Vince Baskerville makes some good points and some not-so-good points:
When deciding on adding new features, take a moment to really think about what it will cost. Every feature has an associated price. One of the many things people hear me say is "What is the negative value?" Meaning, just because it seems to give a temporary gain, is it genuinely worth whatever loss over time?
This is certainly spot on. However...
(...) this is the basic concept behind the 90/10 rule. If you aren't familiar, it is basically a concept that states on average, your user will spend 90% of their time using only 10% of your product. Only 10%. Think about some of your non-tech friends. How many of them continue to use a product like Microsoft Word, yet barely know how to actually do anything outside of the 10-15% range of what it could actually do? Now peer back at your tech. friends, what about products like Adobe Photoshop, or any plethora of computer software? Many software companies keep coming up with feature after feature for every new release, yet almost no-one actually uses them all (and I would put money on how this is a small minority), and frankly it just makes the programs unnecessarily more complex.
This is, unfortunately, a thorough misunderstanding of usage patterns.
Take Excel as an example, bloated software par excellence. Yes, each user uses maybe 5% of Excel. Power users perhaps use 10% of its capabilities. So, should Microsoft chop off 90% of Excel?
The fallacy of this 90/10 rule is that even though 90% of your users only use 10% of your product, they each use a different 10%. Every single feature in Excel is being used by someone. Most of them were paid for, directly or indirectly, by someone who really needed it. Cut a single feature of Excel, and a hundred people across investment bank front office departments will scream out in horror and refuse to upgrade.
Photoshop is another great example. Photoshop is a program with incredible depth. You can spend years learning to use Photoshop and still only use a small fraction of its features. But Photoshop is used for so many different things, that every one of these features is in fact being used. Cut a single one out, and Photoshop will lose sales - possibly a lot of sales. That's why they don't cut features.
On the good side, this cuts both ways. Any specific, niche group of users will only use 10% of the potential feature set. So you can focus on just those users, and build that feature set better than any other player on the market, and therefore capture that niche.
As software development gets continually cheaper, new niches become commercially viable all the time - and you don't need to build all of Photoshop to get them.
If you read this far, you should follow me on twitter here.