Signed on Friday, properly started by Monday.

The problem

The deal closes on a Friday and everybody is delighted. By Monday the person who sold it is off selling something else, the person who will deliver it has not been told it exists, and the client is sitting through the most dangerous week of the entire relationship with absolutely nothing to look at. They signed because they were convinced. Then nothing happened for nine days, and conviction has a shelf life.

Inside the company the work is real and completely invisible. Someone has to raise the invoice, set up accounts, create the folders, add them to the right channels, collect the things you need from them, book the kickoff, write the welcome, tell the team, and put a first task somewhere it will actually get picked up. Twenty-odd steps across six systems. Not one of them is hard. Every one of them is somebody's to remember, and that somebody is reliably the person who was already the bottleneck.

What happens next is the classic failure: partial completion. Eighteen of the twenty steps get done. The two that do not are different every single time, so nobody can learn the pattern. The access nobody granted. The folder nobody made. The thirty-day check sitting in a calendar belonging to somebody who left in June. Your client experiences all of this as a company that is a bit disorganised, which is the precise opposite of what they thought they were buying.

Then it costs you twice. Once now, because a client with a bad first fortnight spends the rest of the relationship watching you rather than trusting you. And once much later, because the economics of retention are brutal in both directions: winning a customer costs five to twenty-five times what keeping one does, and five points of retention moves profit by 25 to 95%. Onboarding is the cheapest retention work there is, and it is the part most companies run from memory.

How it works

Seven things happen between the signature and the end of the first month. The point is not that each is automated. It is that none of them can be forgotten.

1. 01The signature is the trigger, not a notificationThe signed contract starts the sequence directly rather than producing an email somebody has to act on. Everything the pipeline needs was captured during the sale, so nothing has to be retyped and nothing waits for the salesperson to write a handover note. The handover note is the record, and it was being filled in all along.
2. 02The money is handled firstThe invoice or the deposit goes out immediately, with terms, purchase order references and the payment link already on it. Where a deposit gates the start, the pipeline knows that and holds the rest rather than provisioning everything for a client who has not paid. Where it does not, nothing waits. This step is first because it is the one most likely to be deferred and then forgotten.
3. 03Accounts, access and workspace are created togetherLogins, licences, shared folders with the structure already in them, the project in whatever tool you run, the channels, the groups, the calendar invitations. Created from one definition per kind of client, so the fourteenth client this year gets exactly what the first one got, including the parts the first one only got because somebody happened to think of it.
4. 04The client is told what happens next, with dates on itA welcome that names who owns their account, what the first three milestones are, when each lands and what you need from them, with the requests themselves gathered through a short form rather than through an email chain. Most of the anxiety of a first fortnight is not about progress, it is about not knowing what is supposed to be happening.
5. 05Kickoff is booked rather than proposedOffered against real availability from the people who will actually be in the room, held and then confirmed, with the agenda attached and the prerequisites listed. Proposing three times by email is where a week goes, every time, and it is a week spent on the one thing the client is currently measuring you by.
6. 06The first real task lands on a named personNot a project created and left empty. An actual first task on an actual person's list, with the scope the client agreed written in the words it was agreed in. A project nobody has been assigned to is a project that starts whenever somebody notices it, which is usually when the client asks.
7. 07The checks happen because they are scheduled, not rememberedAt seven days that everything provisioned is actually being used, at thirty that the relationship is where it should be, with the question going to a person rather than being a form. Anything incomplete escalates before the client finds it. The thirty-day check is the step every company intends and almost none perform, which is why it is in the pipeline rather than in a calendar.

What it's built from

Five pieces, and all five are probably things you already pay for. What is missing today is not a tool, it is the thing that makes them act in order.

• n8n, the part that decidesSelf-hosted on European infrastructure, and the actual product. Every step, condition, dependency, timer and escalation lives here as something you can open and read. When onboarding changes because you have added a new service, one person changes it in one place, and the next client gets the new version without anybody being told.
• Stripe, for the invoice and the depositThe first invoice or deposit raised the moment the contract is signed, with a payment link on it, and the pipeline reading whether it has been paid where payment gates the start. Invoicing on day one rather than day nine is a cash flow improvement that costs nothing and is almost entirely a question of whether anybody remembered.
• HubSpot, for what the sale already knewThe record the pipeline reads from and writes back to, so onboarding runs on what was agreed during the sale rather than on a handover somebody wrote from memory. It is also where the state lives afterwards: which steps completed, which stalled, and how long the whole thing took, which is a number most companies have never seen.
• Notion, for the workspace and the planFolders, project space and the client-facing plan created from a template per kind of engagement rather than copied from the last one, which is how small errors get inherited for years. Drive, SharePoint, Asana, Linear or whatever you actually use works the same way.
• Slack, for the people sideThe team is told who the client is and who owns them, the channels are created, the first task is announced where the person will see it, and anything that stalls escalates to a name rather than to a channel. A step that fails quietly in an onboarding pipeline is only discovered by the client.

What it's worth

Do this with your own numbers rather than ours. It takes a minute, and it is the only version of it worth believing.

• You onboard 8 clients a month and it takes about three hours of somebody's time each, spread across six systems.24 hours a month, nearly all of it on steps with no judgement in them, performed by whoever is most senior enough to be trusted with the access.
• The first invoice currently goes out nine days after signature and your average first invoice is 8,000 euros.On 8 clients a month that is 64,000 euros sitting nine days further out than it needs to, permanently. Invoicing on the day costs nothing and never stops paying.
• Two steps in twenty get missed on a typical onboarding, and a different two each time.Every client sees at least one thing go wrong in their first fortnight, which is the fortnight they are deciding whether they were right about you.
• You lose one client a year to a start that went badly, worth 30,000 euros in annual revenue.Replacing them costs five to twenty-five times what keeping them would have, on the published figures. A five point improvement in retention is worth between 25 and 95% on profit, and onboarding is the cheapest place to find it.

These are illustrations using published averages and round numbers, not results we are reporting. Your onboarding volume, first invoice value and retention will move every line, which is why the first week of the build counts the steps you actually perform rather than the ones the process document says you do.

The research behind those assumptions

Published figures rather than ours. The first two are the economics of keeping a customer, and the third is how well the thing that keeps them is usually done.

• 5-25xmore expensive to acquire a new customer than to keep an existing one. The Value of Keeping the Right Customers, HBR
• 25-95%is the profit effect of improving customer retention by five percentage points. The Value of Keeping the Right Customers, HBR
• 12%of employees strongly agree that their organisation does a good job of onboarding. The same discipline problem shows up on the staff side. Gallup, Why the Onboarding Experience Is Key for Retention
• 19%of the working week goes on hunting for information, which is what a half-provisioned client account produces for everybody who touches it. McKinsey Global Institute, The Social Economy

What it costs

This is sold as one system at the automation price on the services deck, and it sits in the middle of that range at three to four weeks. What moves it is the number of systems that have to be provisioned and how different your engagement types really are. One kind of client across four systems is the bottom. Four engagement types, two of which have regulatory steps, across nine systems, two of which have no usable interface, is the top.

The running cost is a small hosting line and whatever the underlying tools already cost you, because this does not introduce a new subscription for anybody to log into. That is deliberate. An onboarding system that lives in its own portal is a twenty-first step rather than a replacement for the other twenty.

Outside the build price is somebody watching it, which is the care retainer. The thing worth watching here is drift. Onboarding changes every time the business adds a service or a system, and a pipeline that was correct in March is quietly provisioning the wrong things by September if nobody is responsible for noticing.

What goes wrong, and what happens then

Every one of these is in the build because it happens, not because somebody raised it once in a workshop.

• The deposit never arrivesWhere payment gates the start, the pipeline holds the provisioning rather than completing it and then quietly waiting. The client is chased on a cadence and a person is told on day five. The failure to avoid is a fully set-up client who has not paid and does not feel any urgency about it.
• A step half completesEvery step is checked after it runs rather than assumed to have worked, and a partial failure escalates with the specific step named. The whole value of this is that nothing is silently skipped, so a system that assumes success is a system that has recreated the original problem with more steps.
• The client does not send the things you asked forChased automatically, with the list narrowing to what is still outstanding rather than resending the whole thing, and escalated to the account owner before it delays the first milestone. A kickoff that happens without the prerequisites is a kickoff that has to happen again.
• Two steps want to run in the wrong orderDependencies are explicit, so access is not granted before the contract is countersigned and the kickoff is not booked before the people who need to be in it have been assigned. An onboarding pipeline that fires everything at once is how a client gets a welcome email before the person who will serve them knows they exist.
• A system is down when its step runsRetries with backoff, then holds the step and raises an alert naming it, rather than marking the onboarding complete with a gap in it. Nothing reports success it did not verify.
• This client is genuinely differentSteps can be skipped or added per client by a person, with the reason recorded, and the pipeline continues around the change. A system that cannot be overridden gets worked around entirely within a month, and then you have a pipeline and a shadow process.
• The thirty-day check becomes a formIt goes to a person as a task with the client's actual first month attached: what completed, what stalled, what they have used. The check exists to produce a conversation. Turning it into a satisfaction survey is how companies collect a four out of five from a client who leaves in March.