All notes

SaaS Onboarding Emails Triggered by User Behavior

Build a behavior-triggered SaaS onboarding email sequence with three examples, a trigger-and-exit table, and a worksheet for preventing stale sends.

ActivationNov 12, 202617 min read

A behavior-triggered SaaS onboarding email sequence sends the next useful message based on what a user has completed and what is still missing. Send setup help when setup stalls, suggest a relevant next action after first value, and support collaboration when a teammate joins. Give every message an eligibility check and an exit condition, then check current product state again before sending so someone who has already moved on can skip it.

For B2B SaaS teams responsible for trial conversion, the practical question is: What should this person do next in this account? A day-three email cannot answer that on its own. One account may already be using the product together; another may still need permission to connect its data.

The sequence below includes three email examples and a worksheet you can use to specify your own flow. If you are reviewing several flows, start with the SaaS lifecycle email audit checklist and use this guide for the onboarding portion.

What moment should your onboarding emails move users toward?

Choose an observable product outcome before writing the emails. Account creation and integration setup may be necessary, but they do not always show that someone experienced value.

Velella is a fictional AI support copilot used throughout this Swyftlabs worked example. Its events, users, timings and emails are illustrative, not customer results. A support lead connects a help desk, reviews an AI-generated reply, and approves that reply for a real support ticket. We will use the first approved reply as Velella's candidate first-value milestone. The next collaboration milestone is a teammate approving another reply.

That distinction gives each message a job:

  • Help the workspace owner connect the help desk when setup stalls.
  • After the first approved reply, help the owner invite a colleague to review replies together.
  • When that colleague accepts, help the owner start a shared review.

Before adopting an equivalent milestone in your product, compare early completion with later retention across similar accounts. A useful association makes it a candidate to investigate; it does not prove that forcing the action causes retention. Also check that the action represents the value your customer came for.

Keep this guide focused on eligibility and transitions. For the calendar around a trial deadline, use the trial email sequence guide.

What does a behavior-triggered onboarding sequence look like?

Use a milestone table rather than a list of dates. A delay still has a role: it creates space to act before a nudge. The current milestone decides whether that nudge remains useful.

The following rules are a proposed starting design for fictional Velella. They are not timing benchmarks. Tune them to observed setup time, customer working hours and the reliability of your event pipeline.

Velella onboarding: trigger, recipient, next action and exit
MessageTrigger and eligibilityRecipient and goalSubjectSkip or exit when
1. Setup stallsSix hours after workspace_created, helpdesk_connected_at is still empty and no first reply has been approved. Recheck after any send-window delay.Workspace owner who can connect the help desk. Complete the connection.Connect your help desk to review your first replyThe connection succeeds, first value is reached, or a global suppression applies.
2. First value reachedWhen the workspace's first_reply_approved event is recorded, wait 30 minutes. Send only if teammate_invited_at and teammate_joined_at are both empty for this trial, no global suppression applies, and this message has not already been sent in the trial. Recheck after any additional wait.Owner, after the account's first approved reply. Invite one colleague to review replies.Your workspace approved its first reply. Invite a reviewer.teammate_invited_at or teammate_joined_at is populated for this trial, or a global suppression applies.
3. Teammate joinsReassess when either first value or the first teammate join is recorded. When both milestones are present for the first time in this trial, start a 30-minute delay. Before sending, confirm that this teammate is still in the workspace, has not approved a reply, and no global suppression applies. Recheck after any additional wait.Owner who invited the teammate. Start the shared review that leads to the teammate's first approval.Your teammate joined. Review the next reply together.The joined teammate approves a reply, that teammate leaves, or a global suppression applies.

Global suppressions in this example are an ended trial, a paid upgrade that hands the account to customer onboarding, an unsubscribed recipient, an undeliverable address, a deleted account, or an active human support conversation. An upgrade stops this trial sequence; it should not prevent useful onboarding from continuing in the appropriate customer flow.

These are three possible messages, not three mandatory sends. A fast account can skip all three. The first-value event cancels obsolete setup help, but it can also make the owner eligible for a different message about collaboration.

Record durable milestone timestamps on the workspace and the relevant user. An event says an action occurred; saved state lets a later step ask whether that action is already complete. For example, Loops' onboarding completion recipe sends a completion event and timestamp together, then uses the saved timestamp to filter a separate delayed reminder.

What should you send when setup stalls?

Offer one concrete next step with a direct route back into the product. If permissions are the blocker, another feature tour will not help.

Email 1: connect the help desk

Trigger: Six hours after workspace creation, the help desk is still disconnected and first value has not been reached.

Recipient: The workspace owner with connection permissions.

Exit: helpdesk_connected_at becomes populated, first_reply_approved_at becomes populated, or a global suppression applies.

Subject: Connect your help desk to review your first reply

What to avoid: A generic feature list, an invented completion-time promise, or repeated setup prompts after the connection succeeds.

The bracketed CTA in each example is a copy placeholder. Replace it with a workspace-specific product link that returns the authenticated user to the stated task. Test the destination with the recipient's permissions, including after a fresh sign-in.

If the owner replies, put the account's automated nudges on hold and route the question to a person. Resume only after that conversation is resolved and the account is still eligible. Repeated stalls across many accounts deserve a product investigation as well as better email.

What should you send after a user reaches first value?

Use the result to introduce a useful next action. A celebration alone does not tell the user how to repeat or extend the value.

For Velella, inviting a reviewer makes sense after a solo win. If your product cannot deliver value without a team, collaboration may belong earlier in its onboarding. The milestone order should follow the product's actual dependencies.

Email 2: invite a second reviewer

Trigger: When the workspace's first_reply_approved event is recorded, wait 30 minutes. Send only if teammate_invited_at and teammate_joined_at are both empty for this trial, no global suppression applies, and this message has not already been sent in the trial. Recheck after any additional wait.

Recipient: The workspace owner.

Exit: Skip or exit when teammate_invited_at or teammate_joined_at is populated for this trial, or a global suppression applies.

Subject: Your workspace approved its first reply. Invite a reviewer.

What to avoid: Repeating this invitation prompt after any invitation has been sent in the trial, or implying that an approved reply proves time saved or revenue gained.

In this example, any invitation sent during the trial completes this step, even if it later expires or is revoked. Keep teammate_invited_at as a durable milestone for that trial; revoking an invitation does not clear it. Help with an expired or pending invitation belongs in a separate, capped rule.

The 30-minute pause is an editorial choice for this example. Sending promptly keeps the result fresh; waiting can avoid duplicating an in-app confirmation. Choose a delay that leaves the next task useful, then evaluate product actions after the message. Do not treat either immediate delivery or a one-day delay as a universal rule.

Notice that the email describes a workspace result. The person receiving it may not be the person who approved the reply. This wording avoids claiming that a specific user performed an account-level action.

Who should receive an email when a teammate joins?

Separate invitation delivery, invitation acceptance and collaborative use. They are different events with different recipients.

An invitation goes to the invitee so they can accept it. The onboarding message below goes to the owner after acceptance. It helps the owner turn a new seat into a shared product outcome.

Email 3: start a shared review

Trigger: Reassess when either first value or the first teammate join is recorded. When both milestones are present for the first time in this trial, start a 30-minute delay. Before sending, confirm that this teammate is still in the workspace, has not approved a reply, and no global suppression applies. Recheck after any additional wait.

Recipient: The owner who invited the teammate.

Exit: The joined teammate approves a reply or leaves the workspace, or a global suppression applies.

Subject: Your teammate joined. Review the next reply together.

What to avoid: Sending another invitation, assuming acceptance means adoption, or triggering the same owner email for every new seat.

If the teammate joins before the first approved reply, record the join and reassess collaboration eligibility when first value arrives. Do not rely exclusively on the order in which events reach your email tool. Record the collaboration delay start once per workspace per trial; duplicate milestone events must not restart the delay or enroll another run. If the designated teammate approves a reply or leaves during any wait, suppress Email 3.

For this starting design, cap each message at one send per workspace per trial, to one designated owner. A duplicate event or a second administrator should not create another copy. Limit this flow to one onboarding nudge per workspace in 24 hours. If several messages qualify, choose the most advanced relevant milestone and recheck its conditions when the send window opens.

If an invitation remains pending, support the inviter through a separate, capped rule if needed. Do not put an unaccepted invitee into a general onboarding campaign merely because their email address appears on an invitation.

What if the user finishes while an email is queued?

Check the current completion state after waits and as close as possible to the send action. Configure completion exits as well. An entry-time check alone cannot account for something the user does later.

Consider this fictional Velella timeline:

  1. At 09:00, the setup reminder becomes eligible because the help desk is disconnected.
  2. Delivery waits for the recipient's 10:00 send window.
  3. At 09:45, the owner connects the help desk. The product records helpdesk_connected_at and passes that update to the automation system.
  4. At 10:00, the flow checks the current saved state. Because setup is complete, it skips the reminder and records the suppression reason.

The required outcome is no stale setup reminder when completion state is available before the final send decision. A message already handed to a delivery provider may be beyond your platform's cancellation boundary. Confirm that boundary rather than promising that every queued email can be recalled.

Three separate controls help:

  • Completion state and exits: In Customer.io, custom exit conditions can end a journey when a configured condition is met. Configure the relevant milestone rather than assuming that entry filters alone handle later progress.
  • A current-state check after the delay: Brew's automation documentation distinguishes saved contact fields, read when a condition runs, from the original trigger payload. Updating a field does not itself stop the flow; a later filter or branch must use it.
  • Repeat-entry controls: For event-triggered automations, Customer.io offers entry once, on every event, or once within a specified period, per person. Choose deliberately, and implement this example's workspace-level deduplication separately. Brew also documents independent runs for repeated accepted events, so a completion filter is not a substitute for duplicate-entry protection.

Some systems also support an event wait with a lookback. Hogsend documents lookback behavior for events recorded shortly before a wait begins. A finite lookback covers only that window, and it cannot find an event your system has not received. If completion should lead to a follow-on branch, check that a journey-wide exit does not abort that branch first.

The final-state check above is Swyftlabs' recommended design and acceptance criterion. The platform documentation supports individual controls; it does not establish a universal cancellation guarantee.

How do you test the queued-send check?

Use a test workspace and inspect both journey history and the received email. A branch diagram alone is insufficient.

Swyftlabs queued-send teardown worksheet
TestExpected outcomeEvidence to record
Complete the milestone before enrollmentThe obsolete message never becomes eligible.Product completion timestamp and eligibility decision.
Complete it during the initial delayThe flow exits or skips the message.State ingestion time, exit or filter result, and absence of a send.
Complete it during a send-window or frequency-cap waitThe final state check suppresses the message.Check time, state read, suppression reason and send log.
Send an invitation before first value, then revoke or expire itEmail 2 remains suppressed because teammate_invited_at is populated for this trial.Invitation timestamp, revocation or expiry state, and final eligibility decision.
Send an invitation during Email 2's delay or send-window waitThe final check suppresses Email 2, even if that invitation is later revoked.Invitation ingestion time, saved trial milestone and suppression reason.
Record join before first value, then reverse the order in a second testBoth orders start Email 3's delay once when both milestones are present. An approval or departure during the wait suppresses the message.Milestone times, delay start, membership and approval state, and final decision.
Deliver the trigger twiceAt most one intended message is sent.Event IDs, deduplication result, unchanged collaboration delay start, journey instances and send count.
Deliver completion late or out of orderOnce state arrives, stale future nudges stop. Any earlier stale send is visible for diagnosis.Occurrence time, ingestion time and actual send time.
Complete it after provider handoffThe test identifies whether cancellation is still possible; future nudges stop.Provider acceptance time and observed cancellation boundary.
Unsubscribe, upgrade, end the trial or open a support conversation during a waitThe corresponding suppression prevents subsequent nudges in this flow.Suppression state and journey decision.

Assign an owner to each failed test. Missing completion data is an instrumentation problem; a filter reading the wrong field is an automation problem. Neither is fixed by rewriting the subject line.

What should you specify before building the flow?

Copy this worksheet for each message. The Velella values above are examples; use your product's identifiers and outcomes.

Message and business owner:Workspace/account identifier:Recipient user identifier and required permissions:Qualifying product event or observed stall:Required event fields and occurrence timestamp:Durable completion state and where it is stored:One next action and direct product destination:Delay, timezone and send window:Final eligibility check after all waits:Completion exit and global suppressions:Deduplication scope and repeat-entry rule:Per-account frequency cap and message priority:Reply handling and human owner:Primary product outcome and observation window:Test evidence and known cancellation boundary:

Include workspace_id, actor_user_id, a unique event_id and occurred_at in the event contract, plus the fields needed for the particular milestone. Keep names and field meanings consistent across engineering and lifecycle teams. Use an account-aware state store or the equivalent relationship model your platform supports; do not infer that one person's inactivity means the entire account is stuck.

This example has no automatic second reminder for an unchanged milestone. If you add one, define its different purpose and a firm cap. When someone says the product is not a fit, stop the nudges and respond honestly, including explaining available pause or cancellation options. A reply should lead to help, not another automated pitch.

Treat these onboarding nudges as preference-controlled messages in this design: include an unsubscribe route and enforce suppression. Keep invitation delivery and essential account notices in their own flows so the marketing sequence does not control access to the product.

How should you measure whether the sequence helps trial conversion?

Measure the product action the message is meant to support. For Velella, track connected help desks, workspaces reaching a first approved reply, and joined teammates reaching their first approval. Then examine trial-to-paid conversion for the same account cohorts.

Define the denominator and window before comparing results. For example, the setup message's descriptive completion rate could be:

Eligible trial workspaces that connect a help desk within 48 hours of eligibility ÷ all eligible trial workspaces in that cohort.

The 48-hour window is illustrative. Choose one that fits your setup cycle and report it with the result. Start the clock at the same eligibility point for every group, and count workspaces consistently even if several users are active.

For an experiment, randomly assign eligible workspaces once per trial to the sequence or a holdout before sending, and keep that assignment fixed across users and messages. Keep essential operational notices available to both groups. Compare first-value completion and trial conversion over predefined windows, with uncertainty appropriate to the sample. Comparing people who received a setup nudge with people who skipped it because they had already finished mixes different starting states.

Use opens as a diagnostic, not the activation outcome. Apple's Mail Privacy Protection can prevent senders from learning whether a recipient opened a message, making an open an unreliable measure of product progress. Apple's Mail Privacy Protection explanation

Frequently asked questions

How many emails should a SaaS onboarding sequence contain?

Use one useful message per meaningful gap, with explicit caps. This example has three possible messages, but no account must receive all three. If an account completes the relevant tasks without help, skip the nudges.

Can behavior-triggered onboarding still use delays?

Yes. A delay gives users time to act and can respect local working hours. Check eligibility again after it. Elapsed time tells you when to look; current product state tells you what is still useful.

Does a paid upgrade mean onboarding is finished?

No. It ends the trial-conversion branch in this example. Hand the account to customer onboarding with its completed milestones intact, so the next message reflects the work still needed.

What if the product has no reliable events yet?

Start with one verified milestone and saved completion state. Test ingestion and suppression before expanding the sequence. Add monitoring for missing or delayed events, and pause the affected automation when its eligibility data cannot be trusted.

Where should we start with an existing sequence?

Choose the message most likely to become obsolete while waiting. Write its trigger, recipient, action and exit on one row, then run the queued-send tests. Use the lifecycle email audit checklist to extend the review across your other flows.

If you want a review of your current onboarding, trial and retention flows, get a free Swyftlabs email audit.

Swyftlabs TeamLifecycle email for SaaS
All notes

Find out what
your emails are missing.

Get a free, full audit of your flows, campaigns and deliverability, with what we'd fix first. Report in 7 days.

Get your free email audit
Your email audit

What we'd fix first

FlowsWelcome, trial, failed payment Checked
CampaignsProduct updates and how-tos Checked
DeliverabilitySPF, DKIM, DMARC Checked
Report in 7 days