Making the Switch - Difference or Similarity?
A juice company was considering switching its apple supplier but needed to be confident that the majority of its customers would not notice the change. They had carried out many small scale triangle tests with small numbers of respondents but still lacked confidence to make the change.
We worked with the company to help them define the level of risk they were prepared to tolerate in making the change. They decided that they would not accept the new product if more than 20% of existing customers could tell the difference. In addition they were prepared for a 1 in 10 chance that any conclusion they made about similarity could be wrong (beta = 10%)
What We Did
We ran a triangle test for similarity between samples of the existing juice and samples made with apples from the potential new supplier.
66 respondents took part in the test all of whom were current buyers of the client’s apple juice brand.
What We Found
Of the 66 participants, 27 selected the odd sample correctly in the triangle test. Software calculated that for 27 correct responses from 66 at a beta risk of 10%, the proportion of true discriminators was higher than 20%. This exceeded the proportion of true discriminators that the client was prepared to allow and so the change was not made and the new supplier was not accepted.
This case study is adapted from Sensory Evaluation: A practical Handbook by Sarah E Kemp, Tracey Hollowood and Joanne Hort, Wiley Blackwell.