CHI2006 04.24.2006: Scott Cook, Co-Foudner of Intuit; a Helluva Lot of Faceted Metadata Scott Cook was the opening speaker for CHI2006. I have to say that I was pleasantly surprised by how the preso went, despite a number of logical inconsistencies.

His content generally revolved around using contextual inquiry as a product design tool. His idea seems to be that if you root around with the customers long enough, the "unexpected" pops out and bites your face. I don't know how much I agree with that... but I agree with the spirit of his arguement.

Scott describes the following as a typical product design lifecycle:

Write requirements > Design product > Show product to users

of course, he thinks this is bass ackwards.

Watch users > Design product

Of course this is where is whole premise falls apart ... because based on his anecdotes, it sounds like the typical Intuit product lifecycle is the former. Who needs consistency. This is about IDEAS my friends.

I liked three things most in Scott's presentation.

1. Intuit calls the product approval process "The Greenlight Process". This is far more positive than the standard "Executive Review" process. It's all about getting the green light.

2. At any stage of the Greenlight Process, a product team must simply answer a small number of predefined questions about the product. I am sure that this cuts the review cycle significantly, and prevents the reviewers from biasing their analysis based on their pre-existing opinion.

3. Net-Customer Satisfaction is the key metric they use to measure success. We did this at Navis, and despite the many things Navis did that I feel are wrong, this is the only metric that should matter. The use a 10 point scale 1 - 10.

1 - 6 — Dessenters (d)

7 - 8 — Neutral

9 - 10 — Promoters (p)

They calculate Net Customer Sat thusly:

%p - %d = NetSat

After that, it was pretty much all faceted metadata all the time. It was a fairly interesting day.

Seeing the new eBay redesign (which actually is looking pretty nice) was interersting, and definitely gave me somethings to chew on regarding the display of large quantities of data. They have implemented a system that allows a customer to build and manipulate a query in a fairly intuitive way, integrating search and browse fairly seemlessly.

Much of their work is in line with Flamenco, a newly opensource platform for accessing huge amounts of faceted data.

Also got to check out a Microsoft Research project called Phlat which is built on top of Windows Live and allows you to add tags to any file (e.g.: docs, pics, emails, etc.) It then lets you to build something similar to the query eBay has worked out, though it is far more experimental.

Anyway, all in all, a fairly good day.  ¶ 8:28 AM