Parallel vs serial

[William Pietri]

Yesterday I met a HCI prof at Stanford, Scott Klemmer
<http://hci.stanford.edu/%7Esrk>, who has done some neat research:

http://hci.stanford.edu/research/prototyping

In particular, I thought " / The Effect of Parallel Prototyping on
Design Performance, Learning, and Self-Efficacy
<http://hci.stanford.edu/publications/2009/EffectOfParallelPrototyping.pdf>"/
was relevant here.

They had designers produce 6 versions of an ad. Half started with 1
version and iterated. The other half started with 3, iterated once on
the best 2, and then iterated once more on 1. They took the final ad
from each designer and ran them on MySpace.

It turns out that the parallel efforts yielded noticeably better results
in click-through, time on site, and bounce rate. Interestingly, the
designers also responded better to feedback in the parallel case.

I've often told clients that teams with exactly one idea are at more
risk than teams with zero or multiple ideas, and it's interesting to see
that at least partially backed up by research. This also fits in nicely
with the extensive A/B testing I see more and more teams doing.

William

[Adrian Howard]

On 14 Oct 2009, at 01:40, William Pietri wrote:

> Yesterday I met a HCI prof at Stanford, Scott Klemmer <http://hci.stanford.edu/%7Esrk 
> >, who has done some neat research:
>
> http://hci.stanford.edu/research/prototyping
>
> In particular, I thought " / The Effect of Parallel Prototyping on  
> Design Performance, Learning, and Self-Efficacy <http://hci.stanford.edu/publications/2009/EffectOfParallelPrototyping.pdf 
> >"/ was relevant here.
>
> They had designers produce 6 versions of an ad. Half started with 1  
> version and iterated. The other half started with 3, iterated once  
> on the best 2, and then iterated once more on 1. They took the final  
> ad from each designer and ran them on MySpace.
>
> It turns out that the parallel efforts yielded noticeably better  
> results in click-through, time on site, and bounce rate.  
> Interestingly, the designers also responded better to feedback in  
> the parallel case.
>
> I've often told clients that teams with exactly one idea are at more  
> risk than teams with zero or multiple ideas, and it's interesting to  
> see that at least partially backed up by research. This also fits in  
> nicely with the extensive A/B testing I see more and more teams doing.

[snip]

Interesting - and matches my experiences. I think many design folk  
have already latched on to the fact that parallelism works. For  
example you have the "double diamond" design process promoted by the  
Design Council in the UK (http://is.gd/4jdaz) where you have two  
cycles of exploring multiple solutions and then narrowing them back  
down again.

Cheers,

Adrian
--
http://quietstars.com  -  twitter.com/adrianh  -  delicious.com/adrianh