This question has cropped up a few times. The simple answer is: MindCanvas is a service that depends on a number of tools. Some we built, and then there are other tools we use to run the service. Its what market research companies call full service research – we understand your design question, we collaboratively create the study (using our templates etc.), we have contacts with panel companies if you need a specific type of sample… Once data gathering is complete, you get visualizations and all the data within a 1-2 days.
We use a bunch of tools. These are the tools that you will see/use
–Game-like elicitation methods (OpenSort, TreeSort, DivideTheDollar, VisualChoice, FreeList)
–Interactive Visualizations (Dendogram, VocabularyBrowzer, SimilarityBrowzer, ListMap, OnionMap)
Other software we use at our end
-Off the shelf statistical software: SPSS, Excel, Clustan, data mining software and others – too numerous to list.
-Other tools/utilities we built – for designing studies, data transformation, conversion to XML etc.
The researcher expertise and all this comprises the MindCanvas service.
The question I have been asked, why a service – why not a tool you can use yourself?
We did consider this option right from the beginning, and its never been off the table. We are interested in creating a self-service version of the service – if it makes business sense. But currently, we are focused on building a sustainable business and understanding market demand. If it ever seems like a good business idea, we will definitely go the self-service route.
Second reason, we want to focus on very high quality research. Yes, the service is for rapid research, but without sacrificing the quality. For such research, you need research & statistical expertise. Philosophically, we do not want to build tools that attempt to replace this human expertise. Rather, we wanted tools that augment this expertise (I am very inspired by Douglas Engelbart’s augmentation, not automation).
This is also the reason we did not build a statistical engine. There are great statistical tools on the market, better than anything we could build. We use them and are quite happy with them. However, those engines provide really bad output – ugly graphs and charts. So we focused our energies on building a visualization engine that could take the analysis output and create interactive visualizations. (the way it works – our data gathering system churns out the data in XML format, one of us does the statistics, and feeds it into the visualization engine. We did not attempt to replace the role of the researcher by having an automated analysis-visualization process.)
So what does this mean for a self-service (more automated model)? Currently, I am not convinced that we could put enough intelligence in the system to reduce the role of the researcher. For example, let me describe the role of the researcher in creating a good dendogram. The researcher makes a lot of choices regarding the analysis. In creating a dendogram, there are 5-6 parameters, and 2-6 choices for each parameter. Turns out that all dendograms are not the same. The same data could give you a great, or a terrible design solution depending on the parameters you choose.
But I expect and hope that as we gain for more experience, we will be able work out some best practices or typical scenarios and embed that into a more intelligent system. At this moment in time, we do not have enough experience to make the right choices.
We have been consulting for a little while and have seen our share of problems with the design process. I believe that something like MindCanvas where you have lots of design feedback from engaged users, organized in easy-to-understand, shareable visualizations, can change how design teams collaborate and decisions get made.
As I mentioned earlier, if MindCanvas interests you, drop me a line. We are carefully listening to what people have to say – remember we are in Beta phase right now, still very much figuring it out…