Hi, it's Peggy.
I sign up for a lot of free trials. Partly for work, partly because a welcome sequence is the cleanest example of copy doing a measurable job. The job is getting a new user to the point where the product becomes useful to them. You can watch it succeed or fail in the data.
Last month I signed up for a project-management tool and saved all five onboarding emails. I'm not naming the company. I've changed the product details enough to keep them anonymous, and the sequence is representative of a pattern I see constantly, so treat it as a type rather than a single target. The structure, the timing, and the language patterns are intact.
The sequence has one problem, repeated five times. Every email is timed by the calendar and written about the company. Not one of them reacts to what I actually did inside the product. The single number that predicts whether I'll still be a user in thirty days never appears.
Let me walk you through it.
Your Welcome Sequence Is Talking to Itself
The Setup: Five Emails on a Calendar, Aimed at Nobody in Particular
The product is a project-management tool for small teams. Free 14-day trial, no card required. The kind of product where the trial converts only if the user gets their first project set up and invites one teammate. Call that the activation moment. Everyone in this category has one, and the entire job of onboarding is getting the user to it.
Five emails arrived on a fixed schedule: signup, day 1, day 3, day 5, day 7. I'll quote the load-bearing parts of each, lightly altered, and show what each one is doing and what the data would predict it does.
One thing to hold onto before we start. A welcome sequence has two possible spines: time since signup, or what the user has done. This one is built entirely on the first. That choice is the root of everything below.
Email 1: A Welcome That Buries the Only Thing That Matters
The first email arrives on signup. Here's the open.
Welcome to [Product]! We're thrilled to have you join thousands of teams who trust us to keep their work organized. Founded in 2019, we set out on a mission to make collaboration effortless for teams everywhere.
Then two short paragraphs on the company's values, a note about their support team, and a button near the bottom: "Explore Your Dashboard."
The problem is one of order. The reader just signed up. They have exactly one job right now, which is to create their first project. That instruction is the last thing in the email, phrased as a vague invitation, competing with company history nobody requested.
"Thrilled," "thousands of teams," "founded in 2019," "mission." Every clause is about the company. The reader appears once, as a member of "thousands of teams." A new user reads this and learns that the company is pleased with itself. They don't learn what to do next.
The fix is a single instruction, not a warmer welcome. The first email should name the one action that starts the value clock and remove everything that competes with it. The data on this is consistent: the more a first email asks the user to consider, the fewer users do the one thing that matters.
Email 2: The Feature Dump Nobody Asked For
Day 1. The subject line is "Discover everything [Product] can do."
[Product] is packed with powerful features to supercharge your workflow. Create unlimited projects, assign tasks, set due dates, track time, generate reports, integrate with 40+ apps, customize your dashboard, and so much more.
The body lists nine features as a paragraph, then nine again as bullets, each with a one-line description.
A new user one day in has set up at most one project. They cannot use nine features. They can use the next one. This email treats every reader as identical and presents the full catalog, which is the digital equivalent of handing someone a manual when they asked where the power button is.
There's a measurement point here worth naming. The user who hasn't created a project yet and the user who already invited three teammates are at completely different stages, and this email sends both of them the same nine features. A sequence that can't tell those two users apart is sequencing blind.
The verbs are the usual filler: "supercharge," "powerful," "packed." Those are cosmetic. The real failure runs deeper: the email sends a full catalog to someone who needs one next step.
Email 3: Tips for a User Who May Have Done Nothing Yet
Day 3. Subject: "3 tips to get the most out of [Product]."
Tip 1: Use labels to organize your tasks by priority. Tip 2: Set up recurring tasks to save time. Tip 3: Try our mobile app so you can manage projects on the go.
These are fine tips. They assume a user who has already created a project, added tasks, and is now optimizing. If I signed up on Monday and never came back, this email arrives Thursday telling me to label tasks I never created.
Here is the structural issue laid against the calendar.
Day | Email sent | What it assumes the user has done | What the user may have actually done |
|---|---|---|---|
0 | Welcome | Nothing yet | Signed up |
1 | All nine features | Ready to explore broadly | Maybe created one project |
3 | Optimization tips | Created projects and tasks | Possibly nothing since signup |
5 | Customer results | Engaged enough to care | Unknown |
7 | Upgrade now | Seen the value | Unknown |
Every row in the third column is a guess. The product knows the real answer for every user. None of that knowledge reached the email.
Behavioral state, not elapsed days, is what should decide which email sends. A user who hit the activation moment needs a different day 3 than a user who stalled at signup. This sequence sends them the same one.
Email 4: Proof With Nothing Specific to Prove
Day 5. A customer story.
Teams like yours have seen incredible results with [Product]. One company boosted productivity and saved hours every week after switching. Ready to see what you can achieve?
This is proof in the shape of proof, carrying no actual evidence. "Teams like yours," "incredible results," "boosted productivity," "saved hours." There is no number, no named company, no specific before-and-after a reader could picture.
Compare it to what a usable proof point sounds like: a named role at a named company, the situation before, and one concrete result. "A six-person design studio cut their status-update meetings from three a week to one after moving their projects in." That sentence has edges. The day-5 email has none.
I'll be careful here, the way I always am with claims. I don't know whether this company has real customer results. If they do, this email is wasting them on vague phrasing. If they don't, the fix is upstream of copy: go get one real result before you send a proof email at all.
Email 5: The First Ask, With No Value Established
Day 7. Subject: "Your trial ends soon, upgrade now!"
Don't lose access to your projects! Upgrade today to keep your team organized and unlock all premium features. Choose a plan that's right for you.
This is the first email in the sequence that asks for anything clearly, and it asks for money. The urgency is real, but it's the company's urgency. The trial is ending on the company's calendar, so the company would like the sale.
Nothing in the email references what this specific user built, accomplished, or experienced over seven days. For a user who hit the activation moment, the email could say "your three projects and your two teammates go read-only on Friday." For a user who stalled, an upgrade ask is premature and a re-activation nudge would do more. The sequence sends both the same generic close.
The CTA placement across the whole sequence tells the story. The only clear call to action in seven days of email is the one that asks for a credit card. The four emails that could have driven the user toward value buried their instructions or omitted them.
What the Sequence Should Have Done
The root cause is a single decision. The sequence was built on a calendar, and a calendar can't see the user.
Every product in this category logs whether a user has created a project, added a task, and invited a teammate. That data is the activation moment, and it's the only thing that should decide what the next email says. A sequence wired to behavior sends one path to the user who activated and a different path to the user who stalled. A sequence wired to dates sends everyone the same five emails and hopes.
Here's the same sequence rebuilt on triggers instead of days. The calendar version from the Email 3 table sent five fixed emails to everyone. The behavior version sends whichever email matches where the user actually is.
Trigger (what the user has done) | Email's one goal | Single CTA |
|---|---|---|
Signed up, no project yet | Get them to create the first project | Create your first project |
Created a project, no teammate | Get them to invite one person | Invite a teammate |
Hit activation (project + teammate) | Reflect the value they just unlocked, point to the next useful feature | Try [the next relevant feature] |
48 hours after signup, still no project | Re-activate by removing the friction that stopped them | Create your first project |
Activated, trial ending | Upgrade, naming what they built | Keep your 3 projects active |
Same fourteen days, but every email now references where the user is, and the upgrade ask only reaches users who have something worth keeping.
That's the whole teardown: the emails are timed by the calendar and written about the company, when they should be triggered by behavior and written about the user's progress toward the one moment that predicts retention.
You can fix this at two levels. First, a diagnostic you run on a sequence you already have. Second, a prompt that rebuilds it.
Run This Diagnostic on Your Own Welcome Sequence
Pull up your current sequence and answer these. Each question targets one of the failures above.
What is the single activation moment for this product, the thing a user must do before the product becomes useful? If you can't name it in one sentence, the sequence has no target.
Does email one drive toward that moment and nothing else? Count the competing asks. The answer should be zero.
For each email, what does it assume the user has already done? Write the assumption down. Then ask whether the product knows if that's true.
Is any email in the sequence triggered by behavior, or are all of them timed by the calendar? If all five are time-based, that's the headline finding.
Does any email reference what this specific user built or did? Or could every email be sent to a user who never logged in after signup?
Where is the first clear call to action, and what does it ask for? If the first unambiguous ask is the upgrade, the sequence skipped the job of getting the user to value first.
Does the proof email contain one specific, verifiable result? A named role, a real number, a concrete before-and-after. If it's "teams like yours see great results," it isn't proof.
Most sequences I run this on fail questions 4, 5, and 6 together, because they're the same failure viewed from three angles. The sequence can't see the user, so it can't speak to the user, so it defaults to speaking for the company.
You don't have to grade it by hand. Paste your sequence into this prompt and let Claude run the seven questions for you.
You are auditing a free-trial welcome email sequence. Below it, I'll paste every email in the sequence in order, with the send timing for each.
First, ask me one question if it isn't already clear from what I paste: what is the product's single activation moment (the action a user must complete before the product becomes useful)? Don't audit until you know it.
Then grade the sequence against these seven checks. For each one, give a PASS or FAIL, one sentence of evidence quoted from the emails, and the fix.
1. Activation target: Is there a single, nameable activation moment the sequence is driving toward?
2. First email focus: Does email one drive toward that moment with exactly one call to action, nothing competing?
3. Honest assumptions: For each email, what does it assume the user has already done, and could the product actually confirm that?
4. Trigger type: Is any email triggered by user behavior, or are all of them on a fixed time delay? Name which.
5. User reference: Does any email reference what this specific user built or did, or could every email be sent to someone who never logged in after signup?
6. First clear CTA: Where does the first unambiguous call to action appear, and what does it ask for? Flag it if the first clear ask is the upgrade.
7. Proof: Does the proof email contain one specific, verifiable result (named role, real number, concrete before-and-after)?
End with the single biggest structural problem and the one change that would move the most users to activation. Do not rewrite the emails yet.
[paste the full sequence, with timing, here]
The instruction to withhold the activation moment until you supply it is the important one. Without that, the model invents a plausible activation moment and grades the sequence against its own guess. The audit is only as good as that one input.
The Prompt That Rebuilds It Around Behavior
This prompt won't write a finished sequence on its own, and it shouldn't. It forces the structural decisions the original skipped, then drafts against them. Expect to run it once, answer its questions, and run it again.
You are helping me rebuild a welcome/onboarding email sequence for a free trial. Do not draft any emails yet.
First, establish the structure with me:
1. Ask me what the product is and who the user is. Wait for my answer.
2. Ask me to name the single activation moment: the specific action a user must complete before the product becomes useful. If my answer is vague, push me to make it concrete and measurable.
3. Ask me what behavioral data the product can track (e.g., created a project, invited a teammate, completed a task). I will tell you what's available.
Then propose a branched structure, not a calendar:
- A path for users who have hit the activation moment.
- A path for users who have NOT hit it yet.
- For each email, state its trigger (a behavior or a time-since-last-action), its single goal, and its one call to action.
Rules for the structure:
- Email one drives toward the activation moment and carries exactly one call to action. Nothing competes with it.
- No email assumes the user has done something the data can't confirm.
- The upgrade ask appears only after the user has reached the activation moment, and it references what they actually built.
- Every email is about the user's progress, not the company's history, mission, or feature list.
Show me the structure and wait for my approval before drafting a word of copy.
When I approve, draft each email. Ban this language: thrilled, excited, supercharge, powerful, effortless, seamless, unlock, packed with, and any sentence that opens with the company's name or founding story. For the proof email, leave a clearly marked placeholder for one real, specific customer result and tell me exactly what you need from me to fill it.
Why it's built this way. Three deliberate choices:
It refuses to draft until the activation moment and the available behavioral data are named. Those two facts are exactly what the original sequence never established.
It branches into activated and stalled users. That's the structural change a calendar-based sequence can't make.
It leaves a placeholder for a real customer result. Inventing a customer outcome is the one thing a model must never do.
One honest limitation. This rebuilds the copy and the logic, but the branching only works if your email platform can trigger on product events.
If your tool only supports time delays, you can still apply questions 1, 2, 5, and 6, which improve a time-based sequence considerably. The behavioral branching is the bigger lever, and it needs the plumbing to match.
Want to get the most out of ChatGPT?
ChatGPT is a superpower if you know how to use it correctly.
Discover how HubSpot's guide to AI can elevate both your productivity and creativity to get more things done.
Learn to automate tasks, enhance decision-making, and foster innovation with the power of AI.
The Burnett Matrix
A welcome sequence is the most measurable copy you own. The user's behavior is right there in the product, and the conversion is a clean event.
The sequence I tore down threw that measurement away, timing every send on a calendar and writing every line about the company. The fix lives in the wiring, not the sentences: trigger the sends on the activation moment, and write each email about the user's progress toward it.
Run the seven-question diagnostic on your own sequence this week. Question four is usually where it breaks.
More clicks, cash, and clients,
Peggy Burnett
