The hardest part of losing a job is rarely the afternoon it happens. It is the mornings after. The alarm has nothing to serve, the commute is gone, and a day that used to be structured by other people's expectations opens up empty. People describe those mornings in very different words, but the shape underneath is the same: time stopped arriving pre filled, and nobody handed them a replacement.
Building a peer support product around that moment sounded straightforward. Match people who have lived the same thing, let them talk honestly, and keep it private by default. The mornings problem turned out to be a design problem, not a messaging problem, and it broke our first assumptions in useful ways.
What was built
SupportMeet at https://supportmeet.com is a free peer support site where people share what they are going through and are matched with circles who have lived the same thing, with practical help and professionals or programmes nearby where that helps. Job loss is one of the headline reasons people arrive, and it is the group that taught us the most about timing.
Three choices mattered for the mornings use case. First, circles are matched by lived experience and by stage, not just by topic. Someone in week one after a layoff needs a different conversation from someone in month four, even when the job titles match. Second, the first thing a new member is offered is small and finishable: one short check in, one introduction, one concrete thing for today. Third, nothing is public by default. People speak more honestly about money, fear, and family pressure when the room is closed.
What broke
The first break was the empty prompt. Early versions opened with a general invitation to share your story. Almost nobody did. A blank box at 8 in the morning, on the worst week of your year, is not an invitation, it is homework. Replacing it with specific, low stakes prompts changed completion completely. People will answer a small true question. They will not perform a summary of their crisis on demand.
The second break was advice arriving too early. Well meaning members jumped straight to job boards, resume tips, and contacts. Useful later, corrosive first. Someone who has not yet said out loud that they are scared does not need a funnel, they need to be heard by people who recognise the feeling. The fix was sequencing: listening first, practical help second, and the interface now makes that order explicit instead of hoping members guess it.
The third break was measuring the wrong success. Signups looked fine while morning return visits lagged. The product was being treated as a place to visit once, describe the problem, and leave. What actually helped was rhythm: the same small circle, the same time of day, a reason to come back tomorrow that was about people rather than content. Once check ins were tied to the circle rather than to a feed, the mornings started to fill in.
What the numbers look like
We judge this feature by a plain test. Does a member in their first week after a job loss come back on a second morning, and a third, without being chased by notifications? Return visits in the early days say more than registrations, because the early days are exactly when structure collapses. A second honest measure is whether members move from reading to writing anything at all, even one line. The first line is the expensive one. After it, conversation gets easier.
We also watch what we do not do. No streak pressure, no public counters, no cheerful nudges that pretend a layoff is an opportunity in disguise. The numbers that matter here are quiet ones: repeat check ins, replies between members, and people staying in a circle past the first hard week.
What stayed
A job does more than pay. It tells you when to get up, who expects you, and what a finished day looks like. Losing it removes all three at once, and no single feature gives them back. What a well matched circle can give back is smaller and real: a time, a few people who know the shape of this, and one next step that fits inside a single morning.
The lesson that survived every revision is that support products should be designed around the hardest hour of the day, not the average one. Build for the morning after, keep the first step finishable, let listening come before advice, and measure whether people return when returning is hardest. That is where trust is either earned or quietly lost.
