For the third year running, we fielded a customer feedback survey. A very sincere thanks to all who participated. The quality and quantity of openend responses were tremendous! We are reviewing the comments, and hope to be able to implement many new features and changes based on the suggestions.
“Longer support hours!” was one of the most common themes—especially from our east coast customers in the US. When we moved to Washington State 10 years ago, this put us at a time zone disadvantage with respect to the east coast. Based on a recent hire of another customer support consultant, we will open the office to business and technical support 1.5 hours earlier than before. Starting August 1 our new hours will be 7AM-5PM, Pacific.
Here are a few interesting findings from the 2005 Customer Feedback Survey.
The overall impression of our company is high, and improving. In 2003, 92% of respondents rated the overall quality of interactions with us as Excellent or Good. In 2004 and 2005, 93% and 94% said the same. We’ve plotted the top box (Excellent—FAR exceeded what I normally expect from a software company) percentage below:
Quality of Advice
A key indicator for success is the quality of the help provided. In 2005, 98% of respondents who had an interaction with us either Strongly or Somewhat Agreed that they were confident the support/advice they received was correct. We’ve plotted top box results below:
Conjoint Methods Used
For three years running, we’ve tracked the relative use of Sawtooth Software conjoint methods. In 2005, the momentum continues for increased use of CBC. CBC accounted for 54% of all conjoint/tradeoff projects conducted by customers responding to the survey over the last 12 months. At Sawtooth Software, we’ve been interested in the relative use of our three conjoint packages. The relative proportions are shown below, weighted by projects conducted.
Best/Worst (MaxDiff) Usage
For the first time, we included Best/Worst (MaxDiff) scaling as a response option in the conjoint/tradeoff method usage question. (Although MaxDiff scaling is not in the formal sense a conjoint methodology.) 12% of respondents that conducted preference modeling employed Best/Worst during the year, and it accounted for just under 4% of the total preference modeling projects conducted.
We currently offer a Best/Worst Experiment Designer, and CBC/HB or Latent Class software for parameter estimation for MaxDiff. However, we haven’t yet integrated a complete MaxDiff solution within SSI Web yet. We hope to offer that soon!