I think the key as a business is to avoid being the sort of tool you only need a few times a year. In other words, the best solution to this problem is to cult

I think the key as a business is to avoid being the sort of tool you only need a few times a year. In other words, the best solution to this problem is to cultivate users who use your thing constantly. The epitome of this for me is RedGate's old Ants Profiler. It was the most awesome thing ever when it came out, and a day with it could get you a 1000X performance improvement in your code. But after a few days, you would have optimized everything to within an inch of its life. And the thing came with a 30 day trial. By the end of that month, there was no need to buy it any more, so it got uninstalled and forgotten. Then one year later, you had a new project for a new company that had a performance bottleneck in a critical piece of code. So you'd download a new trial with your new company email address. I actually see this usage pattern with S3stat (one of my products). Users will get excited for a few months, then drift away and cancel, only to pop back on the radar 16 months later. I used to welcome them back and kick off a job to run reports for all those intervening months. Now, I welcome them back and drop a notification on their dashboard with a link to purchase service for those 16 months they were away. It's a service with a real daily workload, regardless of whether you happen to check your reports that day (they still need processing so you can see them when you do eventually come back), so nobody has ever complained about this. Contrast to Twiddla (another of my products), which only uses our resources while people are actually using it. For that, I'm happy to see Teachers cancel their account for the summer. I've even refunded customers who "forgot to cancel" for entire years at a time. So, 500 words later, I guess the answer is that it depends on the business.