"Free forever" is on three of my pricing pages. It's the most expensive line on any of them.
Not in dollars. In constraints. The moment I write those words, a set of decisions gets made for me, and most of them are architecture decisions dressed up as pricing.
The free tier has to cost nothing to serve
Forever is a long time to lose money. So the first rule is that a free user has to be nearly free to keep. MyUptimeBot gives you one monitor at three-minute checks. Paid plans check every forty-five or thirty seconds. That gap isn't a marketing lever; it's the scheduler. One free monitor at three-minute intervals is 480 requests a day, which rounds to nothing. Fifty monitors at thirty seconds is 144,000. The free tier is shaped by what the box can absorb without noticing.
The same logic runs through the other two. Leader Ethos gives away one assessment a quarter. The assessment is forty questions and a scoring function. The expensive part, the detailed report, is a PDF that only exists when someone pays for it. Daily Sleuth and Redacted are free in full, every day, because a daily puzzle is a static file and static files are free to serve.
If your free tier costs real money per user, it isn't free forever. It's free until you can't afford it.
The free tier has to be complete
A free tier that's a crippled version of the product is a demo. Demos are fine, but they're not what the words say.
So the second rule: the free tier does the job. One monitor on MyUptimeBot actually monitors. It alerts by push and email. It logs incidents. Someone running one site never needs to pay me. The free Leader Ethos assessment returns a real profile with real strengths, not a blurred teaser with a "see more" button. The daily puzzle is the puzzle.
Which means the paid tier can't be "the same thing, but it works." It has to be more: more monitors, faster checks, SMS, a team. More assessments, trend lines over time. The paid product is for people who've outgrown the free one, and outgrowing is a feeling you can't fake with a paywall.
The upgrade has to feel like outgrowing
The pricing page on MyUptimeBot says "upgrade only when you outgrow it," and that line is the test for every feature I consider gating. The question isn't "would people pay for this?" It's "does needing this mean you've grown?"
A second monitor means you have a second site. That's growth. Thirty-second checks mean downtime costs you enough that two minutes matters. Growth. SMS alerts mean you have customers who'll notice before you do. Growth.
Removing a watermark? Not growth. Turning off a nag? Not growth. Those are tolls, and a free tier with tolls is a trial with the timer hidden.
What "forever" forbids
No card required. No trial timer. No countdown email on day twelve. No "your free plan is ending" for a plan that has no end. And no quiet downgrade of the free tier two years in, once enough people depend on it, which is the move that turns "free forever" into a lie with a delay on it.
The last one is the hardest to keep, because two years in you'll have a spreadsheet showing exactly how much revenue is sitting in the free tier. The answer is still no. You said forever.
If you're hiring me
When a client says "we want a free tier," I hear a list of technical requirements: per-user metering, a cost model for the free path, a data model that can tell the tiers apart, and features gated on a dimension that means growth. None of that is a pricing-page task. All of it is cheaper to build in than to bolt on.
If your free tier needs a nag screen to convert, it isn't a free tier. Fix the tier, not the nag.