Here’s three criteria I use when deciding whether to build a feature to close a particular deal:
- Is it a very valuable customer, either directly (revenue) or indirectly (reference customer, gives good beta feedback, etc)
- Is the feature useful for many/most customers, even if not a top priority?
- Is the feature relatively inexpensive to build and support?
If the answer to all three questions is “yes” then I’ll typically approve the work. If even one of the answers is “no” then I’m unlikely to greenlight the feature.
There’s always some gray area, of course. For example, a feature that many customers have asked for in the past is suddenly a high priority for one particular prospect. If the feature is cheap to build, we may build it even if the new prospect is not a hugely valuable deal.
A good technique is to use feature development for sales leverage. Instead of the customer demanding “if you don’t build this feature we won’t sign” your sales team can turn it around “if you sign before the end of the week, we can get it into next quarter’s release. Otherwise you may have to wait 3 more months.” If you’re giving a concession to a customer, make sure you get something in return!
Some caveats:
- It’s really important to train the sales team to never commit to feature development on their own. They must escalate to PM.
- As a PM, your job is to understand the underlying customer need behind a feature request. You might be able to satisfy that need using a different feature that’s easier to build, more generally applicable, etc.
- Sometimes you’ll have to walk away from deals where the custom work demands are too high. This is painful for everyone involved, but the alternative is worse: going out of business because you spend all your precious developer time on features that only a few customers want.
- If you’re on the fence, and if the feature is cheap, go ahead and build it. You’ll win friends for life this way. Just make sure that the customer and the salesperson knows they’re getting a special deal that’s unlikely to be repeated.