Tuesday, May 13, 2008

Say No to Feature Creep

Chris Saad has a great post over on his blog: Leadership includes saying no « Paying Attention

Not sure if he just went through a specific scenario to inspire the posting but it's a great read, especially for technical managers.

One of my clients had a problem: they have a list of over 397 enhancement requests to their product. Now, to be fair, some of those are fairly cosmetic and but at least 200 of them require some kind of review for someone to say "no - we are not doing this"

So why don't they? Because when they get in front of their customers, "no" is the hardest thing to say. And I'm a terrible accomplice but because if they ask "can it be done?" , the answer is "yes, it can". The real question should be "SHOULD it be done?"

When some people think about feature creep, they tend to think of it in relevance to a particular function - but in this post, think about feature creep being about the entire product line.

As a result, the client has about 30 open projects that have yet to be prioritized and when they return from customer visits, there might be some more.

Why not add more developers to get them all done?
Before you do that, you have to ask the question : are they all really needed? And then, are they all needed tomorrow?

So we tried to devise an easy way of identifying a real priority. I used to take a listing of all the issues and sorted it based on how many calls we had for a particular feature. The problem with that is that customers tend to be focused on the "issue of the day" as opposed to the "vision for the future", a vision that typically appears as a high priority a few days after we've put our heads down and gotten all the "issues of the day" done.

1. Narrow your focus, for at least a single release.
2. Publicize that focus - make sure no one is unaware of what your current focus is, and make sure no one takes you aware from that.
3. Make that release a short one (no, I don't mean in the sense of Agile with a two week turnaround, although that might not be a bad idea)
4. Have the managers who continually try and re-prioritize items start planning the NEXT release, instead of throwing more irons on the current fire.

There are a lot of benefits to this approach:
a) you have a "focused" release that can be easily marketed.
b) you can deliver a "focused" message (instead of one that's all over the map)
c) there's a finite list of items to work on.

The problem with this approach?
It requires managers who are able to do that - in short, managers/clients who are able to say "no", even if all they have to say is "no - not just yet"

Certainly, hearing "No" is something VFP developers have heard for a while. "64-bit?" no. "VFP 10?" No. "Open EULA?" No. "Access to core code?" No. But we haven't been told no about stopping what we're already doing. There's certainly a limit to what's possible in the "no" world but one "no" does not mean no more "yesses".

What about you?
Have you ever turned away a client or turned off a feature? How did you say No?   What thought pattern did you go through?

1 comment:

Tod McKenna said...

I find the following helpful:

Use a positive "no", whenever possible. I find that saying something like, "we can think about including that in a future release" usually works. If the customer still complains, then likely they see the new item as critical (and therefore should be a priority). If you never hear about it again, well then you win!

I've also used a feasibility graph with success as well. The x axis represents the combination of time and resources needed to implement the feature (most difficult is close to 0,0), while the y represents business value (least valuable is at 0,0). You can often lop off all low valued and difficult tasks using this method.

Great post, by the way. This is a topic that needs much more discussion. As developers, we're often in this situation where we need to say no, but we often can't quite justify it in a way that resonates with the customer!