New members rarely leave because the welcome was cold. They leave because nothing they did got a response, or because there was no obvious first thing to do. Treat the first week as a sequence — arrive, orient, act, get acknowledged, return — and repair whichever step is actually breaking, rather than writing a warmer greeting.
Flow diagram of 5 steps: Ask for a judgement, not a contribution, Point at one specific thing, by name, with a link, Say plainly what happens to what they write, Let the first act be a vote or a read, not a piece of writing, Answer the first post the same day, even in one line.
Ask why new members leave and the answer offered is almost always about warmth: the community was cliquey, nobody said hello, it felt closed. That is occasionally true, and it is the wrong place to look first, because most people who leave in week one were never treated badly by anyone. They completed one or two steps of joining, hit a step with no obvious next action, and quietly stopped. No unpleasantness occurred, which is exactly why the loss is invisible from the inside — there is no incident to point at, no complaint in the moderation queue, just a registered account that never posted again.
It is more useful to treat the first week as five things that each have to happen in order. Someone arrives. They work out what this place is and what counts as normal here. They do something — a reply, a vote, a submission. Someone notices that they did it. And then they come back a second time, which is the only step that actually indicates anything. Each step has a distinct failure, and the repairs are unrelated to each other. A community losing people at the orient step needs clearer signposting and gains nothing from a friendlier tone; a community losing them at the acknowledge step has a response-time problem that no amount of onboarding copy will touch. Working out which step is breaking is most of the work, and it is cheap: for the last twenty people who joined, check how many posted once, and how many of those single posts got a reply.
The five steps of a first week, and the usual failure at each
| Step | What the newcomer needs | The usual failure | What actually fixes it |
|---|---|---|---|
| Arrive | Confirmation this is the thing they were looking for | A landing page describing the mission rather than showing the activity | Put real submissions and live threads where a new arrival lands |
| Orient | To know what normal looks like here before risking a post | A long guidelines document, read by nobody, and no visible examples | Three or four recent threads that demonstrate the tone, plus one short rule |
| Act | One specific, small thing to do | A general invitation to contribute, which assigns no task to anyone | A named open question or a submission that needs a second opinion |
| Be acknowledged | Evidence a person read what they wrote | A first post left unanswered for two days | Anyone replying within a day, even briefly — this rarely needs to be the founder |
| Return | A reason to open the site again on a different day | Nothing changed since their visit, and nothing told them it had | A decision still open, or a reply waiting for them |
A welcome message is worth sending and it is not a retention mechanism, and conflating the two is how communities end up with elaborate greeting rituals and the same drop-off they had before. What a welcome can do is narrow the field of possible next actions from everything to one thing. What it cannot do is substitute for there being something to react to, or for someone responding when the person does react.
Auto-welcomes are cheap, consistent, and read as what they are within about two seconds — new members have seen thousands of them. That is not a reason to remove one; a template that names the single next action is more useful than nothing, and more useful than a warm paragraph that names none. It is a reason not to count it. A community where the only message a new member ever receives is automated has not made contact with them, however friendly the automated text is, and the arrival will correctly conclude that no person has noticed them yet.
Announcing each new member in a public channel feels hospitable and quietly demands a performance: they are now expected to reply, introduce themselves, and be interesting in front of an audience of strangers on their first day. Some people enjoy that. Many joined to read for a fortnight before saying anything, and being spotlit is precisely what makes them close the tab. There is also a privacy cost that is easy to overlook — a public welcome discloses that a named person has joined this particular community, which is not always something they wanted broadcast. A short private message costs the same to write and carries neither problem.
Requiring an introduction post before participating sounds like a gentle first step and is in practice the hardest possible first task: write about yourself, unprompted, with no template, for readers whose expectations you cannot yet judge. The people who clear that bar easily are the ones who were going to post anyway. The requirement mostly removes the quieter contributors, who are not less valuable and are more sensitive to precisely this kind of exposure. Asking for an opinion on something specific is a smaller ask than asking someone to describe themselves.
The size of the first ask determines whether it happens, and most communities pitch it far too high without noticing, because the founder is estimating effort from their own position — where the norms are obvious, the audience is known, and posting costs nothing. For a new arrival, every post carries the risk of misjudging a room they have been in for four minutes.
Reading a submission and saying whether it is too broad takes three minutes and requires no standing in the community. Producing something of your own takes hours and quietly assumes you already know what good looks like here. Start people on the first kind of task. It also gets you something you actually need, which matters — a first task that is obviously make-work for the newcomer’s benefit is transparent and slightly insulting.
A general invitation to join the discussion assigns no task to any particular person and converts accordingly. Naming the submission you want their view on, and saying why theirs, does most of the work — it removes the choice of where to start, which is the step where people stall. The cost is that this does not scale, and for the first fifty members it does not need to.
People hesitate over a first post partly because they cannot tell what it commits them to. Whether a comment can be edited, whether votes are visible, whether a submission is public before anyone reviews it — these are quick to state and, left unstated, are guessed pessimistically. Being explicit that a submission can be withdrawn removes a real hesitation that no amount of encouragement addresses.
Voting on someone else’s idea is a genuine act of participation with almost no exposure, and it teaches the newcomer how ordering works here by doing it rather than by reading about it. Treat that as a legitimate first step rather than as inactivity. Some members will never write anything and are still doing something the community needs, which is registering a preference that the counts then reflect.
This is the step with the worst effort-to-effect ratio in the entire list, and it is the one most often skipped because it is nobody’s job. A one-line reply within a day is worth more than a considered essay three days later, because the arrival is not judging the quality of the response — they are checking whether anyone is there. After the first two or three exchanges this stops mattering nearly as much, and it can stop being a priority.
One step fails far more often than the other four, and it is the least discussed, because fixing it is unglamorous work with no artefact at the end. Nobody writes an onboarding document about replying to people.
The pattern is consistent across forums, review threads and volunteer projects: the strongest available signal about whether someone posts twice is whether anybody responded to their first attempt. It is worth being precise about why. An unanswered post is not read as a scheduling accident, which is usually what it is; it is read as a verdict on the post. From the outside, silence and rejection are indistinguishable, and the person has no way to tell which one they received. That asymmetry is the whole problem, and it is why a one-line acknowledgement outperforms a careful reply that arrives too late.
Suppose thirty members, a third of whom see any given thread on the day it appears, and a third of those reply — fractions that are illustrative rather than measured, but not pessimistic. That is around three replies per thread on average, which sounds comfortable until you notice it is an average: a thread posted on a quiet evening gets none, and if that thread is somebody’s first, they have been silently filtered out by rounding rather than by judgement. This is why the fix has to be assigned to a person rather than left to the general goodwill of the membership. Someone checking once a day for posts by first-time members takes a couple of minutes and is more effective than any change to the welcome copy.
If the founder is always the one who replies, new members learn to address the founder, and the community becomes a hub with an audience rather than a set of people talking to each other. The dependency also has an obvious ceiling and it arrives sooner than expected. The better arrangement is that answering new arrivals is understood to be something several regulars do, which distributes the load and — more importantly — means the newcomer’s first contact is with the community rather than with its owner.
None of this works on an empty community. If there is nothing on the page to react to, the arrival has no available action and a warmer greeting simply makes the absence more conspicuous — that is a cold-start problem, solved by putting real work and an open question in front of the first arrivals, not by improving hospitality. Welcoming also cannot retain someone who came for something this community does not do. A person who arrived expecting an audience for work they have already finished will leave when they discover the process runs the other way, and that is a good outcome rather than a leak to plug: recruiting them harder would have wasted their time and produced a member with a grievance. The honest version of the funnel accepts that a proportion of arrivals should not stay.
There is a limit on the other side too. Attentive welcoming has a failure mode of its own, where a new member is greeted by four people, tagged in two threads and assigned a task within an hour of joining, and it reads as an organisation that needs something from them urgently. Enthusiasm at that volume is difficult to distinguish from desperation, and it sets an expectation of pace the community cannot maintain. One message, one specific thing to look at, and a reply when they respond is the whole intervention.
A note on the evidence behind this page: Kind Channel is new, its community is small, and no programme has aired yet, so nothing above is derived from our own retention data. We do not have enough members for a first-week figure to mean anything, and any percentage we published would be invented. The arithmetic here is arithmetic, with its assumptions stated so you can substitute your own; the rest is the ordinary observable behaviour of small forums, mailing lists and volunteer groups, offered as a practitioner’s account rather than as a measured result. Be sceptical of any new platform that offers you the latter.
Usually because their first post got no response, or because there was no obvious first thing to do at all. Both look like a lack of interest from the outside and neither is. Silence on a first post is read as a verdict on the post rather than as the scheduling accident it normally is, and a general invitation to contribute gives a new arrival no specific action to take.
Yes, provided it names one specific next action rather than only being friendly, and provided nobody treats it as contact with a person. New members recognise a template immediately, so an auto-welcome narrows the field of possible next steps but does not signal that anyone has noticed them. A community whose only message to a new member is automated has not yet made contact.
It is a worse first task than it appears. Writing about yourself, unprompted, for readers whose expectations you cannot judge yet is one of the harder things to post, so the requirement mainly filters for confidence rather than fit and tends to remove the quieter contributors. Asking someone for a view on one specific submission is a much smaller ask and gets you something more useful.
Ideally several regular members rather than the founder alone, with someone checking once a day specifically for posts by first-time members. If the founder answers everything, new arrivals learn to address the founder and the community becomes a hub with an audience instead of people talking to each other. Distributing it also means a newcomer’s first contact is with the community rather than its owner.