What onboarding email is for
A new user of a software product signed up for an outcome, not for email. Most of the work of getting them there happens inside the product: the set-up screens, the empty states, the in-app guidance. Email does the part the product cannot, which is reaching the user when they are not in it, with the one thing that would bring them back to the step they left unfinished. In the AARRR framework, that is the activation stage: the first moment a customer receives the value they came for, set at the action after which people are markedly more likely to stay.
Onboarding sits beside three neighbours that are easy to confuse with it.
| Sequence | Who it is for | Trigger | Goal | Where it is covered |
|---|---|---|---|---|
| Welcome series | A new subscriber to a list | The sign-up to the list | The relationship and a first purchase or use | Welcome series tutorial |
| Lead nurture | A lead who has not bought or started a trial | Stage and behaviour signals | Progress towards a demo, trial or sale | Nurture sequences guide |
| In-app and push onboarding | A user inside, or just outside, an app session | Screens reached and steps not completed | The next step, at the moment of use | Push and in-app messaging guide |
| Onboarding email | A new user of the product, trial or paid | What the user has not yet done | Activation, then a decision to pay | This tutorial |
The measure is activation within an agreed window from sign-up and, for a trial, conversion to paid at a fixed horizon, read by sign-up cohort. Opens and clicks explain how the emails behave; they are not the result. When the product is an app, the in-app side of the sequence, from the path to first value to the moment each permission is requested, is designed in the guide to app onboarding and permission requests. For a service or a physical product, where first value is a visit, an installation or a first delivery, the tutorial on onboarding new customers beyond software designs the path instead.
Before you start
- Product data you can query: account creation, each set-up step, use of the core feature and, for a trial, its start and end dates and plan, all keyed to a user or account identifier.
- A way to send product events to the email platform as they happen, or close to it, with the same identifier the platform uses. The data flow and integrations guide explains the routes; engineering time for this belongs in the plan.
- The in-app and push messages already running, with their triggers, so that email can be fitted around them.
- A consent position for the emails that will be treated as marketing, decided with whoever owns compliance.
- A working welcome series pattern. The welcome series tutorial sets out trigger, spacing, exits and tests; this tutorial reuses them for product events rather than a list sign-up.
- A start for customers who buy through sales. Where the account begins with a deal won in the CRM rather than a sign-up, the tutorial on building a closed-won workflow starts onboarding for every contact on the deal and takes them out of prospect sequences first.
- Someone who can talk to users. The reasons people stall rarely appear in event data on their own.
- An owner and a review date for the sequence.
The steps
Define activation from product data
Write activation as one action, observed in product data, that marks the first moment a user has received the product's core value: a first project created, a first report shared, a first document signed. The AARRR framework's test is the one to use: choose the action after which retention is markedly higher, by comparing users who took it with users who did not, not the action that is easiest to count. Add a window from sign-up within which it counts, set from how long users who stay usually take. The onboarding cadence test defines activation the same way, and the definition should be shared by the sequence, the test and the product team's dashboard. Write it at the top of the sequence's notes, because every email exists to move a user one step closer to it.
Map the steps between sign-up and activation
List the steps a new user has to complete to reach activation, in the order the product presents them: verify the account, connect a data source, invite a colleague, configure the first object, use the core feature. Then read product data for the share of new users who complete each step and how long they take, and mark the steps where most of them stop. Event data shows where users stall but rarely why. A handful of conversations with recent sign-ups who stalled, run as the customer interviews tutorial describes around what they actually did, usually shows whether the obstacle is understanding, effort, missing data or a colleague who never accepted the invitation. Each answer suggests a different email.
Separate service messages from marketing
Classify each planned email before it is written. The promotional versus transactional comparison describes the UK regulator's direct marketing guidance as separating service messages a customer needs from marketing a business wants to send, decided by the content and purpose of the message rather than its label. Account verification, password and security messages are service messages. Help with completing set-up of an account the user has just opened is often service content too. An upgrade offer, a promotion or a push towards a higher plan is marketing, and so is any email that mixes the two. Marketing emails need a recorded basis for each user and an unsubscribe in every message, and an unsubscribe from marketing must never stop the service messages. Decide each case with whoever owns compliance and treat unclear ones as marketing; the SaaS industry guide sets out the wider constraints on commercial email for software. Nothing here is legal advice.
Send the product events the flow needs
The sequence can only react to what the email platform knows. For every step in the map, send an event when the user completes it, with the user identifier, the account, the plan and, for a trial, the trial end date as an attribute. Send the activation event itself, and cancellation and plan changes. Check the delay between an action in the product and its arrival in the email platform, because a sequence that learns about set-up an hour late will send the set-up reminder to someone who has just finished. Keep the user's lifecycle stageA CRM field recording where a contact stands in the relationship, such as subscriber, lead, opportunity or customer., such as trial, active, paying or cancelled, as one attribute defined once and shared with reporting.
Write one email per stalled step
For each step where users stall, write one email that does one job: get the user through that step. Say what the step gives them in terms of the outcome they signed up for, show how to complete it in as few moves as possible, and link straight to the screen where it happens, signed in where the product allows. Answer the obstacle the interviews found: a short guide for understanding, a template for effort, sample data for missing data, a resend button for an invitation nobody accepted. Offer a reply address or a way to book help for the users who are genuinely stuck. Steps where few users stall need no email. Keep the brand's newsletter template, but resist adding product news or offers to an email whose job is a single step.
Trigger each email from what the user has not done
Trigger each email when a user has not completed its step after a wait that gives them a fair chance to do it unprompted, and only if they have completed the steps before it. A user who completes the step never receives its email; a user who completes several steps in one session skips straight past them. This is the behaviour-triggered design the onboarding cadence test compares with a fixed schedule; this tutorial builds it, and the test is how to prove it beats the schedule for your product. Set the waits from the step map, not from a template, and set a sending window in the user's local time. Cap the sequence so that a user stuck at one step receives its email once, with at most one follow-up, rather than a reminder on every day of the trial.
Coordinate with in-app messages
An email and an in-app messageA message shown inside an app during a session the person started, such as a banner, modal or tooltip, needing no operating-system permission. about the same step, arriving the same day, tell the user the product is not paying attention. The push and in-app messaging guide gives the division of labour: in-app messages guide users inside a session, push brings people back, and in an app the first steps usually belong in-app, with push reserved for people who leave before finishing. Email follows the same logic. Let in-app guidance handle a step while the user is in the product, and send the email only if the user leaves without completing it. Put every channel's onboarding messages in one table with their triggers, and apply one frequency cap per user across email, push and in-app. For an app, the tutorial on onboarding messages for new app users builds the in-app and push half of that table.
Handle the end of a trial plainly
For a trial, users need to know when it ends and what happens next: whether the account becomes paid, moves to a free plan or stops, what they will lose access to, and how to choose. Send that information as a plain message ahead of the end date and again at the end, from the trial end attribute rather than a day count, so that a change in trial length does not leave the message on the wrong day; the free trial length test explains why day-numbered schedules break when trial length changes. Any statement about renewal, charges or cancellation must match the terms and be checked by whoever owns compliance; the sources this site cites do not cover notice rules for renewals or charges. Users who have not activated by the end of the trial are the ones to offer help or more time, not a discount.
Set exits on activation, cancellation and unsubscribe
Activation ends the onboarding sequence: the user has reached the goal, and the lifecycle messages for active users take over. Cancellation ends it, with at most a single message that confirms the cancellation and asks why, and never an onboarding nudge to someone who has left. An unsubscribe from marketing ends the marketing emails and leaves the service messages. Conversion to paid ends the trial-end messages but not the step emails, because a paying user who has not activated is the user most likely to become part of next month's churn rateShare of customers at the start of a period who left during it.. Check every exit before each send, not only at entry, as the automation hub requires of any flow, and run the email automation flow launch checklist before the sequence goes live.
Measure activation against a control
Measure the sequence on activation within the agreed window and, for a trial, on paid conversions per sign-up at a fixed horizon, the measure the free trial length test uses because trial-to-paid conversionShare of free trials that became paying subscriptions. read at each trial's own end changes meaning when anything else changes. Read both by sign-up cohortA group of customers who share a starting event and period, tracked together over time., as the cohort analysis tutorial builds it, and only after every user in a cohort has completed the window. Compare against a control: the old schedule while a new sequence is being proved, randomised at sign-up as the onboarding cadence test describes, or a small holdout groupA randomly chosen part of an audience kept from exposure to measure what exposure caused. that receives only the service messages, to show what the sequence adds at all. Keep unsubscribes, complaints and emails per user as guardrails, and review the step map whenever the product's onboarding changes.
A staff rota app's onboarding sequence (illustrative)
Rotafinch replaces a fixed schedule with three triggered emails
Rotafinch, a staff rota app for small cafes, offers a 14-day trial and sent five onboarding emails on days 1, 2, 4, 7 and 12 to every trial user. It defined activation as publishing a first rota within 14 days of sign-up. The step map showed that most trials which never activated had stopped at adding staff: 61% of non-activated accounts had added no one but the owner. It replaced the schedule with three triggered emails, each sent only to users who had not completed its step: add your staff (with a spreadsheet import template), build a first week's shifts, and publish and share the rota. New trials were randomised at sign-up between the old schedule and the new sequence, and paid conversions were read at 30 days from sign-up.
| Sign-up cohort | Sequence | Trials | Activated within 14 days | Activation rate | Paying at 30 days | Paid per sign-up |
|---|---|---|---|---|---|---|
| March | Fixed schedule | 310 | 112 | 36% | 50 | 16% |
| March | Triggered | 305 | 140 | 46% | 62 | 20% |
| April | Fixed schedule | 290 | 101 | 35% | 44 | 15% |
| April | Triggered | 296 | 133 | 45% | 57 | 19% |
| May | Fixed schedule | 300 | 111 | 37% | 48 | 16% |
| May | Triggered | 299 | 141 | 47% | 63 | 21% |
Across the three cohorts the triggered sequence activated 414 of 900 trials against 324 of 900, and 182 were paying at 30 days against 142. Emails per trial user fell from five to an average of 1.7, and unsubscribes fell by about a third. Rotafinch kept the triggered sequence and a small holdout that receives only account messages.
If trial users still rarely pay once the sequence is triggered by behaviour, the diagnosis for trials that do not convert to paid checks the sign-up, sales follow-up, pricing page and cohort reading alongside the emails.
Common mistakes
- Activation chosen because it is easy to count. Signing in is not value; pick the action that predicts staying.
- A fixed schedule for everyone. Users who finished a step are told to do it; users who are stuck wait for days.
- Several jobs in one email. A set-up nudge, a feature tour and an upgrade offer together achieve none of them.
- Events that arrive late. The reminder lands an hour after the user completed the step.
- Email and in-app saying the same thing on the same day. One table and one cap across channels prevents it.
- Trial-end messages timed by day number. A change in trial length leaves them on the wrong day.
- Stopping at payment instead of activation. Paying users who never reached value are the next cancellations.
- Judging the sequence on opens. Activation and paid conversion by cohort are the result; opens are not.
When the sequence is live, the email metrics and testing guide explains how to read its holdout and how much volume a fair comparison of two versions needs.
