Money, risk, and trust are three areas in which respondents often speak less candidly. Anyone who has designed market research in the financial sector knows that a traditional survey on bank satisfaction or insurance preferences produces a smoothed-over, rationalized picture affected by social desirability bias. The question, then, is not whether to research financial institution customers, but how to design a methodology that cuts through the layer of stated declarations.
What sets market research in the financial sector apart from research in other industries?
Finance is a low-involvement category in everyday life – most customers do not think about their personal account for weeks at a time – while also being highly emotionally involving at key decision points: taking out a mortgage, choosing life insurance, or reacting to an investment loss. This asymmetry means that bank customer research must be designed differently from FMCG or retail research. Customers may not remember how they interacted with a mobile app, but they remember exactly when they felt that an advisor did not understand them.
In addition, the financial sector is heavily regulated, which affects fieldwork. Questions about income, debt, products held, or claims history require building a particularly high level of trust and often separate information clauses. Respondents who suspect that a study is a pretext for credit scoring or sales immediately become guarded. This applies both to traditional banks and fintech companies, where there is also the question of willingness to entrust funds to an entity without physical branches.
Insurance research follows yet another logic. The product is abstract, purchased infrequently, and its value becomes apparent only when a claim occurs – an event the customer does not want to imagine. This is why stated willingness to purchase a policy correlates very weakly with actual conversion, and market research in the financial sector, particularly in insurance, requires techniques that bypass rationalization, such as conjoint analysis, pricing experiments, and ethnographic research into the decision-making process.
How should research methods be matched to the characteristics of financial services customers?
The choice of method depends on three variables: the type of decision (transactional vs. high-involvement), the sensitivity of the data, and whether the study is intended to answer “what” is happening (the scale of a phenomenon) or “why” (the decision-making mechanism). In practice, projects in the financial sector most often combine several techniques in a mixed-methods approach.
For quantitative research among bank customers, CAWI surveys using panels with well-profiled respondents work well, as do transactional studies that combine actual behavioral data – app logs and product usage data – with stated declarations. Traditional satisfaction metrics – NPS, CSAT, and CES – are a starting point, but in finance, their interpretation requires segmentation by product life cycle. A customer who has just completed onboarding assesses a bank differently from one who has filed their first complaint, and differently again from one in the fifth year of repaying a loan.
Qualitative research in finance primarily involves IDIs (in-depth interviews), and less often FGIs, because the presence of a group reinforces social desirability bias in money-related topics. In a group, respondents are more likely to claim that they “save regularly” and “compare loan offers” than to admit that they chose a bank because their brother-in-law recommended it. In fintech research, contextual interviews conducted while participants are actually using an app, as well as mobile diaries documenting real-time financial micro-decisions, are particularly valuable.
As Hume’s Institute experts point out, financial services customers rarely speak about the true motivations behind their decisions. Post hoc rationalization dominates, with the choice of cheap insurance explained by “a thorough analysis of the policy terms and conditions,” when in reality it resulted from three minutes spent on a comparison website. This is why research in this sector requires exceptional methodological precision: projective techniques, choice experiments (DCE, conjoint), analysis of purchase journeys based on behavioral data, and triangulation of stated declarations with actual behavior.
In personal finance and household budget management, ethnographic and observational techniques work well. Seeing what a couple’s conversation about spending actually looks like, how bills are sorted, and where insurance contracts are kept provides insights that no survey can deliver. In the segmentation of affluent customers and HNWIs, recruitment through referral networks and interviews conducted by moderators with an appropriate profile are often standard practice, as respondents need to view the interviewer as a peer rather than a survey interviewer.
Typical methodological approaches used by Hume’s Institute in financial services projects include:
- needs-based segmentation combined with behavioral segmentation based on transactional data;
- customer journey research that maps emotional as well as operational touchpoints;
- product concept tests using conjoint analysis – particularly valuable when designing account packages, insurance policies, or loan offers;
- UX research for banking apps and fintechs that combines task-based testing with eye-tracking and post-task interviews;
- reputation and tracking studies measuring trust in a brand on a quarterly or semiannual basis.
What mistakes most commonly occur in financial sector research?
The first and most common mistake is treating stated declarations as a proxy for behavior. In bank customer research, the question “Would you consider changing banks within the next 12 months?” produces percentages in the tens of percent, while the actual propensity to switch providers is usually much lower. Without calibrating stated declarations against real behavioral indicators, research findings can be misleading.
The second mistake is ignoring social desirability bias in financial topics. Respondents overstate their savings, understate consumer debt, and rationalize impulsive insurance purchases. In insurance research, there is particularly often a discrepancy between stated risk aversion and the actual portfolio of policies held. The remedy lies in indirect techniques: questions about “people like you,” projective scenarios, and list experiments.
The third mistake is using an inadequate sample in fintech research. Consumer panels systematically overrepresent digital users, so research into the penetration of a new payment app conducted on a CAWI panel will produce an inflated result relative to the general population. Projects concerning the adoption of financial innovations require combining sources (CAWI + CATI + offline recruitment), or at least clearly communicating the limitations of the sample.
The fourth mistake concerns the interpretation of NPS in finance. This metric works well in categories where customers have a genuine alternative and low switching costs. In retail banking, where inertia is high, a high NPS does not necessarily indicate a real willingness to recommend. This is why, in Hume’s Institute projects, NPS is always supplemented with product engagement metrics and analysis of respondents’ wording in open-ended questions.
The fifth mistake is overlooking the regulatory context and the product life cycle stage. A mortgage satisfaction study conducted three months after disbursement produces entirely different results from the same study conducted after five years of repayment, when the customer has long since rationalized the decision. Without controlling for the variable of “time since decision,” comparisons between segments are risky.
When should quantitative and qualitative methods be combined in financial services research?
In the financial sector, a mixed-methods approach is the rule rather than the exception. The sequence most commonly used by Hume’s Institute involves an exploratory qualitative phase – understanding the customer’s language, mapping barriers, and developing segmentation hypotheses – followed by quantitative validation on a representative sample. In some projects, this is followed by a third, in-depth phase – returning to qualitative research to explain unexpected quantitative findings.
Criteria worth considering when deciding whether to combine methods in personal finance projects, bank customer research, or fintech research include:
- whether the topic is sensitive or abstract for respondents – if so, a qualitative phase is essential to calibrate the language used in the questionnaire;
- whether behavioral data exist that can be triangulated with stated declarations – if so, the study should be designed as a hybrid of a survey and transactional data analysis;
- whether the objective is segmentation – purely declarative segmentations in finance are unstable, so behavioral variables should be included;
- whether the research is intended to support a design decision relating to a product, UX, or communication – in that case, the qualitative phase should both precede and follow the quantitative phase.
Regardless of the chosen configuration, fieldwork quality is crucial in the financial sector: interviewer training in sensitive questions, respondent verification – particularly in research involving affluent and SME segments – quality control of recordings and transcripts in qualitative research, and transactional data audits in behavioral projects.
Frequently asked questions
How can financial services customers be researched without social desirability bias?
Use indirect techniques rather than direct questions: projective scenarios involving “people like you,” list experiments, conjoint analysis, and choice experiments. It is also crucial to ensure full anonymity at the procedural level and, in qualitative interviews, to use a moderator whose profile creates a sense of being free from judgment. Wherever possible, stated declarations should be triangulated with behavioral data.
What should be measured in a bank satisfaction study?
A single metric such as NPS is not enough. The standard is to measure satisfaction at touchpoints – onboarding, complaint handling, remote channels, and advisors – Customer Effort Score for operational processes, and product engagement indicators such as the number of active products and login frequency. Additional value comes from segmenting results by the stage of the customer life cycle and the type of event preceding the study.
How should willingness to switch financial services providers be researched?
Declarative questions about switching intentions are significantly overstated in finance because of behavioral inertia. Better results come from analyzing switching barriers – cognitive, procedural, and relational switching costs – and testing specific offer scenarios through conjoint analysis, where respondents must make an actual choice. This can be supplemented by analyzing historical data on actual provider changes in comparable segments.
Ask about a financial sector research project
If you are planning research among customers of a bank, insurer, or fintech and would like to discuss how to select a methodology for a specific decision-making problem, contact the Hume’s Institute team – we will select an approach suited to the specifics of your category and the sensitivity of the topic.