In early 2015, Stewart Butterfield, the co-founder of Slack, told First Round Review about a number his team had noticed. They looked back at which teams had stuck with Slack and which hadn’t, and picked a line: any team that had exchanged 2,000 messages had really tried the product. Then came the striking part. Of the teams past that line, he said, 93% were still using Slack.
For a team of about 10 people, he added, 2,000 messages was roughly a week of normal chat.
That story gets repeated a lot, usually as “get users to 2,000 messages and they’ll stay.” That is not quite what it says, and the difference is the subject of this post. What Slack found was a line between people who had actually experienced the product and people who had only signed up. Everything you do in the first minutes and days of a new user’s life is about moving them across that line.
That line has a name: activation.
What activation means
Activation is the moment a new user first gets the value your product promised. Not the moment they sign up. Not the moment they log in. The moment the product does its job for them, once.
Some examples:
| Product | Signed up | Activated |
|---|---|---|
| Invoicing app | Created an account | Sent a first real invoice to a client |
| Team chat | Created a workspace | The team actually talks there instead of email |
| Scheduling tool | Connected a calendar | Someone booked a meeting through their link |
| Analytics tool | Installed the snippet | Saw their own data in a chart and answered a question |
| An API | Got a key | Made a call and got a useful result back |
For my own small CV scoring API, the first value is simple to name: a developer sends one CV and one job description and gets a fit score back. Getting a key is not that. Reading the docs is not that. Only the first real result is.
An activation rate is the share of new signups who reach that moment, usually within a set window like 7 or 14 days.
Activation rate = signups who reached first value within N days / all signups in that period
Why care so much about one early moment? Because people who never get value almost never come back. You cannot retain someone who never started. In Cohort Retention: The Chart That Tells the Truth, the first big drop in a retention curve, from month 0 to month 1, is mostly activation failure. People signed up, poked around, didn’t get it, and left.
A useful way to see the early life of a user is as four steps:
- Signup. They create an account.
- Setup. They do the work the product needs before it can help: import data, connect a tool, invite people.
- First value. The product does its job once. This is activation.
- Habit. They come back and get value again, on their own, week after week.
Each step loses people. The rest of this post is about losing fewer of them.
The “aha moment”, and how to read the famous ones
The aha moment is the informal name for activation from the user’s side: the point where they think “oh, this is useful.” Founders love the idea because it suggests one magic action that turns visitors into loyal users.
Two examples come up in almost every talk on the subject.
Facebook’s “7 friends in 10 days.” Chamath Palihapitiya led Facebook’s growth team in its fast-growth years. In a talk published in October 2012, he described the team’s big realization as getting any new user to seven friends within ten days of joining. Accounts of the talk say the team got there by comparing users who stayed engaged with those who didn’t, and noticing the friend count in the first days. He also framed the job as getting people to an aha moment as fast as possible.
Slack’s 2,000 messages, from the story above.
Both are real, sourced claims. Both are also routinely stretched. Here is how to state them carefully:
- They are correlations, not proven causes. Teams that sent 2,000 messages stayed. That might be because messaging made them stay. It might also be because teams that already needed Slack badly were the ones that sent 2,000 messages. Facebook users with seven friends early on stayed. But people whose friends were already on Facebook would both reach seven friends quickly and have a reason to stay. The number marks the line. It does not prove that pushing someone over it will keep them.
- They come from a specific product at a specific time. Seven friends meant something for a social network at that time. It means nothing for your invoicing app.
- The exact figures were reported in talks and interviews, not published studies. We don’t have their data, their cohort definitions or how they picked the cut-off. Treat the numbers as anecdotes from people who were there.
That doesn’t make them useless. What they teach is a method: look at what retained users did early that churned users didn’t, find the simplest action that separates them, then design the product to get more people to that action. Then check whether retention actually moves. If it does, you may have found a cause. If it doesn’t, you found a symptom.
Copy the method, not the number.
Find your own activation event
You don’t need a data team for this. You need two lists: users who stayed, and users who left. Then you compare what each group did in their first few days.
Here is a worked example with a fictional company, Ledgerleaf, a simple bookkeeping and invoicing app for freelancers. Everything about Ledgerleaf is made up for this post, and all the numbers are checked in Python.
Over three months, 1,200 people signed up. Four weeks after signing up, 130 of them were still using the product (retained). The other 1,070 had stopped (churned).
The founder picks four things a new user might do in their first days and counts how many in each group did them:
| Early action | Retained who did it (of 130) | Churned who did it (of 1,070) |
|---|---|---|
| Connected a bank account | 98 (75%) | 610 (57%) |
| Sent a first invoice within 3 days | 117 (90%) | 123 (11%) |
| Invited their accountant | 26 (20%) | 11 (1%) |
| Watched the product tour | 52 (40%) | 470 (44%) |
Now flip it around. For each action, what share of the people who did it were still around four weeks later, compared with people who didn’t?
| Early action | People who did it | Retained if they did it | Retained if they didn’t |
|---|---|---|---|
| Connected a bank account | 708 | 13.8% | 6.5% |
| Sent a first invoice within 3 days | 240 | 48.8% | 1.4% |
| Invited their accountant | 37 | 70.3% | 8.9% |
| Watched the product tour | 522 | 10.0% | 11.5% |
Read the rows one at a time:
- Connecting a bank account looks good at first. Three quarters of retained users did it. But more than half of churned users did it too. It’s a common setup step, not the moment of value. Weak signal.
- Sending a first invoice within 3 days separates the groups sharply. 90% of retained users did it; only 11% of churned users did. People who did it stayed at 48.8%. People who didn’t stayed at 1.4%. Strong signal, and it’s common enough (240 of 1,200 people) to build around.
- Inviting an accountant has the highest retention of all, 70.3%. But only 37 people did it. It probably marks a small group of very serious users. Interesting, but too rare to be the main target.
- Watching the tour doesn’t help at all. People who watched it were slightly less likely to stay. More on that below.
So Ledgerleaf’s activation event becomes: sent a first real invoice within 3 days of signing up.
A good activation event separates retained from churned users clearly, is common enough to matter, is the product doing its job (not a setup chore), happens within days, and fits in one sentence.
And a warning: with small numbers, these gaps can be noise. If you have 30 signups, you have a guess, not a finding. Write down the candidate, then check it again with the next few cohorts. And remember the correlation trap from above. The real test is whether getting more people to the event raises retention. That’s an experiment, not a spreadsheet.
Time to value
Once you know the activation event, measure time to value: how long it takes a new user to get from signup to that event.
Time to value = time of first value - time of signup
Use the median (the middle user), not the average, because a few people who come back after three weeks will drag the average up and hide what’s typical.
Time to value matters because attention runs out. Someone who just signed up is at their most curious. Every minute of setup, every form, every “check your email to continue” spends that curiosity. If first value is two days away, most people will be gone before it arrives, not because they decided your product is bad, but because life got in the way.
The title of this post is a goal, not a rule. For many simple tools, the first value can happen in the first five minutes. For heavier products, like accounting software that needs real data, it might take a day. The question is always the same: what is the shortest path from “I signed up” to “this did something for me”, and what is standing in it?
To find what’s standing in it, walk through your own signup as a stranger, with a timer. Better, watch three real new users do it, over a video call, without helping. Write down every step, every pause, every moment they look confused. You will be surprised how long it takes. Founders know their product too well to see the friction.
Onboarding: remove steps, don’t add tours
Onboarding is everything between signup and first value. There are two very different ways to “improve” it.
One is to add help. A welcome video. A product tour with seven tooltips. A checklist of things to do. An email series explaining features.
The other is to remove work. Fewer fields. Fewer required steps. Smarter defaults. Doing for the user what they would otherwise do by hand.
The first is easier to build and feels productive. The second almost always works better, because a tour explains the steps. Removing a step means nobody has to do it.
Look back at Ledgerleaf’s table. The tour was watched by 522 people and didn’t help retention at all. That’s fictional, but it’s a pattern worth checking in your own data. A tour asks a new user to learn your interface before they’ve seen why it’s worth learning. Most people click “skip,” or click “next” seven times without reading, and arrive at the same empty screen.
Ways to remove steps:
- Ask for less at signup. Email and password, or a single sign-in button. Company size, role and “how did you hear about us” can wait, or be skipped forever.
- Delay setup that isn’t needed for first value. Ledgerleaf needs a client name and an amount to send an invoice. It does not need a bank connection, a logo or tax settings first. Let people add those after they’ve seen it work.
- Fill in what you can guess. Currency from their country. Time zone from their browser. A sensible default template.
- Let them do the real thing right away. If your product sends invoices, the first screen after signup can be “send your first invoice,” not a dashboard.
- Make the first result fast. If your product needs data to be useful, import it for them, or let them start with a small piece of it.
One honest test for every onboarding step: if we deleted this, would fewer people reach first value? If the answer is no, delete it or move it later.
None of this means help is bad. A one-line hint next to a confusing field beats a five-minute video nobody watches. Help with steps that must exist; don’t explain steps that shouldn’t.
Empty states and sample data
An empty state is what a screen looks like when there’s nothing in it yet: no projects, no invoices, no data. For a new user, almost every screen is an empty state. It’s also where many of them quit.
A bad empty state says “No invoices yet” in grey text in the middle of the page. It’s accurate, and it gives the user nothing to do.
A good empty state does three things:
- Says what will appear here. “Your invoices will show up here, with who’s paid and who hasn’t.”
- Gives one clear next action. A single button: “Create your first invoice.”
- Shows what it looks like when it’s full. A small preview or illustration of a filled-in screen, so the user can picture the value.
Sample data goes one step further. Instead of an empty screen, the new user sees a realistic example already filled in: a demo project, a few made-up invoices, a chart with example numbers. Now they can click around and see how the product behaves before putting in their own data.
It works best when it’s clearly labelled as an example, easy to clear in one click, and leads to the real action: “This is a sample invoice. Send your own in 30 seconds.”
Be careful with one thing: sample data can create a fake aha moment. Someone sees a nice chart of made-up numbers and feels good, but nothing has happened for them yet. Count activation on real use, not on time spent looking at the demo.
A worked example: buy traffic, or fix activation?
Here is the decision many founders face, with Ledgerleaf again. Every number is fictional and checked in Python.
Ledgerleaf today, per month:
- 10,000 people visit the website.
- 4% sign up: 400 signups.
- Half of them finish setup: 200.
- 40% of those send a first invoice: 80 activated. That’s an activation rate of 20% (80 of 400).
- Half of activated users are still active at week 4: 40 retained.
- For simplicity, assume people who never send an invoice don’t stay. The earlier table showed only 1.4% of them did, so this barely changes the answer.
The founder has two options for next month.
Option A: buy more traffic. Spend $12,000 on ads for 10,000 extra visitors ($1.20 a visit). Ad visitors are a bit less interested than people who found Ledgerleaf on their own, so they sign up at 3%: 300 extra signups. They go through the same leaky funnel: 150 finish setup, 60 send an invoice, 30 extra retained users.
That’s $400 per extra retained user ($12,000 / 30), and the $12,000 has to be spent again next month to get the next 30.
Option B: fix activation. Spend a few weeks on the onboarding work from the sections above: cut setup to the minimum needed to send an invoice, make “send your first invoice” the first screen, add a sample invoice and a better empty state. Suppose this raises setup completion from 50% to 80% and first-invoice rate from 40% to 60%.
Same 10,000 visitors, same 400 signups:
| Step | Today | After fixing activation |
|---|---|---|
| Signups | 400 | 400 |
| Finish setup | 200 (50%) | 320 (80%) |
| Send first invoice (activated) | 80 (40%) | 192 (60%) |
| Activation rate | 20% | 48% |
| Retained at week 4 (50% of activated) | 40 | 96 |
96 retained users instead of 40, from the same traffic, with no ad spend. That’s 2.4 times as many. Option A, for $12,000, got to 70.
And the improvement keeps paying. It applies to every future visitor, from every channel, every month.
Now the part that founders often miss. Fixing activation also makes ads cheaper. Run Option A after Option B: the same 300 ad signups now go through the better funnel. 240 finish setup, 144 send an invoice, 72 retained. The cost per extra retained user falls from $400 to $166.67 ($12,000 / 72).
| Extra retained users from $12,000 of ads | Cost per extra retained user | |
|---|---|---|
| Ads before fixing activation | 30 | $400.00 |
| Ads after fixing activation | 72 | $166.67 |
The same money buys 2.4 times as many real users. That’s why the usual order is: fix the leak, then pour in more water. Paying for traffic into a funnel that loses 80% of signups before first value means paying for people to leave.
Two honest caveats about this example:
- The improvements are assumed, not promised. I picked 80% and 60% to make the arithmetic clear. Your real gains could be smaller, or bigger. Measure them.
- Activation is not the whole story. In this example only half of activated users stay to week 4. Once activation improves, the next leak is between first value and habit, and that’s a product question: is the value strong enough to come back for? No onboarding trick fixes a product people don’t need. That’s a product-market fit problem.
As Distribution Beats Product argues, distribution brings people to the door. Activation decides how many come in, and every channel is worth more once it works.
How activation connects to retention and fit
It helps to see where activation sits next to the other ideas in this series.
- Cohort retention shows how many people from each signup month are still around later. Activation mostly decides the first drop in that curve. If you fix activation, the curve starts higher. Whether it flattens depends on the value being worth coming back for.
- Product-market fit shows up as curves that flatten and users who would be very disappointed to lose you. Good activation can’t create fit, but bad activation can hide it. If 80% of signups never see the value, your fit survey and retention numbers will look worse than the product deserves, because most of the people you’re measuring never really tried it.
- Unit economics. A higher activation rate means more paying customers from the same marketing spend, which lowers your customer acquisition cost (CAC, what it costs to win one customer). See Unit Economics in Plain English.
A simple way to track all of this: put the activation rate for each signup cohort next to its week-4 or month-1 retention. When you change onboarding, newer cohorts should show both numbers going up. If activation rises but retention doesn’t, you may have optimised a step that only looked important. Go back to the comparison table and look again.
Key ideas
- Activation is the moment a new user first gets the value your product promised. Signups are not users until they get there.
- Famous aha moments like Facebook’s 7 friends in 10 days and Slack’s 2,000 messages are reported correlations, not proven causes. Copy the method, not the number.
- Find your own activation event by comparing what retained and churned users did in their first days. Pick an action that separates them clearly, is common enough to matter, and is the product doing its job.
- Measure time to value with the median, and shrink it. Curiosity runs out fast.
- Good onboarding removes steps rather than adding tours. Help where a step must exist; delete the steps that don’t.
- Design empty states and offer sample data so new users always know the next action, but count activation on real use.
- Fixing activation usually beats buying traffic, and it makes every future ad dollar go further. Fix the leak, then pour.
Sources: First Round Review, “From 0 to $1B: Slack’s Founder Shares Their Epic Launch Strategy”, February 26, 2015 (Stewart Butterfield on the 2,000-message line, 93% still using Slack, about a week of messages for a 10-person team); Startup Archive, “Chamath Palihapitiya on the growth principles that got Facebook to billions of users”, quoting his talk “How we put Facebook on the path to 1 billion users” (video published October 2012; 7 friends in 10 days, getting users to an aha moment quickly); Tren Griffin, “A Dozen Lessons on Growth”, 25iq, February 10, 2017 (summary that the threshold came from comparing engaged and unengaged cohorts). Ledgerleaf and all its numbers are fictional; the arithmetic was checked in Python.
Related: Cohort Retention: The Chart That Tells the Truth, Product-Market Fit: How to Know You Have It, and Distribution Beats Product.



