Subject: Re: success metrics -- user adoption rates
From: Paul Rohr <>
Date: Fri, 11 Jun 1999 22:25:56 -0700

Stephen's criticisms of my problematic metric are a good clarification, so 
I'll let them stand as is.  

At 12:21 PM 6/9/99 +0900, Stephen J. Turnbull wrote:
>One problem with the contest is that it doesn't measure "good," it
>measures "actual".  

Actually, I guess I was trying to really measure just two things:

  - whether we were doing better or worse than people expected, and
  - how diverse a range of expectations we'd set. 

Insofar as we had a reasonably tight cluster of guesses at about half our 
actual numbers, plus a few really diverse outlying points, I'm pretty happy 
with the results.  They're pretty sparse, but I'm happy enough for now.  
I'll be even happier the day we exceed the expectations of the highest 

>Next time, would you do the contest in terms such
>as "how many downloads do we need to pay our mortgages?" or "If Redhat 
>was distributing this product under their brand, how many downloads do 
>you think they would get?"  :-)

Unfortunately for you, neither of those questions are likely to elicit 
answers which help make people feel better about my products or my company, 
so I'll pass.  ;-)

>But you're doing the community a real service by educating
>a few people about what a real FSB is doing in terms of volume.  

Thanks.  Of course, as a 100% GPL business writing new software, "doing the 
community a real service" is an inescapable part of our mission.  

>I bet
>there are several people on this list who do "custom" rather than
>"pre-packaged," and have little idea of what kind of adoption rate
>something like AbiWord could get.

Of course, as you pointed out earlier, those numbers are highly contingent 
on the specific circumstances of each FSB.  Our press presence is still 
effectively nil, and our only distribution channel is an email list and a 
website sitting behind a T1.  Each of those factors has clearly been a 
constraint for us.  

As I alluded to before, with either more features or Red Hat's marketing 
budget, those numbers would definitely change.  A lot. 

>I wonder if all the FSBs out there shouldn't be doing something like
>it.  :-)

Doing market research in public can be a risky thing, so I can understand 
that not everyone's free to disclose their results.  

Anyone who does a similar analysis, please send me a copy of your results.  
Preferably not under NDA, but I'll take 'em any way I can get 'em.  :-)


PS:  I didn't have any private guesses from members of this list.  I suppose 
that means that fsb expertise isn't too relevant in this kind of contest,