Traffic is up. Signups look fine. Revenue is flat. If that's your dashboard, the leak isn't where you keep looking. It's below the funnel you're staring at, in the gap between "created an account" and "reached the moment your product proved itself."
Key takeaways
The five most common SaaS UX mistakes: signup forms that ask too much too early, login walls before any value, slow pages, no guided path to first value, and hidden pricing. Each is checkable, each has a number behind it, and each belongs to a product designer, not a marketer.
That gap has a number. Across 62 B2B SaaS companies benchmarked by Userpilot (who, disclosure, sell onboarding tooling), average activation is 37.5%, with a median around 37% [C1]. Read that slowly: at well-run companies, companies organised enough to measure activation at all, roughly six in ten signups never reach first value. They sign up, poke around, and quietly leave.
Founders usually respond by pouring more into the top: more ads, more content, more outbound. But if the bucket loses 60% between signup and value, more water isn't the fix. The bucket is.
The good news: these five conversion killers aren't mysterious. Each is specific and checkable, each has a real number behind it, and each has an owner. Here they are, in the order your users hit them.
The best-studied friction in the digital economy isn't SaaS signup. It's ecommerce checkout, where Baymard Institute has spent more than fifteen years watching people abandon purchases. Their numbers: average cart abandonment sits at 70.22% across 50 studies [C4], and 17% of US online shoppers have abandoned an order over a "too long / complicated checkout process" [C5]. The average US checkout flow carries 23.48 form elements when 12–14 would do the job [C7]. Even by 2024, the average checkout still asked 11.3 form fields when most sites need only 8 [C8]. Baymard's estimate of the prize: the average large ecommerce site can gain a 35.26% increase in conversion rate through better checkout design alone [C6].
Let's be straight about what these numbers are: checkout studies, not SaaS signup studies. Don't quote "70% of your signups abandon" in your board deck; that's not what the data says. But the mechanism transfers cleanly, because a signup form is a checkout form wearing different copy: a stranger, mid-decision, being asked for information before they've received anything. Every field is a small tax on a commitment they haven't fully made yet.
Audit your own signup flow today. Count the fields. For each one, ask: do we need this before the user experiences value, or are we collecting it now because it was convenient to build it that way? Company size, phone number, "how did you hear about us" — all of it can wait. Cut to the minimum viable ask; defer the rest to a moment when the user has a reason to answer.
Nielsen Norman Group made the case back in 2014, and nothing about human psychology has changed since: forced registration before a user gets anything "has high interaction cost and defies the reciprocity principle" [C11]. You're demanding a favour (an email, a password, a commitment) from someone you haven't given anything to yet. People reciprocate generosity. They resent tollbooths.
The fix is sequencing, not surrender. Show the value first: a sandbox, a public demo, a template gallery, a result computed from one input. Then ask for the account when the user has something worth saving.
There's a related trade-off worth naming honestly, because you'll see the numbers quoted at you. In First Page Sage's dataset of 86 companies (71% B2B, through Q3 2025 — note this is an SEO agency's own client data), opt-in free trials convert organic signups to paid at 18.2%, while opt-out trials requiring a credit card up front convert at 48.8% [C9]. That looks like an argument for the card wall. It isn't, automatically: card-required trials convert entrants at a higher rate partly because far fewer, far more committed people enter. It's a filter, not a fix. If your constraint is intent (lots of tyre-kickers, limited support capacity), the gate may serve you. If your constraint is volume, you're starving an already thin funnel. Choose deliberately; don't inherit the default.
In 2019, Deloitte studied 37 leading brand sites and 30+ million sessions for Google and landed on a figure that still gets quoted for a reason: a 0.1 second improvement in mobile site speed increased conversion rates by 8.4% for retail sites [C10]. Retail, 2019 — label it honestly. But the direction holds everywhere it's been measured: speed is not polish, it's conversion infrastructure. A tenth of a second, an 8.4% swing. That's the sensitivity of a stranger's patience.
Early-stage SaaS ships heavy pages for a boring reason: nobody owns performance. The marketing site accretes tracking scripts, the app bundles everything on first load, and every sprint adds weight because no sprint is ever about removing it. Meanwhile your prospect is on a phone, on hotel wifi, deciding whether you're worth three more seconds. Run your signup path through a page-speed test this week. If the numbers embarrass you, that's a product task with a measurable payoff, not a someday item.
Here's where most of that 37.5% activation gap [C1] actually lives. The signup succeeds. The confirmation email fires. And then the product opens on an empty screen with fourteen navigation items and no opinion about what the user should do first.
Users don't push through that. In a 2020 Wyzowl survey (216 respondents, general app users rather than B2B specifically — small sample, but the finding is stark), 8 in 10 users said they've deleted an app because they didn't know how to use it [C12]. Not because it lacked features. Because they couldn't find the door.
The cheap, research-settled fix is consistency. Nielsen Norman Group's usability heuristic #4: "Users should not have to wonder whether different words, situations, or actions mean the same thing. Follow platform and industry conventions" [C2]. Jakob's Law says the same thing from the user's side: users spend most of their time on other sites and products, so they prefer yours to work the way everything they already know works [C3]. Your onboarding doesn't need to be inventive. It needs to be familiar, and it needs to guide the user to one first win. Not a feature tour, not a six-step checklist covering every module. One task that produces the moment where the product earns its keep. Everything else in the first session is a distraction from that moment.
The number here is refreshingly current: in January 2025 TrustRadius data reported by eMarketer, pricing transparency is the No. 1 change B2B tech buyers want from vendors, cited by 45% [C13]. Not better demos, not more case studies. Just tell me what it costs.
Early-stage SaaS hides pricing for a copied reason: the big vendors do it. But "talk to sales" works for enterprise vendors because they have enterprise sales teams and brand gravity; a buyer will book the call because they need the product and there are three vendors in the category. You have neither the sales capacity nor the pull, so your hidden pricing page doesn't create conversations; it ends them. Buyers assume the worst number and shortlist someone else before you know they existed.
And no, this doesn't contradict the follow-conventions advice from Mistake 4. Jakob's Law says match the patterns your users know and expect; it is not context-blind imitation of companies whose go-to-market you don't share. The convention your buyers actually want is a public price.
Look back at the five: form friction, value sequencing, page weight, first-run flow, pricing presentation. None of these is a marketing problem. None is solved by a graphics designer making things prettier, or by more ad spend pushing more people into the same leaks. They're product design problems — flows, states, defaults, friction — and if you're weighing which role to hire first, this is the case for a product/UX designer.
The tempting alternative is "we'll just A/B test it." Here's the arithmetic problem, stated illustratively: detecting even a substantial lift, say 20%, with statistical confidence typically needs thousands of visitors per variant. At early-stage traffic, that's months per test, per change. You'd spend a year learning what the research above already settled. The pre-scale playbook is the opposite: fix the five known leaks first, on evidence that already exists, and save experimentation for big swings once you have the volume to measure them.
If nobody on your team owns this work, that's the actual gap, and it's a well-scoped one. A UX audit of your activation funnel, from landing page to first value, is exactly the kind of contained first task DevOD was built for: a UX engineer who does product and UX design on one subscription alongside developers and QA, with a 15-minute fit call and a one-task Proof of Quality before you subscribe, first task delivered within 5 business days on a 3-day task cycle, $3,495/month for a single role, add or change roles any month, cancel any time.
Because the uncomfortable, fixable truth is this: the traffic isn't leaking. The product is.
Userpilot's activation rate benchmark of 62 B2B SaaS companies puts the average at 37.5%, with a median around 37% [C1]. If you're near that, you're typical. And typical still means roughly six in ten signups never reach first value.
It depends on your constraint. In First Page Sage's client data (86 companies, 71% B2B), opt-in trials convert organic signups to paid at 18.2% versus 48.8% for card-required trials [C9]. But the card wall is a filter, not a fix: far fewer, far more committed people enter. Gate for intent; open up for volume.
The best evidence comes from checkout rather than SaaS signup: 17% of US online shoppers have abandoned an order over a "too long / complicated checkout process" [C5], and the average US checkout carries 23.48 form elements when 12–14 would do [C7]. Every premature field taxes a commitment the user hasn't made yet.
In Deloitte's 2019 study for Google (37 brand sites, 30+ million sessions), a 0.1 second improvement in mobile speed increased retail conversion rates by 8.4% [C10]. It's retail data, but the direction holds: speed is conversion infrastructure, not polish.
Usually not first. Detecting even a large lift with statistical confidence typically needs thousands of visitors per variant, which at early-stage traffic means months per test. Fix the research-settled leaks first and save experimentation for big swings once you have the volume to measure them.
