Importance-Performance Analysis

Both are used and they answer slightly different questions. Stated importance, as in the original Martilla and James design, asks respondents directly and is simple but can be flat, redundant across correlated attributes, or inflated by social desirability. Derived importance infers each attribute's weight from how strongly it relates to overall satisfaction in a regression or similar model, which often discriminates better but assumes a correct model and adequate variation. Many practitioners plot derived importance against performance, or compare stated and derived grids, because attributes that look important when asked but have little impact on satisfaction reveal a meaningful gap worth investigating.

There is no single correct rule, and the choice matters because it determines quadrant membership. Using the grand means of importance and performance places each attribute relative to the average attribute, which tends to spread points across all four quadrants and is the most common convention. Using scale midpoints judges attributes against an absolute standard, which can pile most attributes into one quadrant if overall ratings are high or low. The key is to choose a rule deliberately, justify it, and apply it consistently across waves so results remain comparable; reporting both placements is a useful robustness check.

Voice of the Customer, in Griffin and Hauser's framework, supplies the validated, structured list of customer needs that should populate the IPA grid's attributes, so the two are complementary inputs. The Kano model adds something IPA lacks: it distinguishes must-haves, performance needs, and delighters, recognizing that satisfaction responds nonlinearly and asymmetrically to different attribute types. IPA assumes roughly linear, symmetric importance, so for categories with strong basic-versus-delighter dynamics, combining IPA's prioritization grid with a Kano classification gives a richer picture than IPA alone, ensuring you neither neglect a must-have nor over-invest in a feature whose absence is merely tolerated.