Three hours left. The temptation now is to keep improving, because improving is comfortable and shipping is not. Resist it. The remaining hours belong to the pre-flight list, the send, and the reading of what comes back.
Pre-flight
Twenty minutes, in order, on the deployed URL — never on localhost:
- Open it on a real phone, over mobile data rather than your home network.
- Open it in a browser you never use, logged out, with no extensions.
- Walk the whole spine as a stranger. Land, input, wait, result, share. Do not use the shortcut you built for yourself at hour eight.
- Break it on purpose. Empty input, absurd input, the back button mid-flow, a double-clicked submit.
- Check the share preview by pasting the link into a message to yourself.
- Check the tab: real title, real favicon, no framework default.
- Confirm you can be replied to. An email address, a form, a social handle — a way for the first person who likes it to say so.
Fix only what breaks the spine. Everything else goes in LATER.md, including the two things you noticed just now that are bothering you.
Ten minutes of analytics
Use PostHog. One snippet, and you have product analytics, funnels, and session replay in the same place — which matters tomorrow, when "where did they stop?" becomes the only question you care about.
Its free tier is genuinely generous. A product launched today, and most products a year from today, never pay anything — so this is not a decision to agonise over or a cost to weigh against the value. Install it, take the free tier, move on.
Ten minutes means the snippet and three events, and nothing else:
- Arrived — someone landed on the page.
- Started — someone began the one action your promise is built on.
- Finished — someone got the result.
Three numbers, and you can see the shape of the funnel by tomorrow morning. Turn session replay on while you are in there; on day one, watching two people fail is worth more than any chart.
Everything else PostHog does — cohorts, experiments, feature flags, dashboards — is real and useful and none of it is today. Building a full analytics setup at hour twenty-two is a way of avoiding the send.
Send it to ten people by name
Not a broadcast. Ten individual messages, each one sentence, each to someone who has a reason to care:
Made a thing today that turns a receipt photo into an expense line — took me a day. Would love to know if it works for you: [link]
Individual messages get opened, replied to, and forwarded. A single post to a large audience mostly gets scrolled past, and it gives you nothing to read afterwards. Ten real reactions are worth more on day one than a thousand impressions, because reactions come with reasons attached.
Post publicly too if you want — the communities where your users already are, not the ones where builders congratulate each other. But send the ten personal messages first, and send them before you feel ready, because you will not feel ready.
Read behaviour, not compliments
People are kind. Kindness is noise. What you are reading for is:
- Where they stopped. The step people abandon is your real bug, and it is almost never the step you were worried about.
- What they called it. If three people describe your product in words you did not use, those are your words now. Change the headline tonight.
- What they asked for next. Not to build it — to learn what they thought they were buying.
- Whether anyone came back. One unprompted second visit is a stronger signal than fifty first visits.
"This is cool" is not data. "I tried it with my own file and it did not handle the second page" is worth the whole day.
Write it down while it stings
Within twenty-four hours, before the details soften, write half a page: what you shipped, what you cut and whether you were right to, where the hours actually went versus where you planned them, and the one thing you would do differently.
This is the only artefact of the sprint that compounds. The product may or may not deserve a second day — most will not, and that is a successful outcome, not a failed one. You spent a day to find out something that would otherwise have cost you a month.
Then close the laptop. You built and launched a product today. Most people take a quarter to not do that.