Activation is the only metric that matters at seed

An activation metric measures the share of new users who reach a first real moment of value, such as sending a first invoice or seeing their own data in a dashboard, within a set time after signing up. Activation rate is that number divided by everyone who signed up in the same period.

Our position is that at seed it is the number to steer by, ahead of signups, MRR and even retention. That is a strong claim and it has limits, which we will get to. First, the case for it.

Why activation beats the other numbers at seed

Signups measure your marketing. They tell you the ad, the post or the launch worked. They say nothing about whether the product did, and they are the easiest number to inflate with spend. A seed company with rising signups and flat activation is paying to fill a bucket with a hole in it.

MRR is too small and too noisy to read. At seed, a large share of revenue often comes from customers the founders sold personally, on terms nobody will repeat. One account landing or leaving swings the month. It matters for the bank balance. It is a poor guide to whether the product is working for the next hundred strangers.

Retention is the truth, but it arrives late. Whether people are still using you after three months is the real test of product-market fit, and it is what Series A investors look at hardest. The trouble is the wait. A change you ship today shows up in three-month retention in, unsurprisingly, three months, and a seed company gets very few three-month cycles before the next raise.

Activation is the front edge of retention. It shows up within days, it responds to design changes within a release or two, and when it is defined well it predicts who will still be around later. That combination of speed and meaning is why it deserves the top of the dashboard. You can run ten experiments on activation in the time it takes to read one retention cohort.

Defining your activation event

The activation event is a single action a user takes that shows they have got real value. Choosing it is the hard part, and most teams pick something too early, like completing onboarding, because it is easy to track. A finished setup wizard proves the user was patient, and tells you little else.

• It happens early. Ideally in the first session, at most in the first week. If it takes a month, it is a retention measure.
• It is value to the user, in their terms. Something they would describe to a colleague as the product working for them.
• It separates stayers from leavers. Users who reach it should retain clearly better than users who do not. If there is no difference, pick another event.
• It is one action, with a time window. "Sent a first invoice within seven days of signup" beats a weighted score of six behaviours.

The famous examples were found with enormous amounts of data, in hindsight. Chamath Palihapitiya has described Facebook's growth team focusing on getting new users to seven friends in ten days. Stewart Butterfield told Andrew Chen, for The Cold Start Problem, that Slack treated a team that had exchanged 2,000 messages as one that had really tried the product. At seed you will not have the data to find a number that precise. Pick a sensible event, check it against retention as soon as you have a few cohorts, and change it if it fails the test.

Worked examples by product type

Two patterns run through the table. The weak definitions are all steps on the way to value, chosen because they are easy to count. The better ones involve the user's own data, their own customers or their own team, because that is when the product stops being a demo and starts being theirs.

Products with more than one kind of user need more than one event. When we rebuilt Great People, an HR performance platform, we defined eight user demographics and gave each its own flows, and added a mobile app so employees get value from the platform as well as the employer. An employer and an employee reach value at different moments. Averaging those into one activation rate hides whichever one is broken.

Measuring it without a data team

You need two timestamps per user: when they signed up and when they first did the activation event. Any product analytics tool will give you that, and so will a query against your own database. Then:

• Group new users by the week they signed up.
• For each weekly cohort, count how many reached the activation event inside the window.
• Divide by the cohort size. That is your activation rate for the week.
• Once you have a few months of data, compare retention for users who activated against those who did not. That is the test of whether you chose the right event.
• Split by acquisition channel. Paid users often activate at a different rate from referrals, and the blended number hides it.

For a sense of scale, Lenny Rachitsky's 2022 survey with Yuriy Timen, which drew more than 500 responses, put the average activation rate at around a third and the median at around a quarter, with SaaS products a little higher. Treat that as a loose reference only. Everyone defines the event differently, so your own trend week on week is worth far more than anyone's benchmark.

The design changes that typically move it

Activation is where design has its most direct effect on a seed company's numbers, because almost everything between signup and the activation event is a flow someone designed, or forgot to. The changes we reach for most:

• Remove steps before value. Every setup screen, preference question and verification gate before the first win costs a share of users. Move what you can to after the activation event. Our signup flow audit covers the checks in order.
• Never start on an empty screen. Sample data or a pre-filled template lets a new user see the product working before they have put anything in. We wrote up how to design empty states properly.
• One obvious first action. The first screen should point at the activation event and nothing else of equal weight.
• Show progress and let people resume. Anything that needs documents, data or a colleague will be finished later. Make later easy.
• Ask for the invite at the moment of value. A teammate invite lands better straight after a user's first success than as step two of onboarding.
• Match the model to the time it takes. If your activation event takes weeks, a fourteen-day trial will fail regardless of design. We covered how to choose between a free trial, a demo and freemium.

Where the argument breaks

Activation is a proxy, and proxies can be gamed. Lower the bar for what counts and the number goes up while nothing improves. Push users through the event with tricks, and you get the action without the value behind it, which is why the retention check above is not optional.

It also cannot create demand. If the people signing up are the wrong people, or the problem is not painful enough, a beautifully designed path to the activation event will simply deliver more of them to the point where they decide it is not for them. Activation tells you whether the product works for the people who arrive. Whether enough of the right people arrive is a separate question, and at seed it may be the bigger one.

And in some businesses it fits awkwardly. Sales-led enterprise products may have a handful of new accounts a quarter, where a conversation with each one beats any rate. Marketplaces need activation on both sides, and one side being healthy says little about the other. In those cases the principle still holds, find the earliest real sign of value and design towards it, but the single tidy percentage does not.

None of that weakens the core point. At seed you have little time, little data and a product still changing every week. You need a number that moves fast enough to learn from and means enough to steer by, and activation is the one that does both.