Advisers spend around 15 hours per client on onboarding, and more than half of that time has nothing to do with the client. It goes on chasing documents, re-keying data, drafting follow-up emails and waiting for systems that don’t talk to each other. That’s administration, and working harder won’t fix it. Redesigning the process will.
This playbook walks through how to do that, step by step, in a way that stands up to FCA scrutiny and doesn’t need a big technology budget or an IT department.
Where the onboarding time goes
Most of the lost time sits in the middle two of four stages, so be precise about where the hours go before you touch any software.
The four stages are: initial data collection (fact-find, ID checks, source of wealth documents), internal processing (keying information into the back office, compliance checks, building the file), suitability assessment and documentation (research, drafting, review), and client communication (welcome letters, progress updates, confirming next steps). In a typical firm, stages two and three are where most of the non-client-facing time builds up. Stage one is slow because it depends on clients sending things back at their own pace. Stage four is slow because it’s done by hand, over and over, for each client.
The cost is easy to underestimate. For a firm onboarding 50 new clients a year, half of 15 hours per client is roughly 375 hours of admin. At a conservative £50 an hour for adviser or paraplanner time, that’s £18,750 a year in process overhead, before you count the growth it’s holding back. Automation works differently at each stage, so targeting the wrong one first wastes effort.
What level of help you need
Most onboarding bottlenecks need Level 1 or Level 2 help, not engineering. Place each one at the right level before you look at any tool.
- Level 1: education. Problems you can solve with a better prompt, a better template, or fifteen minutes of setup in a tool you already pay for. If your CRM has a workflow builder nobody uses, or your back office has a document request feature sitting idle, that’s a Level 1 fix. Free, and available today.
- Level 2: integration. Two or three systems need connecting so data moves between them automatically instead of being re-typed. This is the most common real need in IFA onboarding. Make, n8n or Zapier handle the configuration. Expect days to a few weeks, and a cost in the low hundreds to low thousands of pounds.
- Level 3: custom build. Real engineering: custom pipelines, complex logic, orchestrated AI. Most IFA onboarding doesn’t need it. If a vendor says yours does, ask them to explain why Level 2 won’t work.
Solve at Level 1 first, then Level 2 if you need it. A firm that thinks it needs a custom build often needs a two-hour configuration session and a better template library.
Map the process before you automate it
Automating a broken process gives you a faster broken process, so the mapping comes first whatever technology you end up using. It isn’t a technology decision or a procurement exercise. It’s a simple audit.
Spend two or three hours tracing one recent onboarding from first contact to completed file. For each step, write down who does it, how long it takes, what triggers it, which system it lives in and what can go wrong. Before suggesting anything, I start the same way: map what the firm already uses and who owns each part. You’re looking for three things: purely mechanical steps (copying data from one place to another), emails that get written from scratch every time, and hand-offs where work sits waiting for someone to pick it up.
Those are your automation candidates. Steps that need a decision, regulated judgement, discretion or professional knowledge aren’t. Keep those human. The aim is to take away the mechanical work that stops people doing the parts only they can do.
What to automate, and in what order
Start with document collection, then data transfer, templates and meeting capture:
- Document collection. A secure digital portal, even a simple one, that sends automatic reminders and tracks what’s arrived will save more time, faster, than almost any other single change. Clients who get a clear, professional request with an easy upload return documents sooner, and the adviser spends less time chasing. Most modern back-office platforms include this, so if yours does and you’re not using it, that’s a Level 1 fix.
- Data transfer between systems. Most re-keying happens in the gap between the fact-find tool and the back office. A Level 2 integration in Make or n8n can usually move the structured data across automatically once a fact-find is marked complete, so manual entry becomes the exception instead of the default.
- Templated communications. A library of well-written, compliance-reviewed templates for routine onboarding emails (after the first meeting, document request, progress update, welcome to the service) cuts drafting time and makes the client experience more consistent. Usually a Level 1 change. AI can help draft the templates, but a person checks them for regulatory accuracy and tone before they’re used.
- Meeting capture. Tools that record a conversation and draft a follow-up note or fact-find summary can save real time after meetings, but the adviser must review and confirm the output before it goes into any client file or informs any recommendation.
A workflow built on these four can send a document request, track whether it’s been done, chase it after 48 hours and flag it to the adviser if it’s still missing after five days. The adviser doesn’t get involved until something needs their judgement. The document collection playbook covers that first step in more detail.
Which touchpoints to keep human
Keep a person on every touchpoint where a decision is made. Automation only hollows out a relationship when it’s applied to everything, including the moments where a human presence matters. The test isn’t whether a step can be automated. It’s whether automating it takes away something the client would notice and value.
A simple rule works: informational touchpoints get automated, decisional touchpoints stay human.
| Automate it (informational) | Keep it human (decisional) |
|---|---|
| Document requests and upload chasing | The first conversation |
| Status updates and progress tracking | The risk discussion |
| Appointment and deadline reminders | The recommendation discussion |
| Data transfer between systems | The welcome call after the fact-find |
| ID verification workflows | Any moment of genuine client concern |
The venture firm a16z describes a “barbell” model for AI in client services that fits here: automate the high-volume, low-value interactions so you can put human time into the ones that matter most. A progress-chasing email can be automated, because the client wants to know where things stand and doesn’t particularly want to ring your office to find out. The welcome call after the first fact-find shouldn’t be. That’s when a client decides whether they trust you.
Start with new clients
Introduce the new process to new clients first, because they have no habits with your firm yet. Existing clients who’ve always had a paper pack and a phone call will notice the change. New clients will just treat whatever you set up as normal, and it’s much easier to set a digital-first expectation at the start than to reset an old one later.
In practice, make automated document collection, digital ID checks and progress-tracking portals standard for new clients before rolling them out to your existing book. It also gives you a clean read on how clients respond.
Transparency and data handling
Clients dealing with automated communications should always know they can reach a person, and anything that looks like a conversation should make clear it isn’t a human. That means automated emails signed clearly by the firm (not dressed up as a personal email from the adviser), a visible route to the adviser or their support team, and no doubt about when a client is dealing with an automated step and when with a person. Done well, this builds confidence: clients see an organised process that knows when their situation needs a real conversation.
Any automated system collecting or processing client personal data has to work within your existing GDPR framework. The EDPB has published guidance noting that vague or over-broad consent in professional services can create ambiguity and compliance risk [1]. Make sure your digital onboarding captures specific consent for a stated purpose, not a blanket agreement buried in a welcome email.
What the FCA expects when you automate client-facing processes
Consumer Duty applies directly. It requires firms to deliver good outcomes for clients, which includes an onboarding process that’s clear, fair and not misleading, clients who understand what they’re being asked for and why, and a firm that can show real engagement and informed consent at each stage [2].
Done properly, automation helps with all of that. Consistent communications reduce misunderstandings, a documented workflow is easier to audit than one that lives in someone’s head, and automatic reminders stop clients slipping through the cracks. It can also work against Consumer Duty. A process that sends every client down an identical path regardless of complexity or vulnerability fails the duty’s focus on the individual. A system that chases documents with no way for a client to say they’re struggling creates exactly the friction the duty is meant to remove. And any automated message a client might reasonably read as advice, rather than admin, carries regulatory risk.
Five things keep onboarding automation defensible:
- Write down your process map, before and after. The FCA expects firms to show their processes produce good client outcomes, and a written process map is where that starts.
- Review every automated communication against Consumer Duty standards. Every templated email or message is a regulated communication. Have your compliance officer or another suitably qualified person review them before they go live.
- Build in vulnerability identification. Your intake should give clients a clear, easy way to say they need extra support, and pass that to a person promptly. An automated system that doesn’t is a Consumer Duty gap.
- Make human review explicit at regulated decision points. Any AI-assisted output that feeds a suitability assessment, a recommendation or a regulated document goes through a logged human review. “The system drafted it and I checked it” is a legitimate workflow. “The system handled it” isn’t.
- Check the process regularly. Every six months, look at how the workflow is performing: are clients finishing onboarding faster, and where does it stall? Consumer Duty requires ongoing monitoring of outcomes, not a one-off design review.
Using AI tools in onboarding
AI tools now exist for specific onboarding tasks: summarising meeting notes, drafting first versions of suitability-related documents, and pulling structured data out of uploaded documents. Used well, they cut a lot of time from the processing and documentation stages. Used carelessly, they create compliance risk, because any AI output needs a person to review it before it’s used in a regulated context. That review is the step that keeps the process defensible, so it doesn’t get skipped on busy days.
Choosing the vendor matters too. Gartner has noted that a significant share of tools marketed as AI agents in 2026 are older automation with a new label. Before building an onboarding workflow around any AI tool, check the capability is real, that the contract is clear about your data rights, and that your workflow would survive the tool being discontinued or changing direction. Depending on one vendor is a business continuity risk for any regulated firm. The guide to spotting AI-washing has the questions to ask.
What onboarding automation achieves
Intelliflo’s data from firms using integrations like wealthLink and zeroKey shows around a 20% increase in average clients per adviser from the efficiency gains. Don’t take that as a guarantee for your firm. It’s a direction, and it makes sense: if an adviser spends seven or eight hours of each fifteen-hour onboarding on mechanical tasks and you recover most of that, you’ve changed their capacity.
The more interesting result isn’t cost or capacity. Dunbar’s number, the anthropological idea that people can keep up roughly 150 stable relationships, is relevant here: an adviser looking after 75 or more clients is already near the limit of what they can manage as real relationships. When six or seven hours per new client come back, that time goes into the conversations that build the relationship: longer discovery meetings, more proactive annual reviews, better preparation. Automating the admin makes the relationship better, because the adviser’s attention isn’t being used up by process.
The firms that get there start with an honest process map, fix the Level 1 things first, and build the Level 2 integrations carefully, instead of chasing a system that promises to do everything.
This article is for information only. It isn’t regulated financial advice or a compliance opinion. Consult a qualified compliance professional for advice specific to your firm.
Sources
[1] EDPB, ‘Guidelines 1/2026 on processing personal data for scientific research purposes under Article 89 GDPR’, European Data Protection Board, 2026. Available at: https://www.edpb.europa.eu/our-work-tools/documents/public-consultations/2026/guidelines-12026-processing-personal-data_en [Note: public consultation document; the guidance on consent specificity cited here reflects the draft position. Firms should monitor for the finalised version before treating it as settled regulatory requirement.]
[2] Financial Conduct Authority, Consumer Duty (PS22/9), final rules and guidance (July 2022), https://www.fca.org.uk/publications/policy-statements/ps22-9-new-consumer-duty