to find elements that satisfy all targets.
  6. In the qualitative variant, take the word clusters into follow-up interviews or a participatory session to recover context and relate clusters to product properties.
  7. Design by combining brand, mainstream and perceptual elements, then verify the final design with the same semantic differential on a fresh sample.

Sources

  • DUXU 2022: UX Research, Design, and Assessment (Soares et al., 2022)
  • Emotionally Intelligent Design (Pavliscak, 2018)
  • Handbook of Human Factors and Ergonomics (Salvendy & Karwowski, 2021)

Also known as

  • Kansei engineering / perceptual image mapping with brand identity
  • Perceptual image design
  • Product perceptual image research
  • Affective engineering
  • Kansei clustering
  • Emotion-word clustering with follow-up

Categories

  • Analytics & Behaviour
  • Survey & Psychometrics