Retirement Advice at Scale: How Regulation, Smaller Pension Pots and Automation Are Rewriting the Business Model

Retirement advice is becoming an operating-model problem

Retirement planning is usually described as a conversation about investment returns, tax and longevity. For advisory firms, however, the harder 2024 question was operational: how can a regulated business give more clients genuinely personal decisions when the average pension relationship is becoming smaller, more fragmented and more demanding? The answer affects staffing, technology, product governance and the economics of every client segment.

An article published by FT Adviser on May 2, 2024 reported research conducted by FT Longitude for BNY Mellon Investment Management. Nine in ten advisers and planners surveyed expected thinking about retirement solutions to change significantly during the following five years, while 88% said industry attitudes and approaches needed to change faster. Those expectations were not a prediction about one product. They described pressure on the whole service model.

The strategic challenge is to standardize repeatable work without standardizing the client. Data collection, reconciliation, cash-flow calculations, evidence trails and routine monitoring can be made consistent. The choice of income pattern, risk capacity, guarantees and contingency reserves still requires judgment. Firms that confuse these two layers either create an expensive manual factory or an efficient process that produces unsuitable outcomes.

Why the accumulation and decumulation split is insufficient

Accumulation has a relatively clear direction: contributions enter a portfolio intended to grow over a long horizon. Decumulation reverses the cash flow, but not in a simple mirror image. A retired household may draw income, preserve a reserve, support relatives, meet care costs, change housing and revise plans after market movements. The portfolio must finance decisions whose timing and size are uncertain.

The research found that 88% of independent financial advisers considered a clean division between accumulation and decumulation overly simplistic. That finding matters because many systems and product ranges are still organized around a single retirement date. Real clients may reduce work gradually, continue earning after taking pension income or alternate between withdrawals and periods of preservation. A business process built around one irreversible switch cannot represent those paths well.

Richard Parkin, head of retirement at BNY Mellon Investment Management, argued that the industry needed a new approach to retirement income. The commercial implication is that advisers need a service able to revisit decisions rather than merely implement an initial withdrawal. Review capacity, not just initial recommendation capacity, becomes the scarce resource.

Fragmented pension pots change unit economics

Generation X reached the retirement market with a different asset pattern from many predecessors. Defined-benefit provision was less dominant, employment changes created multiple pension accounts, and responsibility for converting savings into income shifted toward individuals. Each separate pot can require identification, valuation, fee comparison, beneficiary checks and transfer analysis before an adviser can see the complete household position.

More people may therefore need advice while bringing smaller average balances to each engagement. That combination is difficult for a firm priced as a percentage of assets. The work required to collect facts and demonstrate suitability does not fall in proportion to portfolio value. A client with five modest pots can require more administrative effort than a wealthier client with one consolidated arrangement, yet generate a lower fee.

This does not make smaller clients commercially irrelevant. It means the service must be designed deliberately. Firms need clear eligibility rules, modular stages, predictable review cycles and controlled exceptions. If every case begins as an entirely bespoke project, scarce specialists spend time recreating checklists instead of resolving the decisions that genuinely differ.

Consumer protection is part of product design

The Financial Conduct Authority examined this market in its thematic review of retirement income advice, first published on March 20, 2024. It found that most reviewed files showed suitable advice, but some firms did not adequately consider sustainable income or provide the right information. A small number of consumers lost valuable guarantees or incurred unnecessary charges.

Those findings convert abstract compliance into design requirements. A retirement workflow should make guarantees visible before transfer choices, require evidence for withdrawal assumptions, display charges in the decision context and trigger escalation when facts are incomplete. Quality cannot depend on one attentive employee remembering every hazard at the end of a long case.

Consumer Duty strengthens that principle. A firm must be able to explain how its service supports good outcomes, offers fair value and communicates in a way clients can understand. The evidence is generated throughout the process: what alternatives were considered, which assumptions were tested, where uncertainty remained and how the client responded. Compliance records should therefore be a by-product of good work, not a separate reconstruction after the recommendation.

A scalable service separates routine production from judgment

Advisory firms can map each case as a sequence of inputs, calculations, judgments, communications and controls. Routine production includes requesting statements, normalizing holdings, checking missing fields, preparing cash-flow scenarios and scheduling reviews. Judgment includes interpreting a client's priorities, assessing capacity for loss, resolving conflicts and deciding whether a guarantee should be retained.

The distinction determines what technology should do. Automation is valuable when a task has stable inputs, explicit rules and a verifiable output. Human review is essential when objectives conflict, facts are ambiguous or a decision carries irreversible consequences. A workflow should expose these boundaries instead of pretending that every step is equally automatable.

A practical control map

• Capture data once, identify its source and record when it was last verified.
• Reconcile pension values and benefits before generating recommendations.
• Apply consistent assumptions, then show advisers where an override occurred.
• Route guarantees, vulnerability indicators and unusual withdrawals to specialists.
• Keep calculation versions and client communications in one evidence trail.
• Monitor whether the implemented plan behaves within the agreed boundaries.

This map makes productivity measurable. Management can see whether delay comes from missing client information, provider response times, repeated data entry, specialist queues or adviser rework. Investment can then target the bottleneck rather than purchase technology because competitors have done so.

Cash-flow modelling is a conversation, not an oracle

Modern cash-flow tools can project income, spending, tax, inflation, investment returns and life expectancy across decades. They help clients see why a withdrawal that appears affordable today may reduce resilience later. They also let advisers compare retirement dates, spending paths and contingency reserves consistently.

A projection is still conditional on assumptions. A smooth average return can conceal damaging early losses, while one inflation rate cannot represent every household expense. Longevity is not known, and future policy can change tax or pension rules. Presenting one precise line as a forecast creates false confidence.

A stronger process uses scenarios and stress tests. It shows a central path, adverse market sequences, higher inflation, longer life and exceptional spending. The adviser explains which decisions can be reversed and which cannot. The client can then choose a plan with visible margins rather than chase an apparently optimal number.

Automation can refresh these scenarios when values change, but governance must control assumptions and disclosures. Model owners should approve parameters, document updates and test calculations. Advisers should understand enough of the model to challenge surprising results. Software increases capacity only when users can rely on its logic.

Artificial intelligence needs a narrow business case

Artificial intelligence can classify documents, extract values, summarize correspondence and highlight inconsistent data. Used carefully, it can shorten the administrative stage of a case and help an adviser search a complex record. It may also prepare alternative explanations for clients with different levels of financial confidence.

These uses do not transfer accountability. Extracted values need validation, summaries can omit qualifications, and generated language may sound certain when the evidence is weak. Sensitive financial data also raises questions about access, retention, suppliers and model training. A firm needs approved use cases, human sign-off and a route for reporting failures.

The best early targets are bounded tasks with observable errors. Document extraction can be compared with the original statement. A missing-field alert can be verified. A draft meeting summary can be reviewed by the adviser who attended. Fully autonomous recommendations combine too many assumptions and consequences for an immature control environment.

Management should measure corrected errors, processing time and downstream rework, not only demonstrations of speed. If staff must repeatedly repair an automated output, the system has moved effort rather than removed it. Productivity means a lower total cost for a dependable outcome.

Specialization and outsourcing reshape the value chain

Cost pressure and wider demand encourage firms to decide which capabilities they truly need to own. Portfolio construction can be delegated to a discretionary fund manager, platform administration can be purchased, and specialist retirement modelling can be centralized. Advisers then concentrate on client discovery, planning choices and continuing relationships.

Christian Markwick, head of adviser support at The Verve Group, connected changing retirement needs with the pressure to rethink service. Douglas Kearney, investment and finance director at Intelligent Pensions, described technology as a way to create consistency and capacity. Their perspectives point to an operating ecosystem rather than a single all-purpose adviser.

Outsourcing does not outsource responsibility. A firm must understand a supplier's mandate, charges, conflicts, resilience and data controls. It must monitor whether the service remains appropriate for its clients. Standard portfolios can improve consistency, but they cannot replace household-level decisions about income timing, reserves and guarantees.

The commercial advantage comes from clear interfaces. Each participant knows which data it receives, which decision it owns, how exceptions are escalated and how performance is reviewed. Poorly defined outsourcing merely adds handoffs. Well-defined specialization allows expertise and systems to be shared across more clients.

Segmentation should change service, not fairness

A scalable firm cannot provide the same meeting frequency and analysis depth to every client regardless of need. Segmentation is therefore necessary, but asset value alone is a weak guide. Complexity, vulnerability, income dependence, guarantees and proximity to irreversible decisions can matter more than wealth.

A lower-complexity segment may receive digital data collection, standardized scenarios and scheduled access to an adviser. A complex segment may require specialist modelling and more frequent reviews. Both should receive a service designed to produce an appropriate outcome, with transparent charges and an escalation route when circumstances change.

Fair value analysis should include the cost to serve, the benefits clients actually use and the risks the service manages. Cross-subsidy can be legitimate, but it should be understood. A firm that promises intensive support to every customer while pricing only by assets may eventually ration attention informally, which is worse than an explicit service design.

Capacity has to be measured end to end

Adding a planning tool can make one calculation faster while leaving the whole case unchanged. Provider statements may still arrive late, administrators may rekey values, advisers may correct incomplete fact finds and compliance teams may review inconsistent files. Local automation is not the same as system capacity.

Useful measures follow a case from first request to implemented plan: elapsed time, active staff time, waiting time, first-time completeness, rework, exception rate and review completion. Firms should separate simple and complex cases so averages do not hide a failing segment. Client understanding and outcome indicators belong beside efficiency measures.

These metrics also support investment choices. If statement collection dominates delay, integration and provider connectivity may matter more than a new forecasting interface. If adviser rework is high, templates or training may be the answer. If specialist escalation is the constraint, centralizing expertise can increase throughput.

A staged transition reduces operational risk

A firm does not need to replace its entire advice stack at once. It can begin by mapping one retirement journey, removing duplicate data fields and defining mandatory controls. The next stage can automate document intake and scenario preparation while retaining the existing approval process. Only after error rates and outcomes are understood should the business expand automation.

Four stages for management

1. Define client segments, service promises and the decisions that require human judgment.
2. Create a governed data model and a single evidence trail for each case.
3. Automate bounded tasks, measure corrections and resolve the largest bottleneck.
4. Expand carefully, monitoring client outcomes, fair value and operational resilience.

Staff involvement is essential. Administrators know where data fails, advisers know where conversation changes a decision, and compliance teams know which evidence is repeatedly missing. A design imposed without those perspectives may look efficient on a process chart but generate workarounds in practice.

Clients also need an explanation of the hybrid model. They should know when a calculation is automated, where an adviser exercises judgment and how to correct inaccurate data. Transparency can turn technology from a source of suspicion into visible quality control.

The new model standardizes process and personalizes decisions

The retirement market in the United Kingdom is not simply moving from human advice to software. It is dividing work more intelligently. Machines can organize records, repeat calculations, enforce checkpoints and keep monitoring schedules. People can interpret priorities, explain uncertainty and take responsibility for consequential recommendations.

That division can improve access only if economics and governance are designed together. Smaller, fragmented pots require lower production cost. Consumer protection requires reliable controls and understandable communication. Personalized retirement paths require flexibility. Optimizing one requirement while ignoring the others produces either an unaffordable service or a scalable but weak one.

The research reported in 2024 captured a broad recognition that change was overdue. The durable response is not a fashionable tool or a new label for the same portfolio. It is an operating model in which data, workflow, specialists, suppliers and advisers form one accountable system.

For business leaders, the practical test is straightforward: can the firm serve one more appropriate client without adding the same amount of manual work or reducing the quality of judgment? If the answer improves while evidence, resilience and client understanding remain strong, automation is creating real capacity. That is how retirement advice can scale without becoming impersonal.