Home - Product Management Resources

When Dropbox filed to go public in 2018, it reported more than 500 million registered users and about 11 million paying ones, according to its S-1 filing. Roughly 2 percent. The free tier was not a pricing decision that happened to stay on the books. It was the company's acquisition engine, its biggest cost line after the paid infrastructure, and the reason the business existed at the size it did. Most product teams decide on a free tier the way they decide on a font: someone proposes it, nobody objects, and it ships. The questions below are the ones I would want answered before that happens, because a free tier is far easier to launch than to take back. Does freemium actually convert better than a free trial? Not by as much as people assume, and the answer depends on where you measure. The ChartMogul Conversion Report, built from a survey of about 200 software products, puts the "good" range for freemium at 3 to 5 percent of free signups converting to paid within six months, and "great" at 8 to 12 percent. Free trials sit slightly higher on the same measure. Trials that require a credit card up front land around 25 to 35 percent. The interesting part is what happens at the top of the funnel. Per 1,000 website visitors, the report's freemium example produced about 90 signups and roughly 5 paying customers. The free trial example produced 45 signups and 3.6 payers. Freemium wins on volume, trials win on rate, and the end result is closer than either camp admits. So the honest answer is that freemium is not a conversion tactic. It is a bet that a larger pool of free users creates something a smaller pool of trial users cannot: word of mouth, team invitations, data, or habit. If your product creates none of those, you are paying to host people who were never going to buy. What should the free tier give away, and where should it stop? This is the strategic question, and it is where the fence goes. The fence is the limit that turns a free user into a paying one, and its placement decides who your product actually serves. Slack placed its fence on history. Its 2019 S-1 explained that free organizations often upgraded as they engaged more deeply, frequently because they wanted to search beyond 10,000 messages or needed single sign-on. Both limits scale with success. A team that barely uses Slack never hits the message cap. A team that runs its operation in Slack hits it fast, and by then the product is holding knowledge they cannot afford to lose. The fence sits exactly where the value compounds. Evernote placed its fence on convenience. In June 2016 the company limited its free Basic plan to two synced devices and raised prices on its paid tiers at the same time, giving existing users with more devices 30 days to cut back. The limit had nothing to do with how much value a user was getting. It caught anyone with a phone, a laptop and a tablet, including people who opened the app twice a month, and the reaction was predictably angry. A useful test: does the fence get closer as the customer gets more value? If yes, upgrading feels like a natural next step. If the fence is arbitrary, upgrading feels like a ransom. Who is the free tier really for? Answer this one in writing, because the answer changes what you build. Running IT operations at a large telecom, I was on the other side of this more than once. We ran a network monitoring tool on its free tier for close to two years. The cap was a device count, and we simply kept the deployment under it, monitoring the systems that mattered most and leaving the rest on older tooling. The vendor almost certainly counted us as an active free account with upgrade potential. We were never going to pay, because the fence was on a number we controlled, and the product never gave us a reason to cross it. What the vendor never learned was that the person using the free tier (me, an operations lead) was not the person who signed purchase orders. A free tier aimed at practitioners only converts if something carries the product from the practitioner to the budget holder: an admin feature, a security requirement, a report the executive wants. Slack's single sign-on limit did exactly that, because SSO is a requirement that comes from IT and security, not from the people chatting. If you cannot name the moment a free user hands your product to someone with a budget, the free tier is serving users, not customers. That can still be fine, but call it what it is: marketing spend. What does a free tier cost that doesn't show up on the dashboard? Infrastructure is the visible cost. The invisible ones tend to be larger. Free users file support tickets. They generate abuse. They anchor your product's perceived value at zero, which is a problem I wrote about in Pricing Is Product Strategy. And they shape the roadmap, because the loudest feedback often comes from the largest group, and on a freemium product the largest group pays nothing. Heroku is the clearest public example of a free tier outliving its purpose. In August 2022, the Salesforce-owned platform announced it would end its free plans that November, with its general manager writing that teams were spending "an extraordinary amount of effort to manage fraud and abuse." The free tier had built Heroku's reputation among developers a decade earlier. By 2022 it was consuming the engineering time that paying customers needed, and removing it cost Heroku a large share of the goodwill it had once bought. That is the asymmetry worth respecting. A free tier is cheap to add and expensive to remove, because every user who joined under it treats it as a promise. So, should you have one? A free tier earns its place when three things are true: The product gets better as more people use it, through invitations, shared workspaces, or network effects. Otherwise volume is just cost. The fence tracks value. The limit should be hit by your best free users, not your most casual ones. There is a path to a budget holder. Some feature, limit or requirement must move the product from the person who loves it to the person who pays for it. If one of the three is missing, a time-limited trial is usually the better call. It is easier to extend a trial into a free plan later than to claw a free plan back, as Heroku and Evernote both learned in public. And if distribution is the real constraint on your growth, decide that first; freemium is one answer to that problem, not the only one. Before the team ships a free tier, write down the fence, the user it is for, and the moment that user turns into revenue. If those three sentences are hard to write, the free tier is not ready, no matter how simple it looks on the pricing page.

Recent Posts