SaaS User Onboarding

The first session decides most of what happens next. A user who reaches something useful in their first ten minutes comes back on day two, and a user who does not usually never opens the tab again. That is the whole argument for spending engineering time here rather than on the next feature.

Start with a list, not a design. Write down every single thing a new user must do between clicking signup and getting value, including the invisible steps like waiting for a sync or finding an API key.

Audit the setup work, then delete most of it

Most onboarding projects begin by making the existing steps prettier. The higher-return move is removing steps entirely, and the audit is what makes that decision possible.

For each step on your list, ask three questions. Can we infer this instead of asking? Can this wait until after the first success? Can we do it for the user?

Setup step	Usual treatment	Better treatment	Effect on time to value
Company size and role	Required form field	Enrich from the email domain, confirm later	Removes a screen
Data import	Ask the user to upload a CSV	Offer three sample datasets plus an import option	User sees value before doing work
Invite teammates	Step two of onboarding	Prompt after the first successful action	Large. Invites before value feel like a favour.
Connect an integration	Blocking step	Optional, with a clear reason attached	Removes the biggest drop-off point
Choose a plan	Before entering the product	After the trial produces something	Removes a decision the user cannot yet make
Profile photo and preferences	Onboarding checklist item	Delete entirely	Small but free

The teammate invite step deserves its own paragraph. Asking someone to invite three colleagues before they have seen anything works is asking them to stake their reputation on a product they have not evaluated. Move it after first value and invite rates typically rise rather than fall.

The three onboarding patterns and which product needs which

There are three, and picking the wrong one is the most common structural mistake.

The guided tour walks a user through the interface with tooltips. It fits products where the value is in a workflow the user cannot guess at, and it fails when users click through it to reach what they came for. Slack’s original Slackbot conversation is the good version of this: it taught by making you do the thing rather than by pointing at it.

This checklist plus designed empty states fits most B2B tools. Each empty screen explains what goes there and offers the one action that fills it. Linear does this well, with a first run that assumes you will use the keyboard and teaches the shortcut in the moment you need it rather than in a modal at the start.

The done-for-you import or template fits products where the blank page is the problem. Notion’s template picker and Canva’s design-first empty state both skip the blank canvas entirely, so the user starts with something to edit rather than something to create. Figma does the same by making the file, not the tour, the first thing you meet.

Pattern	Fits	Fails when	Build cost
Guided tour	Complex workflows with non-obvious sequence	Users want one specific thing and skip the tour	Low with a tool, medium in-house
Checklist plus empty states	Most B2B tools with several entry points	Checklist items are invented to fill it	Medium, needs design work per screen
Done-for-you import or template	Products where a blank canvas blocks the user	Templates do not resemble the user's real work	High, needs content as well as code

Empty states are the real onboarding surface

An empty state is a screen a user will see when they are most confused and least committed. Treat it as the highest-priority design work in the product.

Four things belong on every empty state: a one-line description of what lives here, the single most likely next action as a button, a secondary path for people who want the other option, and an example. The example does more work than the copy. Showing a populated version of the screen, even as a static image, answers the question the user actually has, which is what this is supposed to look like when it works.

This is why I would fix empty states before buying a tour tool. Tour tools run 500 to 3,000 dollars a month and their most common use is explaining a screen that should have explained itself. Spend the first sprint on the screen.

Design for the second session

Almost every onboarding review I have seen optimises session one and ignores session two, where a large share of the loss actually happens. The user came back, the initial context is gone, and the product shows them the same screen with nothing obviously next.

Three things carry a user into the second session. Leave a visible unfinished thread, such as a draft, a partial import, or a report that finished processing. Send an event-triggered message when something completed while they were away. And make the second session start where the first one ended rather than at a dashboard.

Measure it explicitly. Day one to day two return rate is a better early predictor of retention than almost anything else you can read in week one, and most teams do not have it on a chart.

The onboarding email sequence, triggered on events

Day-based sequences send the wrong message to most people. Day three’s “here is how to import your data” arrives for a user who imported on day one and for a user who has not logged in since signup, and it is wrong for both.

Wiring this needs product events flowing into your messaging tool, usually through Segment or a direct integration with Amplitude or Mixpanel. That is roughly a week of engineering and it is the highest-return week in most onboarding projects. The full sequence with copy is in the SaaS onboarding email sequence, and the drafts are in SaaS onboarding email templates.

Measuring: one activation event, three funnel steps

Pick one activation event and defend it. It should be the action that correlates most strongly with 90 day retention in your own data, not the action that is easiest to move.

Find it by comparing retained and churned cohorts across candidate events. In most products, the winning event involves the user putting their own data in and getting something back. Profile completion never wins. Neither does “viewed the dashboard”.

That last row is the honest one. If users activate, return once, and then stop, onboarding is not your problem and no amount of checklist design will fix it. Say so out loud rather than running a fourth onboarding redesign.

The tradeoff nobody mentions

Aggressive onboarding optimisation can raise activation and lower revenue quality. Remove enough friction and you activate people who were never going to pay, which inflates your activation rate, fills your product with low-intent accounts, and makes your paid conversion rate look worse.

Watch activation rate and trial-to-paid together. If activation rises 12 points and trial-to-paid falls 4, you have probably optimised for the wrong population. The fix is usually qualification at signup rather than more friction later, which is covered properly in how to design a SaaS free trial.

Where to start on Monday

Run the setup audit. One spreadsheet, one row per step, three columns for infer, defer or automate. Most teams find two or three steps they can delete outright in an hour of discussion, and deletions ship faster than redesigns.

Then pick your single activation event, instrument the four funnel steps, and only after that decide whether you need a tour tool. Use the SaaS onboarding audit checklist to structure the review, SaaS onboarding teardowns to see how other products solved the same screens, and the time to value definition to keep the team using the term consistently.

Once activation is stable, the next two levers are a focused 30 day activation sprint and, if your product has any multiplayer dimension, designing a SaaS referral loop. Both sit downstream of a first session that works, which is why this comes first. The wider programme context is in SaaS growth marketing.