AI-powered market research for the financial sector: how Bradesco Seguros uses synthetic data for regulated decisions

The financial sector has more customer data than any other industry, but uses less of that data for marketing decisions because every research project with real personal data requires months of regulatory approval. Synthetic data solves this paradox: Bradesco Seguros validated 600 respondents in 48 hours.

The financial-sector paradox: more data, fewer data-driven decisions

A retail bank has behavioral data from millions of customers. It knows what each customer buys, where they buy, how much they spend, when they pay bills, and which credit products they hire. In data volume, it is the richest sector in the market.

And yet marketing decisions in financial institutions rarely start from data validated with consumers. They start from historical product-behavior analyses and market intuition calibrated by years of experience.

The reason is paradoxical: precisely because the sector holds so much sensitive customer data, it faces the most restrictions on using that data for research. Each research project involving personal customer data requires legal, DPO, and sometimes SUSEP or Central Bank compliance approval.

Synthetic data eliminates this paradox.

What are the specific challenges of market research in the financial sector?

Why is traditional market research so difficult in financial services?

Four structural blockers: LGPD and Central Bank regulation restrict collection and use of customer data for research; legal approval cycles add weeks; product-decision timelines do not match traditional research timelines; and niche audiences are difficult to recruit.

Blocker 1: LGPD plus sector regulation

LGPD already requires documented legal basis, adequate consent, and DPAs with research suppliers. In financial services, Central Bank and SUSEP rules add additional layers for account-holder and insured-customer data. Each study with real customer data becomes a legal project before it becomes a research project.

Blocker 2: approval cycles that do not match decision speed

A new credit or insurance product may need positioning validation in three weeks to meet product launch calendars. Approval for research using customer data can take four to eight weeks. The positioning decision often reaches market without external validation.

Blocker 3: niche audiences are hard to recruit

Researching private-banking clients, insured people in complex life products, or SME-credit takers requires specialized recruitment that may take longer than the decision cycle allows. These profiles are rarely available in conventional research panels.

Blocker 4: fintechs move faster

Fintechs do not carry the same legacy structure as traditional banks, but enter the same markets with products validated in weekly cycles. Established institutions that need months to validate a product lose market timing to smaller and more agile competitors.

How Bradesco Seguros uses Synthetic Personas

How do insurers use Synthetic Personas for marketing decisions?

With synthetic data that preserves the statistical properties of real data without using customers’ personal information. Bradesco Seguros generated 600 Synthetic Personas in 48 hours, with R$1.20 per Persona versus R$150 to R$300 per qualitative interview, and a launch 10.5 times faster.

Bradesco Seguros needed to validate the positioning of a new product before committing launch budget. The traditional qualitative-research process, including legal approval and fieldwork with real insured customers, would take three to six months.

With Galaxies’ Synthetic Personas, positioning was validated with 600 synthetic respondents in 48 hours. Cost per respondent fell from R$150 to R$300 per interview to R$1.20 per Persona. The responsible director documented “60 times more qualitative responses in a quarter of the expected time.”

The launch happened 10.5 times faster than the previous cycle, with full LGPD compliance and no personal data from real insured customers in the process.

Why Synthetic Personas are especially suited to financial services

Why is synthetic data ideal for the financial sector?

Three specific reasons: LGPD compliance by design, because synthetic data is not personal data under Article 5, I; the ability to simulate customer profiles that cannot be approached directly; and zero risk of leaking sensitive financial customer data. Galaxies is ISO 27001 certified.

• Compliance by design: synthetic data is not personal data under Article 5, I of LGPD. The output does not require additional DPA, specific legal basis, or DPO approval per project. Marketing teams access Personas without extra compliance per interaction.

• Profiles that cannot be approached directly: private-pension customers, insured people in complex life products, or SME-credit customers with specific profiles. Representative Synthetic Personas are created from the institution’s historical data without exposing individuals.

• Zero risk of sensitive-data leakage: a research-data leak in a financial institution creates serious reputational and regulatory damage. With Synthetic Personas, there is no personal customer data in the process. The risk is eliminated at the source.

Five marketing decisions in financial institutions validated with Synthetic Personas

How are Synthetic Personas used for financial-sector marketing decisions?

Five applications: positioning a new credit or insurance product before regulatory approval; testing campaign messages by income segment and investor profile; simulating customer journeys in the app before UX redesign; mapping sales objections for training briefs; and researching price acceptance for insurance products.

• Product positioning before regulatory approval: the product team validates how the market perceives a new credit product before submitting it to the Central Bank for approval. With Synthetic Personas, validation happens in 48 hours and guides adjustments before the regulatory process.

• Message testing by income segment and investor profile: the same private-pension campaign resonates differently with conservative and moderate investors. Synthetic Personas segmented by investor profile reveal which message generates more consideration in each group.

• App journey simulation before UX redesign: before committing development budget, product teams validate which navigation flow creates less friction for each user profile. Synthetic Personas reveal pain points by digital-familiarity level.

• Objection mapping for manager training: which objections does the customer raise when a manager presents an SME credit product? Synthetic Personas identify the most frequent objections by customer profile, guiding sales-force training briefs with data.

• Price-acceptance research for insurance products: what monthly premium do different insured profiles perceive as adequate? Predictive simulation by cluster reveals the indifference point and the price that discourages adoption by income profile and region.

Frequently asked questions

How do insurers use AI for consumer research?

With Synthetic Personas that preserve statistical properties of real data without using customers’ personal information. Bradesco Seguros generated 600 Personas in 48 hours, at R$1.20 per Persona versus R$150 to R$300 per interview, and launched 10.5x faster with full compliance.

Are synthetic data compliant with Central Bank and SUSEP regulations?

Yes. Synthetic data is not personal data under Article 5, I of LGPD and does not involve direct customer-data collection. The Galaxies methodology uses anonymized input data and generates Personas that do not correspond to any real individual. Galaxies is ISO 27001 certified.

How does the financial sector validate regulated-product positioning?

With Synthetic Personas representative of the product’s target audience. The team submits positioning briefs to Personas before regulatory approval and receives the signal in 48 hours: which attributes create value, which objections emerge, and which segment shows the highest intent.

How does synthetic data reduce research costs in financial institutions?

Bradesco Seguros reduced cost per respondent from R$150 to R$300 per qualitative interview to R$1.20 per Synthetic Persona. With delivery in 48 hours versus four to eight weeks of fieldwork with regulated personal data, the decision-timing impact is also significant.

Does Galaxies have security certifications suited to financial services?

Yes. Galaxies is ISO 27001 certified, the international standard for information-security management. This certification documents formal, auditable, globally recognized security controls, often required in financial-sector supplier approval processes.