Customer onboarding is where many businesses feel the tension between efficiency and warmth most sharply. The first weeks decide whether a new client feels looked after or processed. Yet the same weeks are full of repetitive work: collecting documents, creating accounts, sending the same welcome information, scheduling a kick-off call. Automating the repetition is sensible. Automating the relationship is not. The skill lies in knowing which is which.
This article describes how to automate the administrative parts of onboarding so that the human parts get better, not thinner.
Why automated onboarding feels impersonal
Impersonal onboarding is rarely caused by automation itself. It is caused by automation applied to the wrong things, or applied carelessly. The usual symptoms are familiar from the customer's side.
- They are asked for the same details by three different people or forms, a pattern covered in where customer information gets lost during onboarding.
- They receive a stream of generic messages that ignore what they have already told you.
- Nobody seems to know who is responsible for them, because tasks move between teams without a named owner.
- When something goes wrong, the only route is a no-reply address.
- Timing is off: a message about step four arrives before step two is complete.
None of these is a technology limit. Each is a design choice, and each can be avoided.
Separate administrative steps from relationship moments
Begin by listing every step in your current onboarding, then sort each into one of two groups.
- Administrative steps that follow a fixed rule: creating the customer record, issuing a form, checking that required documents are present, setting up a login, sending a receipt, booking a standard session, reminding about an outstanding item.
- Relationship moments where the customer needs to feel understood: the first conversation, clarifying what they actually want, explaining something unusual in their situation, handling a worry, celebrating the first result.
Automate the first group and protect the second. A useful test is to ask whether the customer would be disappointed to learn that a person did not do it. A document checklist reminder: no. A welcome call from the person who will run their account: yes. The wider approach to structuring this work is in how to design a customer onboarding workflow that scales.
Collect information once and use it everywhere
The most impersonal thing a business can do is ask for something it has already been given. Automation can prevent this, if the information is captured once into a single record that every later step reads from. Customers' details held in separate forms, inboxes and spreadsheets are exactly the disconnected stores that IBM describes in its overview of data silos, with the inconsistency that follows.
Under the Protection of Personal Information Act, this also has a legal side. Section 13 requires personal information to be collected for a specific, explicitly defined and lawful purpose, and section 18 requires the responsible party to take reasonably practicable steps to make the person aware of what is being collected and why. An onboarding form is the natural place to do both properly: ask only for what the service genuinely needs, say in plain words what it will be used for, and record the answers once. Customers tend to read a short, clear explanation as respect rather than red tape.
Use what you know to personalise sensibly
Personalisation in onboarding does not need clever technology. It needs the process to respond to facts you already hold.
- Use the customer's own words. If they said their priority is getting invoices out faster, the welcome message should mention invoices, not a generic list of features.
- Branch the sequence. A sole proprietor and a company with five branches need different steps, so do not send both the same twelve-step checklist.
- Name a person. Messages should come from a real team member who can be replied to, even if a system sends them.
- Adjust to pace. If a customer completes a step early, skip the reminder. If they stall, let a person follow up rather than sending a fourth automated nudge.
The measure of success is that a customer reading the message feels it was written for them, which is a matter of accuracy and relevance more than of tone.
Build in human checkpoints
Automation should hand work to people at the right moments, not hide them. Examples include:
- A person reviews and approves anything that changes the customer's contract, pricing or credit terms. Microsoft's Power Automate approvals feature illustrates the pattern: a flow pauses and asks a named person to approve or reject before continuing.
- A person is alerted when a customer has not progressed for a set number of days, or has replied to an automated message with a question.
- A person makes the call or sends a short personal note at defined milestones, such as the first week and the first month.
It also helps to give the customer an obvious way to reach a person at any point. A visible "talk to someone" option costs little and removes the feeling of being trapped in a machine. The cost of leaving onboarding as a manual list of emails is considered in manual customer onboarding checklists and the cost of delayed revenue, and the reasons the arrangement fails as volumes rise are in why customer onboarding breaks as the business grows.
Test it as the customer
Before launch, run a real or dummy customer through the whole sequence and read every message in order, on a phone. Look for repeated requests, broken personalisation such as a blank first name, messages that arrive out of order, and any point where a question has nowhere to go. After launch, ask new customers one question at the end of onboarding: was there anything you were asked for twice? Their answers show where the process still leaks. Also monitor for failed steps, since an onboarding that stalls silently is the least personal experience of all.
An illustrative example
The following is a hypothetical case. A Stellenbosch bookkeeping firm onboards about thirty small-business clients a quarter. Previously, an administrator emailed a list of required documents, chased by phone, and the bookkeeper first met the client after the paperwork was done. Clients often had to resend the same documents because they were saved in different folders.
The firm redesigns the process. A single online form captures business details, VAT status and the purpose of the engagement, with a short explanation of how the information will be used. Documents are uploaded to a client folder, and an automatic checklist shows what is outstanding. Reminders are sent twice, from the administrator's name, and a third is sent to her as a task instead. As soon as the form is submitted, the bookkeeper receives the answers and books a thirty-minute call to discuss priorities before any paperwork is chased. The administrative load drops, and the first human contact now happens earlier and with more preparation, which clients mention in their feedback.
The point
Onboarding feels impersonal when customers are made to repeat themselves, are ignored by a system or cannot find a person. Automate the rule-based steps, capture information once, personalise from facts you hold, and put people at the moments that matter. Done well, automation gives your team more time for the conversations that make a new customer confident that they chose well.
Sources and further reading
- Section 13 Collection for specific purpose - POPIA (last verified 10 October 2026)
- Section 18 Notification to data subject when collecting personal information - POPIA (last verified 10 October 2026)
- What Are Data Silos? - IBM (last verified 10 October 2026)
- Get started with Power Automate approvals - Power Automate (last verified 10 October 2026)
Related Mellorca reading
- Why Customers Keep Being Asked for the Same Information
- How to Build a Business Application Inventory That Is Actually Useful
- How to Move Operational Knowledge Out of People's Heads