Why 54% of SaaS products choose freemium
Our analysis of 1,000 SaaS pricing pages found freemium is now the dominant acquisition model. The more useful finding is where it stops working.
Our analysis of 1,000 SaaS pricing pages found freemium is now the dominant acquisition model. The more useful finding is where it stops working.
When we started reading pricing pages, we expected freemium to look like a bet — a model chosen by a confident minority with the runway to give product away. It is not a bet any more. Slightly more than half of the products in our sample offer a permanently free tier, and among the ones that do not, the overwhelming majority substitute a time-boxed trial. Charging at the door has become the unusual choice.
54.1% offer freemium. That figure is doing a lot of work, though, because “freemium” covers two quite different things. In roughly two-thirds of cases the free tier is a real, indefinite plan with a hard limit — seats, records, projects. In the rest it is closer to a trial wearing a free plan’s clothes: functional, but capped so tightly that nobody stays.
Shares sum past 100% — most products run more than one entry path.
We expected category to be the strong signal — developer tools free, finance tools paid. It was not. Tier count was. Among products with four tiers, 67% include a free plan; among three-tier products, 41%. The free plan is functioning as the bottom rung of a ladder rather than as a separate product, and you need enough rungs above it for the climb to make sense.
A free tier without a credible next step is not an acquisition model. It is a cost centre with a signup form.
Two patterns showed up repeatedly in the products that offered freemium and then walked it back. The first is the limit nobody hits — generous enough that the free plan is simply the product, with a paid tier that adds administration rather than capability. The second is the limit nobody understands: a cap expressed in a unit the buyer does not yet have a feel for, so the upgrade moment never arrives as a decision, only as an interruption.
The products that made it work shared something dull: the limit was in the same unit as the value. Seats when the value is collaboration, volume when the value is throughput. When the meter measures the thing the buyer already cares about, the upgrade explains itself.
We are testing the next part of this live. The Experiment Network is currently running CTA language across 47 sites, and the first result — “Start Free Trial” outperforming “Get Started” by 28% — suggests the word “free” carries weight well past the pricing page.