Financial services is the most heavily researched industry on earth. Banks, insurers, and wealth managers spend billions annually on consumer research, market analysis, and product testing. They also operate under the most stringent data protection regimes of any sector, which means that the more they need to understand their customers, the harder it becomes to actually ask them anything. This is, to put it charitably, a structural problem. Synthetic research may be the first methodology that makes the problem easier rather than more expensive.
The Industry That Knows Everything and Understands Nothing
There is a peculiar paradox at the heart of financial services research. No industry collects more behavioural data per customer. Every transaction, every login, every abandoned mortgage application, every time someone hovers over the "close account" button generates a data point. The average retail bank holds more behavioural information about its customers than most consumer brands could dream of possessing. And yet, when it comes to understanding why customers behave the way they do, what they actually want from their financial products, or how they would respond to a new offering, the industry remains remarkably dependent on the same research methodologies it was using in 1995.
The reason is not incompetence. It is compliance.
Financial services firms operate under regulatory frameworks that treat customer data with a severity that other industries would find bewildering. GDPR, the California Consumer Privacy Act, the Gramm-Leach-Bliley Act, PSD2, and their various national equivalents create a labyrinth of restrictions on how customer data can be collected, stored, processed, shared, and used for purposes beyond the original transaction. The compliance burden is not theoretical. Fines for data protection violations in financial services have exceeded $5 billion globally since 2018, and the reputational consequences of a breach can dwarf the regulatory penalty.
The practical effect is that the industry with the most data is often the least able to use it for research purposes. Customer journey analysis requires anonymisation protocols that strip out the contextual detail that makes the analysis useful. Focus groups with real customers trigger data processing agreements, consent workflows, and legal reviews that can delay a study by months. Even something as simple as surveying customers about their satisfaction with a new app feature requires navigating a compliance process designed to prevent the misuse of financial data, not to facilitate product improvement.
The result is a research bottleneck that has become so normalised that most financial services professionals no longer recognise it as abnormal. Product teams at fintechs wait weeks for research that would take days in consumer goods. Insurance companies test new policy structures with internal stakeholders rather than representative customers because the compliance overhead of external research exceeds the project timeline. Wealth management firms rely on relationship managers' anecdotal impressions of client preferences because structured research at scale is prohibitively slow.
Synthetic research does not merely offer a faster alternative. It sidesteps the compliance constraint entirely. When your research participants are AI-generated personas rather than real individuals, there is no personally identifiable information to protect, no consent to obtain, no data processing agreement to negotiate, and no risk of a breach. The research is, by construction, compliant. This is not a marginal advantage. For an industry where compliance is the binding constraint on research velocity, it is transformative.
What Financial Services Firms Actually Need to Test
The use cases for synthetic research in financial services are broader than most practitioners initially assume. The instinct is to think of it as a substitute for customer satisfaction surveys. It is that, but it is also considerably more.
Customer Journey Testing
The financial services customer journey is, by any reasonable standard, among the most complex in any consumer-facing industry. Opening a bank account involves identity verification, credit checks, product selection, terms acceptance, and onboarding. Applying for a mortgage involves all of the above plus income documentation, property valuation, affordability assessment, and a decision timeline measured in weeks. Each step in these journeys represents a potential abandonment point, and the cost of abandonment in financial services is high because the customer acquisition cost is high.
Synthetic research allows product teams to test proposed journey changes before committing to development. A neobank considering a redesigned onboarding flow can simulate how different customer segments would respond to each screen, each form field, each piece of required documentation. An insurer redesigning its claims process can test whether a simplified digital workflow would satisfy customers or whether the perceived reduction in human contact would erode trust at precisely the moment when trust matters most.
The critical advantage is iteration speed. A traditional user testing programme for a mortgage application redesign might involve recruiting 30 participants, scheduling sessions over three weeks, conducting the tests, analysing the recordings, and presenting findings. The synthetic equivalent takes minutes and can be repeated with different demographic segments, different journey variants, and different competitive contexts without additional recruitment.
Pricing Page Optimisation
Pricing in financial services is where rationality goes to die. Consumers who will spend twenty minutes comparing the unit price of washing-up liquid will accept a mortgage rate without checking whether a competitor offers 15 basis points less. They will scrutinise the annual fee on a credit card while ignoring the foreign exchange markup that costs them three times as much over the course of a year. The gap between how financial services firms present pricing and how consumers actually process pricing information is vast, and it represents one of the largest untapped opportunities for conversion optimisation in the industry.
Synthetic research is particularly well suited to pricing page testing because it can simulate the cognitive and emotional responses that drive financial decision-making. How do consumers react when they see a monthly fee expressed as a daily cost? Does displaying the total cost of a loan over its lifetime increase or decrease application intent? When comparing two savings accounts, do consumers focus on the headline rate, the AER, or the terms and conditions around withdrawal penalties?
These are questions that A/B testing can answer eventually, with enough traffic and enough time. Synthetic research answers them before the page is built.
Financial Product Messaging
The challenge of financial product messaging is that you are, in most cases, trying to make something inherently dull sound appealing without making it sound frivolous. Nobody wakes up excited about a new current account. Nobody shares an Instagram story about their pension contributions. The emotional register available to financial services marketers is narrow, and the compliance constraints on what they can claim are strict.
Synthetic research helps by testing how different messaging frameworks land with specific audiences. Does a millennial audience respond better to a savings product positioned around "financial freedom" or "building security"? Does an affluent retiree react more positively to wealth management messaging that emphasises growth or preservation? When a fintech describes itself as "the bank that works for you," does that register as empowering or patronising?
The value here is not just in finding the right message but in understanding why it works. Synthetic personas provide qualitative reasoning alongside their responses, explaining the associations, concerns, and motivations that shape their reaction. This explanatory depth is what transforms message testing from a binary pass/fail exercise into a genuine insight about how your audience thinks about money.
Compliance-Safe Research
This deserves its own section because it is, for many financial services firms, the single most compelling reason to adopt synthetic research.
The compliance challenge in financial services research is not merely about data protection. It extends to the research content itself. Financial promotions regulations in the UK, SEC marketing rules in the US, and their equivalents elsewhere impose strict requirements on how financial products can be described, what claims can be made, and how risk must be communicated. Testing marketing materials with real consumers can itself trigger regulatory scrutiny if the materials are deemed to constitute a financial promotion shown to the public before approval.
Synthetic research eliminates this concern. Testing a draft advertisement with AI personas is not a financial promotion. There is no "public" to whom the material is being communicated. The compliance team does not need to review research stimuli for regulatory compliance because the stimuli are being shown to software, not people. This may seem like a technicality, but anyone who has experienced a three-week delay while compliance reviews a set of focus group discussion guides will appreciate its practical significance.
Financial Literacy Testing
A use case that is underappreciated but genuinely important. Financial services firms have both a regulatory obligation and a commercial interest in ensuring that their customers understand the products they are buying. Mis-selling scandals, from payment protection insurance in the UK to subprime mortgages in the US, have their roots in a gap between what firms believed they were communicating and what customers actually understood.
Synthetic research can test whether product descriptions, terms and conditions, and disclosure documents are comprehensible to their intended audience. Not comprehensible to a compliance officer or a product manager, but to someone with average financial literacy who is trying to decide whether to take out a personal loan. The ability to simulate comprehension across different education levels, age groups, and financial experience profiles is something that traditional research can do in theory but rarely does in practice because of the cost and time involved.
The Competitive Landscape: Enterprise Versus Accessible
The synthetic research market in financial services has, until recently, been dominated by platforms designed for enterprise buyers. This is unsurprising. Financial services is an enterprise industry, and the vendors who arrived first built products that matched the procurement expectations of their target customers: long sales cycles, custom implementations, opaque pricing, and dedicated account management.
Simile is perhaps the most prominent example of this approach in the financial services context. The platform lists Wealthfront and Banco Itau among its customers, both significant names in wealth management and retail banking respectively. Wealthfront, the automated investment service with over $50 billion in assets under management, uses synthetic consumer research to test product positioning and customer communication. Banco Itau, the largest bank in Latin America by market capitalisation, represents the kind of enterprise relationship that validates a platform's credibility in the sector.
Simile's model is enterprise-first: no public pricing, no self-serve access, no published API. If you want to use Simile, you speak to a sales team, negotiate a contract, and receive a managed research experience. For a firm like Banco Itau, with dedicated research budgets and procurement processes built for this kind of engagement, the model works well. For a Series A fintech with twelve employees and a product manager who wants to test three onboarding variants before Friday, it is not a realistic option.
Aaru occupies a similar position, with its partnership with EY providing an enterprise distribution channel that lends institutional credibility. The EY relationship is notable because it signals that the established consulting firms are incorporating synthetic research into their advisory offerings, which in turn accelerates adoption among the consulting firms' financial services clients. When your EY engagement manager recommends a synthetic research platform, the internal approval process is considerably smoother than when your product manager discovers one on Twitter.
The enterprise model has genuine advantages for large financial services firms. Dedicated support, custom model training, integration with existing research workflows, and the comfort of a named account team all matter when the buyer is a regulated institution with specific requirements around data handling, audit trails, and vendor risk assessment.
But the enterprise model also has a structural limitation: it excludes the vast majority of the financial services industry. There are approximately 10,000 banks in the United States alone, most of them community banks and credit unions with no dedicated research function and no budget for an enterprise synthetic research contract. There are thousands of fintechs, insurtechs, and wealthtechs at various stages of growth, most of which need customer insight more urgently than the large institutions and have less capacity to obtain it through traditional channels. There are independent financial advisers, mortgage brokers, and insurance agents who would benefit enormously from understanding how their clients perceive their services but who have never had access to any form of structured research.
This is where the accessible end of the market becomes relevant. FishDog offers a self-serve platform with transparent pricing, a full REST API, and results delivered in minutes rather than days. The relevance for financial services is not just cost, though cost matters. It is the combination of speed, accessibility, and the compliance advantage described above. A product manager at a neobank can design a study, recruit synthetic personas filtered by demographics and financial behaviour, run the study, and read the results during a single work session. No procurement process. No vendor risk assessment. No three-month sales cycle.
The API capability is particularly significant for financial services firms that want to integrate research into their product development workflow rather than treating it as a separate, periodic activity. A fintech building a new savings product can programmatically test messaging variants, onboarding flows, and pricing presentations as part of their sprint cycle, with research results feeding directly into product decisions rather than arriving in a PDF three weeks after the decision was made.
The Compliance Advantage, Examined Honestly
It would be irresponsible to discuss synthetic research in financial services without addressing the compliance dimension with some rigour, because the advantage is real but not without nuance.
The core proposition is straightforward: synthetic research involves no real individuals, therefore no personally identifiable information is collected, processed, or stored. This eliminates the most significant compliance constraints on financial services research. No data subject access requests. No right-to-erasure obligations. No data processing agreements with third-party recruiters. No consent management. No risk of a data breach involving customer information, because no customer information is involved.
For firms operating under GDPR, this is not a trivial benefit. The regulation requires a lawful basis for processing personal data, and "market research" is not automatically a sufficient basis when the data subjects are customers of a regulated financial institution. The legitimate interest assessment, the data protection impact assessment, and the associated documentation requirements create a compliance overhead that can double the timeline of a traditional research project. Synthetic research bypasses all of it.
For firms subject to banking secrecy laws, the advantage is even more pronounced. In jurisdictions where the mere disclosure that someone is a customer of a particular bank constitutes a breach of banking secrecy, conducting external research with customers requires legal gymnastics that most product teams would rather avoid. Synthetic research requires no gymnastics at all.
The nuance is this: the compliance advantage applies to the research methodology, not to the application of its findings. If a synthetic research study reveals that customers would respond positively to a new product feature, the subsequent development and marketing of that feature remain subject to all the usual regulatory requirements. Synthetic research does not create a compliance-free zone for product development. It creates a compliance-free zone for the research that informs product development, which is a meaningful but bounded advantage.
There is also a question of regulatory acceptance. Financial regulators have not, to date, issued formal guidance on the use of synthetic research in place of traditional consumer testing for regulatory purposes. If a regulator requires evidence that customers understand a product's risk profile before it can be marketed, it is not yet clear whether evidence derived from synthetic personas would satisfy that requirement. The prudent approach, for now, is to treat synthetic research as a complement to, rather than a replacement for, the specific forms of consumer testing that regulators explicitly require. For all other research purposes, which constitute the vast majority of the research a financial services firm conducts, the compliance advantage is unambiguous.
Who Benefits Most, and Why the Answer Is Not Obvious
The intuitive assumption is that the largest financial services firms would benefit most from synthetic research, because they have the most complex products, the most stringent compliance requirements, and the largest research budgets to redirect. The intuition is wrong, or at least incomplete.
The firms that benefit most are those where the gap between research need and research capacity is largest. In large banks and insurers, dedicated research teams, established vendor relationships, and substantial budgets mean that research happens, even if it happens slowly. The improvement from synthetic research is incremental: faster, cheaper, less compliance friction, but additive to an existing capability.
In mid-market financial services firms, fintechs, and specialist providers, the gap between need and capacity is often absolute rather than relative. These firms do not have a research function that synthetic research would accelerate. They have no research function at all. Their product decisions are informed by internal opinion, competitor observation, and the occasional conversation with a friendly customer. Synthetic research does not improve their research capability. It creates it from nothing.
A community bank in Ohio considering whether to launch a mobile banking app. An insurtech in London testing whether its claims process is comprehensible to non-native English speakers. A wealth management startup in Toronto wondering whether its robo-advisory messaging resonates with millennials who distrust traditional finance. A mortgage broker in Sydney trying to understand why clients abandon applications at the income verification stage. None of these firms have ever commissioned a formal research study. All of them make decisions daily that would benefit from structured consumer insight. Synthetic research, at an accessible price point and with minutes rather than months of turnaround, brings them into the research economy for the first time.
There is a second category of beneficiary that is worth noting: the individual product manager or marketing director within a large institution who wants to run research without engaging the formal research procurement process. In a large bank, commissioning a research study through official channels can involve vendor selection, procurement review, legal review, compliance review, budget approval, and a timeline measured in quarters. Synthetic research, accessed via a self-serve platform and paid for on a credit card, allows the individual to answer their question now and present findings to their colleagues as a fait accompli rather than a proposal. This is not how enterprise procurement is supposed to work, but it is how enterprise innovation frequently does work.
A Realistic Assessment of Limitations
Intellectual honesty demands a discussion of what synthetic research cannot do in financial services, because the limitations are real and pretending otherwise would undermine the credibility of the genuine advantages.
Synthetic personas are calibrated on population-level data. They represent how people in a given demographic and psychographic segment tend to think, feel, and behave. They do not represent how your specific customers think, feel, and behave. For a bank with ten million retail customers, the synthetic approximation is likely to be close. For a wealth management firm with 200 high-net-worth clients, each with idiosyncratic preferences shaped by decades of personal financial history, the approximation may be less useful.
Synthetic research also struggles with products that are genuinely novel. When testing messaging for a new current account or a redesigned insurance claims process, the synthetic personas have sufficient context to generate meaningful responses because these are products and processes that exist in the training data. When testing a genuinely unprecedented financial product, one that consumers have no frame of reference for, the synthetic responses may default to generic heuristics rather than the nuanced, sometimes irrational reactions that real consumers exhibit when confronted with the unfamiliar.
Regulatory-mandated consumer testing, as noted above, may not be satisfiable with synthetic data. Where a regulator requires evidence that real consumers understood a product disclosure, synthetic evidence is not, at present, a substitute.
Finally, the emotional dimension of financial decision-making is difficult to simulate fully. The anxiety of taking on a large mortgage, the grief-tinged complexity of claiming on a life insurance policy, the excitement and terror of a first investment: these emotional states shape consumer behaviour in ways that demographic and psychographic calibration captures imperfectly. Synthetic research is excellent at simulating considered responses to financial products. It is less reliable at simulating the irrational, emotionally charged responses that drive some of the most consequential financial decisions.
These limitations do not invalidate synthetic research for financial services. They define its appropriate scope. The vast majority of financial services research questions fall comfortably within that scope. The minority that do not are generally identifiable in advance.
Where This Goes Next
The trajectory is reasonably clear, even if the timeline is not.
Financial services firms will adopt synthetic research first for use cases where the compliance advantage is most acute: product messaging testing, journey optimisation, pricing research, and internal concept validation. These applications involve no regulatory-mandated consumer testing requirements, no edge cases around novel products, and no emotional complexity that strains the synthetic methodology. They are also, not coincidentally, the applications where traditional research is slowest and most expensive relative to the value of the insight produced.
As adoption increases and the methodology proves its reliability within the sector, the scope will expand. Regulators will eventually issue guidance on synthetic research, and that guidance will likely permit its use for a broader range of purposes than the current ambiguity allows. Enterprise platforms will build financial-services-specific features: compliance audit trails, regulatory reporting integrations, and synthetic persona calibrations trained on financial behaviour data. Self-serve platforms will become the default tool for the thousands of mid-market financial services firms that currently conduct no research at all.
The firms that move early will have an advantage that compounds over time. Not merely the advantage of better-informed decisions, though that is real. The advantage of building an organisational muscle, a habit of testing assumptions with data rather than opinions, a culture where "let us run a quick study" is as natural a response to uncertainty as "let us schedule a meeting." In an industry where the difference between a well-designed product and a poorly designed one can be measured in billions of dollars of customer assets gained or lost, the research habit is not a nice-to-have. It is a competitive necessity.
The industry that knows everything about its customers' transactions is finally acquiring the tools to understand their motivations. The irony is that it took artificial consumers to make that possible.
Full disclosure: Phillip Gales is co-founder at [FishDog](https://fish.dog), a synthetic market research platform that competes in the market described in this article. His analysis is informed by direct experience but shaped by commercial interests that the reader should weigh accordingly. Competitors including Simile and Aaru are represented based on publicly available information, and readers evaluating platforms for financial services research should conduct independent due diligence.