Do Personas Matter?

I build pretty inexpensive web applications. Consulting, project management, design, custom programming, deployment, testing and training all together run between $10,000 and $50,000 with the vast majority of the projects being between $10,000 and $25,000. If you have rich custom functionality required, it's always difficult to provide it at those kind of price points. I often find there are many elements I'd like to include in the project but that would add to the cost and therefore make the projects infeasible for my current client base.

On the one hand you could argue that personas are not required. You don't NEED a picture of "Bill, 42, traveling salesman, married, 2 kids (Sue 8, John 12), aging Labrador at home, uses internet to share photos with parents in Ohio, lives in exurb outside of Dallas, drives Lexus, open to technology where it can help his life (loves sat nav in car), but needs support (spend 6 hours in showroom learning how to use it)." Clearly such a picture isn't absolutely required to be able to build a site. However, having such a clear picture of a target user makes it much easier when trying to make design decisions as you just need to ask yourself "what would Bill want" and it's much easier to find the answer. Also, you can sketch out some key personas in an hour for most apps as long as you're speaking to someone who knows their target audience, so the personas themselves don't need to be that expensive to deliver (and you can easily get some random photos from Flickr). In fact, as I think about it, I wonder if it wouldn't be pretty easy to put together some generic baseline personas that could then be customized for any given client, tightening up the development process still further. In fact, as I think about it a little more, maybe a persona generator could be useful. Imagine if you could select demographics, story vignettes that suggest different psychographic profiles (from Myers Briggs to NLP style memory access modalities) and a place to add just a little bit of flair, perhaps specifically helping you with questions to help position them in terms of decision making strategies, whether they're innovators or laggards for technology adoption and the like. Might help to make persona development both more efficient and more valuable - especially when working with clients without a background in developing personas. But I digress . . .

I used to have a slightly expanded discovery process which started with business intent, then moved on to a list of roles and from there came up with the use cases for each role. Sometimes I find I now take a short cut, just going from business intent to use cases and making up the roles on the fly and associating one to each use case, but I could see going back to having an "audiences" discussion with clients and then using personas to make each role more tangible. I'm not sure how much benefit it would provide for the fairly small projects I build, but it might be worth giving it a shot and finding out.

What has your experience been? What are the smallest projects you've successfully used personas on and what are some of the examples of decisions that were either improved or simplified by having a more tangible representation of the target audience?