Only 19.2% of new users complete the onboarding checklist a SaaS product puts in front of them, and the median product manages 10.1%, according to Userpilot's 2025 benchmark of 188 companies on its platform. The same report puts average activation at 37.5% across 62 B2B companies. SaaS onboarding flow design therefore starts from a hard fact: six in ten people who sign up never reach the moment the product was built for, and the screens between sign-up and that moment are where they leave.
This guide walks through the seven screens most B2B onboarding flows contain, puts the published drop-off data next to each one, and ends with an order for cutting. The numbers come from Userpilot, Chameleon, ChartMogul, ProductLed, Navattic, Nielsen Norman Group and Baymard, and every vendor figure is labelled. Our SaaS onboarding statistics post holds the full benchmark tables; this one is about the screens.
The seven screens in SaaS onboarding flow design
Almost every self-serve B2B product uses a subset of the same seven screens. Naming them matters because the drop-off data is collected per element, not per flow, and the cut list at the end refers to them by number.
| # | Screen | Job |
|---|---|---|
| 1 | Sign-up form | Create the account with the least that is needed |
| 2 | Verification or payment wall | Confirm the email, or take a card |
| 3 | Welcome survey | Ask the two or three questions that change the next screen |
| 4 | Workspace setup | Name the project, connect data, import or use a sample |
| 5 | Product tour | Point at the interface |
| 6 | First action or empty state | Get the user to do the one thing that predicts retention |
| 7 | Checklist | Hold the next three to five tasks across sessions |
Not every product needs all seven. A tool that works on a sample dataset can skip screen 4. Most can skip screen 5, as the evidence below shows.
Screens 1 and 2: the sign-up form and the wall behind it
How much friction is normal. Userpilot's 2023 State of SaaS Onboarding review of 100 SaaS tools defined a frictionless sign-up as one that asks only for name, email and password and lets the user in without confirming the email or entering a card. By that definition 79% of tools had a friction-based sign-up, up from 76% in 2022, and 53% to 54% offered sign-up with a Google account. These are teardown counts: what products do, not what converts.
Fields, not steps. There is no public benchmark for SaaS sign-up form length, so the nearest proxy is Baymard's checkout research, which found the average 2024 checkout used 11.3 form fields while most sites need only 8, and that the number of fields affects usability far more than the number of steps. Translated to sign-up: an email step followed by a password step costs less than one screen with five inputs. Our form conversion statistics post has the field-level abandonment data.
The card decision. ChartMogul's January 2026 survey of 200 software products, run with ProductLed, found 20% of free trials require a card up front. Those trials convert at about 30% free-to-paid within six months, more than five times the trials without a card, against a median of 8% across all models. The catch is in the funnel maths below. These are self-reported survey estimates, not verified billing data, and the report says so.
| Model | Sign-ups | Paying customers | Free-to-paid rate |
|---|---|---|---|
| Freemium | 90 | 5 | 5.5% |
| Free trial, no card | 45 | 3.6 | 8% |
| Free trial, card required | 35 | 10.5 | 30% |
Source: ChartMogul, The SaaS Conversion Report, February 2026, 200 products surveyed
Email verification. We could not find a published, independent drop-off figure for an email confirmation step placed before first use; the vendor numbers in circulation trace back to case studies, not a dataset. Treat it as unmeasured and verify later: let the user in, show the product, and confirm the address when they first invite someone or need to receive something.
Screens 3 and 4: the welcome survey and the workspace setup
Nielsen Norman Group's position is blunt: avoid app onboarding where possible and spend the effort on usability. The exceptions it allows are precisely screens 3 and 4, where the product needs information from the user to start, or where what it does depends on the user's context. A survey that feeds a CRM field and nothing else is a form with a friendlier label.
What products do. Userpilot's blog, citing its State of SaaS Onboarding research, says 33% of SaaS companies have no welcome screen at all. In its MarTech-focused review, 76% of companies used some form of microsurvey in the welcome flow. Both are teardown counts, not conversion data.
How long people give a modal. Chameleon's 2025 benchmark, built on more than 550 million end-user interactions on its platform, puts the average modal dismiss rate at 37.5%. Of the modals that are dismissed, 38% are closed in under four seconds. Text-only modals complete at 44%, text with an image at 29%, and text with video at 21%. Modals triggered by a user action complete at 47% to 51%, against 39% for modals that fire on page load. A welcome survey is a modal, so keep it to text and make the first question the one that changes the flow.
One question per page. The GOV.UK Design System's question-page pattern asks one question per page so users can focus on it, and allows grouping only where research supports it. A three-question welcome survey becomes three light steps, each with a reason the question is being asked.
Setup should be optional. Screen 4 is where products ask for an integration, an import or a project name before showing anything. Where sample data can stand in, offer it first and let the real connection wait. Navattic's 2025 report, drawn from about 28,000 interactive demos on its platform, found ungated demos had a 10% higher engagement rate than gated ones, and a form placed mid-demo engaged 9.7% better than one at the start. That is pre-sign-up data from a demo vendor, but it points the same way: show value, then ask.
Screens 5 to 7: the tour, the first action and the checklist
Tours. Chameleon's 2024 report, covering about 300 million interactions in 2023, put the average product tour completion rate at 33.5% and found five-step tours falling to 21.6%. Its 2025 edition says tours beyond five steps lose more than half of users, four-step tours complete best, a progress indicator improves completion by 12%, and tours the user starts get about double the engagement of tours that fire automatically.
Nielsen Norman Group tested the underlying question in a 70-person remote study of four iPhone apps with deck-of-cards tutorials. The 35 users who read the tutorial succeeded on 91% of tasks; the 35 who skipped it succeeded on 94%, not a significant difference. The tutorial group rated the tasks harder, 4.92 against 5.49 on a seven-point ease scale, and that difference was significant. Userpilot's 2023 teardown found 35% of tools used a linear tour, 29% an interactive walkthrough and 36% no in-app guidance at all.
The first action. Screen 6 is the one that matters. Rachitsky and Timen define the activation milestone as the earliest point in onboarding that shows the product's value and predicts long-term retention, and set a bar: users who reach it should retain at least twice as well as users who do not. They warn against defining it as completing sign-up. Across 500+ responses the SaaS median was 30% and the average 36%; the 60th percentile counts as good and the 80th as great. Userpilot's 2025 data puts the B2B average at 37.5% and median time to value at 1 day, 1 hour and 54 minutes.
Checklists. Userpilot's 188-company sample completes checklists at 19.2% on average and 10.1% at the median. Chameleon reports that 23% of users click into its Launcher, 27% when it opens with a welcome state, and that users who do interact complete about five items in a session. A checklist is still worth having because it survives the session boundary that kills tours, but it should be short, start with one item ticked, and lead with the activation action.
| Element | Metric | Figure | Source |
|---|---|---|---|
| Product tour | Average completion | 33.5% | Chameleon 2024 |
| Product tour, 5 steps | Completion | 21.6% | Chameleon 2024 |
| Product tour, over 5 steps | Users lost | More than half | Chameleon 2025 |
| Modal | Dismiss rate | 37.5% | Chameleon 2025 |
| Modal with video | Completion | 21%, against 44% text-only | Chameleon 2025 |
| Checklist | Completion, average / median | 19.2% / 10.1% | Userpilot 2025, 188 companies |
| Launcher | Users who click in | 23%, 27% with welcome state | Chameleon 2025 |
| Whole flow | Activation, average / median | 37.5% / 37.04% | Userpilot 2025, 62 companies |
Sources: Chameleon Benchmark Reports 2024 and 2025; Userpilot Benchmark Report 2025. All are vendor platform data.
What to cut first
The order below weighs the evidence against how much each screen costs the user before first value. Cut from the top.
| Order | Cut this | Keep instead | Evidence |
|---|---|---|---|
| 1 | Linear product tour (screen 5) | Two or three tooltips on the first-action screen | 33.5% completion; no task gain in NN/g's study |
| 2 | Email confirmation before entry (screen 2) | Verify on first invite or export | Unmeasured, and it sits in front of value |
| 3 | Survey questions nothing branches on (screen 3) | One to three text-only questions that change the next screen | 37.5% modal dismiss, 38% within four seconds |
| 4 | Mandatory integration or import (screen 4) | Sample data, real connection as a checklist item | Ungated demos engage 10% better |
| 5 | Profile and preference items (screen 7) | Three to five items, activation first, one pre-ticked | 19.2% completion, five items per session |
| 6 | Video in any onboarding modal | Text, with video in the help centre | 21% completion against 44% text-only |
| 7 | The card wall (screen 2) | Cut for volume, keep for sales-ready trials | 30% against 8% conversion, but 35 against 45 sign-ups per 1,000 |
Sources: Chameleon 2024 and 2025; Nielsen Norman Group 2020; Navattic 2025; Userpilot 2025; ChartMogul 2026
The card wall sits last because it is a business decision, not a usability one. ChartMogul's data says a card-required trial produces about three times the paying customers per visitor of a no-card trial, and 80% of free trial products add human touchpoints such as customer success for self-serve enterprise sign-ups. If the trial is handed to a salesperson, the card may be filtering the right people. If the product grows on volume, it removes the users who would have found value on day 3.
Measure before you redesign
ProductLed's survey of more than 600 SaaS businesses found activation tracked only 34% of the time. ChartMogul's 2026 survey adds that product teams own activation 49% of the time while sales (28%) and growth (26%) own free-to-paid conversion, a split the report blames for activation getting too little attention. In the same survey, 43% of participants had improved conversion in the previous twelve months, most by 10% to 25%.
The minimum instrumentation is four events: account created, first action completed, checklist item completed with the item name, and tour or modal dismissed with the step number. Read them against ChartMogul's timing data: median daily trial conversion sits near zero until the trial's final day, and most conversions land around day 7. A flow that has not delivered first value by day 2 has spent most of its budget.
Onboarding sits inside the application, not on the marketing site, and we build applications in Next.js; our Next.js and Supabase SaaS post covers the auth and billing plumbing the first two screens depend on. On an onboarding project we would wire the activation event and the time-to-value timer before designing screens, so the first release can be compared to the figures here.
What this means for a founder designing the flow
Design the flow backwards from screen 6. Decide the one action that predicts retention, check it against the twice-the-retention rule, and make every earlier screen shorter or conditional on that action. Let users in before confirming email, ask only questions that branch, swap the tour for tooltips on the first-action screen, and keep the checklist to five items with the activation action at the top. Decide the card wall on your sales motion, not on a best-practice article. Then measure for two weeks before changing anything else: a 14-day trial is mostly decided by day 7, and the benchmarks above are vendor platform data, not your users.
We build applications in Next.js and would scope an onboarding flow as part of the product build, with the instrumentation in the first release. Our web app development service page explains how we scope it.


