Strategy — fourth of five IA Summit 2007 themes

(Here’s the fourth post on the 2007 IA Sum­mit. You can find the first one that intro­duces the series and describes the first theme ‘tan­gi­ble’ here, the sec­ond one on ‘social’ here and the third one on ‘web of data’ here.)

Like oth­er design dis­ci­plines, IAs are typ­i­cal­ly brought in to solve a prob­lem. The extent to which the design prob­lem is defined and expli­cat­ed is a huge deter­min­ing fac­tor in the suc­cess of their under­tak­ing. More often than not, an IA would take a prob­lem and run with it, not think­ing whether this is the right prob­lem to solve, or even a prob­lem at all!

This has always seemed like a sil­ly sit­u­a­tion to me. Some of the most enjoy­able ses­sions at the sum­mit there­fore were the ones that dis­cussed ways in which IAs can join in on strate­gic think­ing. This way, we can help dis­cov­er the actu­al prob­lem that needs solv­ing, which gives us a bet­ter chance of actu­al­ly deliv­er­ing a suc­cess­ful and valu­able solu­tion.

Gene Smith and Matthew Milan dis­cussed con­cep­tu­al mod­els (which I’ve been play­ing around with for a while) and the more involved rich map­ping, from soft sys­tems think­ing. Key take­away for me was when mod­el­ling a sys­tem we should also describe its con­text (includ­ing the project itself). Oth­er good stuff by peo­ple of Crit­i­cal Mass (Milan again togeth­er with Sam Lad­ner) was pro­vid­ed in the form of ‘back­cast­ing’, a very visu­al brain­storm­ing method to be used in a work­shop ses­sion with a client in order to envi­sion desired project out­comes and map paths from the cur­rent sit­u­a­tion to those out­comes (notes at The Chick­en Test).

Peo­ple from Avenue A Razor­fish (Gar­rick Schmitt, Marisa Gal­lagher) talked about their frame­work for tying togeth­er lots of dif­fer­ent user research such as click stream analy­sis, search logs, eye track­ing and oth­ers. This remind­ed me of Jared Folkmann’s excel­lent talk at last year’s Euro IA Sum­mit in Berlin.

Final­ly, I attend­ed one nice talk (by James Robert­son) on the val­ue of con­tex­tu­al enquiries, which if noth­ing else has made me all the more deter­mined to try this myself the next time an oppor­tu­ni­ty presents itself.

Using concept models to design for the web of data

Flickr concept model by mApplogic

I’m lucky enough to be doing some con­cept­ing and inter­ac­tion design work for a social web site. This pre­sent­ed me with the oppor­tu­ni­ty to inte­grate some stuff I found while read­ing on social soft­ware, and the web as platform/network. Here’s how I’ve been inte­grat­ing some of it.

I was inspired by the con­cept mod­el of the Flickr ecosys­tem I saw in Luke Wroblewski’s pre­sen­ta­tion on social inter­ac­tion design (which was done by Bryce Glass) to try and cre­ate one myself. Coin­ci­den­tal­ly there’s a whole chap­ter in Dan Brown’s book (which Peter was smart enough to pur­chase and was lying around the office) on cre­at­ing con­cept mod­els.

One of the things I want­ed to do is make the site play nice with the web of data. To that end, I decid­ed to apply Tom Coates’ 3 basic page types to the design of the site. So what I did was first cre­ate a con­cept mod­el (of course fol­low­ing some research of the site’s busi­ness and user goals) and then look at the nouns and verbs in the mod­el. For each noun I cre­at­ed a sin­gle object view page and a list view page. For each verb I cre­at­ed a manip­u­la­tion inter­face page. Of course, all list type pages would get RSS feeds in the even­tu­al site.

For instance if you have a mod­el that states ‘Review­er rates Book’ then you’d end up with a page for each review­er and book, a page to list review­ers, a page to list books and a manip­u­la­tion inter­face for rat­ing a book.

Doing this result­ed in a nice list of pages that I could then analyse for com­plete­ness and/or redun­dan­cy. Of course this only works if your con­cept mod­el accu­rate­ly reflects what the site should achieve. If your mod­el sucks, your list of pages will too.

Anoth­er caveat lies in the fact that a con­cept mod­el tends to be very effec­tive for map­ping the func­tion­al aspects of a site, but not very suit­able for cre­at­ing an overview of its con­tent (which is often more push ori­ent­ed). If the kind of site you’re cre­at­ing involves more infor­ma­tion archi­tec­ture than inter­ac­tion design you might want to do some addi­tion­al con­tent inven­to­ry work and fold that into the page list.

One last chal­lenge would be orga­niz­ing these pages in a coher­ent whole (beyond cou­pling lists to sin­gle items to inter­faces). I can imag­ine I’d attempt some card sort­ing to achieve that.

Final­ly, for cre­at­ing the con­cept mod­el I used the spe­cial­ized (and free) tool Cmap­Tools which is pret­ty nice in that it goes beyond visu­al­ly mod­el­ling the con­cepts but actu­al­ly track­ing the state­ments you implic­it­ly make when link­ing con­cepts to each oth­er.

Any­one else have expe­ri­ence with try­ing to inte­grate some of the stuff Coates was talk­ing about in their design of a site?