I created a full-text-search bookmarking app called historious (http://historio.us). It started out as free for everyone, since I just built it for myself and people found it useful, but I quickly had to charge, after realizing that people had more than my 10-20 bookmarks (some people had 80 THOUSAND).
I changed it to 2,000 free bookmarks, and unlimited paid, which worked well. Then I changed it to 300 free, then to 50 free. The reason why I did this is this:
As per my analytics package, on average, it takes less than two minutes for 95% of my users to upgrade to paid. Basically, from what I can see, users register, try it out with one or two bookmarks, then upgrade immediately if they like it. If they don't upgrade then, they are not very likely to ever upgrade.
Thus, it doesn't make sense for me to support free users (this is very expensive, since it stores the entire page for everything you bookmark, twice), so the free tier is more of a "trial" tier now.
I'm not sure how many people who would have upgraded if they'd tried it saw the limit and said "I won't use this, the free limit is too low", but, based on the data (and it sounds reasonable to me as well), they are not many. People reasonably want a short trial run, and then they either upgrade or don't.
1. How did you deal with existing free users already exceeding your new limits ? (and how did they react) ? 2. Did you see a measurable increase in upgrades after doing this ? 3. Did you see a measurable decrease in free users of the service after doing this ?
Thanks.
I have also needed to cull unused accounts (of free users who haven't logged in in months). This provided a great boost to the service (saved me from needing to keep upgrading the server for the past few months), but it has also produced a few false positives of people who hadn't used their accounts for a few months and came back. Luckily, I keep backups they can just restore.
2. The last limit change was a few days ago, so I haven't had a chance to gather significant data on this. I guess we'll know in a few months.
3. I will also track this, that is a good question. I anticipate that a few users will be deterred from using the service for free, but that might not necessarily be bad (I am not sure).