Recently, I had the privilege of presenting a paper at the SKIM/Sawtooth Software European Conference and Training Event. This was my first speaking opportunity at a European conference, so I took careful notes of the event, and I’d like to share them with you.
What Is the SKIM/Sawtooth Conference?
This conference is known throughout the research industry as a world-class event, with opportunities to learn from experts and get practical instruction in many of today’s best conjoint research methodologies and applications. From the presentations to the breakout sessions and the conclusion, SKIM/Sawtooth Conference 2016 was very well-planned. It brought together stalwarts from the research industry, sharing their views on one platform. The thing that I loved about this conference was that, despite a very high attendance, they maintain an excellent level of content and don’t get overshadowed by vendor pitches. For me, such conferences are an inspiration to continue in cutting-edge and innovative research work.
All previous SKIM/Sawtooth conferences have mainly focused on advancing conjoint and its application. This conference was no different; it was divided into two halves, the first being devoted to impactful real-time examples, while the second focused on the best practices for operating and optimizing research techniques for:
- Conjoint on Mobile
- Conjoint in Loyalty Programs
- Guidelines for HB Modeling
- Segmentation Approaches
Debuting New MaxDiff Innovations at the Conference
As you probably know, Absolutdata has been developing an interesting variation of MaxDiff. I was scheduled to present our innovative usage of this technique in redefining a loyalty program, and my presentation session was scheduled for just before lunch. I was glad to see full attendance, even though everyone was super hungry! Our presentation covered the introduction of a new methodology, which adds multiple anchor points in a MaxDiff exercise. This technique is known as ‘Dual-Anchor MaxDiff’. The dual-anchor improvement solved two noteworthy challenges: it tackled a larger number of attributes than was previously possible in MaxDiff; and it provided multiple importance groups, which was also not possible in earlier MaxDiff exercises.
The case example which I presented was for a multinational hotel chain. This innovation helped them redesign their loyalty program, breaking the benefit statements into bundles of varying importance. The information was then used to create a three-tier loyalty program that featured customers’ preferred rewards.
A Day Full of Ideas on Conjoint
The conference included many sessions about Conjoint on Mobile. That certainly establishes how hot this topic is, and I was very keen to learn some different approaches. The presenters talked about their findings and recommendations for study specs (like the optimum number of screens, estimation guidelines, etc.) In my opinion, though, some time must pass before we can generalize these specs; they may depend on different geographies, different cellular technology stages and social strata.
Another interesting and unique presentation revolved around isolating preferred features using a single-card conjoint with two attributes. The preferences were utilized to develop a reward framework for customers. This distinctive methodology, though appreciated by one and all, raised a lot of questions on its broad applicability.
However, my biggest take-away was that HB can easily be used on limited data and can deliver amazing results.
A comparison of age-old segmentation techniques was intriguing. In this session, the latent class method emerged as the winner (due to low error and practical success rate) from among many methodologies. Besides its thought-provoking results, this particular presentation had an impeccable storyline with crisp examples. I’d say it was one of the best of the lot.
Overall, SKIM/Sawtooth was a very enriching experience. It filled the analytical part of my brain with ideas that I can’t wait to implement in my projects!