How to Measure Brand Trust for a Bank or Insurer

By Yamm Labs · Updated October 2026 · 13 min read

The quarterly board pack has a slide called “Brand health”. Awareness is up two points, consideration is flat, and a green bar labelled “trust” sits beside the logo. The chief financial officer asks what the trust bar measures, whether it moved because of the new campaign or despite the app outage in August, and how the question was worded. The room goes quiet, because the tracker was bought as a black box. If you have to measure brand trust for a bank or insurer and then defend the number, the work starts with the definition and the data plan, long before anyone fields a survey.

What you need before you start to measure brand trust

You need three owners, two approvals and four sets of inputs before the first question is written. Brand trust is a customer’s willingness to rely on your firm with their money, data or claim at a moment when they cannot check the outcome for themselves. That definition matters because it tells you what to ask and which behaviour to watch.

Name the owners first:

1. The brand or marketing head owns the study, the questionnaire and the board slide.
2. The insights or analytics lead owns the sample, the fieldwork vendor and the scoring method.
3. The data protection officer, or whoever holds that role in your firm, owns the consent and retention decisions.

Two approvals come before fieldwork. Compliance signs the questionnaire and any customer contact, and IT security signs any new tool that touches customer data, whether a survey platform, a panel vendor or a listening tool. Procurement often joins at this stage, so ask early whether a vendor already on the approved list can do the work.

Collect four inputs: any past tracker data with the exact question wording, complaint and grievance data from the service team, app store ratings and review text for the last year, and a log of events that could move trust, such as outages, fraud alerts, fee changes, leadership news and campaign launches. The event log is the input teams forget, and it is the one that explains most surprising movements later.

Trust you cannot define in one sentence is trust you cannot measure in one quarter.

Phase 1: trust has to be split into parts before it can be scored

If you measure brand trust with a single “Do you trust this brand?” question, you get a number with no lever attached. When it falls, nobody knows whether customers doubt your competence, your honesty or your intentions, and each problem has a different fix.

Management researchers Roger Mayer, James Davis and David Schoorman proposed in 1995 that trustworthiness rests on three factors: ability, benevolence and integrity. The split maps well onto financial services, where a customer can believe a bank is competent and still suspect it of selling them a product they do not need. We use plainer labels in questionnaires: ability, integrity and goodwill.

1. Write two statements for each part, phrased the way an Indian customer would say them, and ask for agreement on a five-point scale.
2. Add one overall reliance question, such as “I would keep my salary account here if a friend told me to move”.
3. Add one likelihood-to-recommend question on a 0 to 10 scale, so you can compare with service surveys the bank already runs.
4. Add one behavioural intent question tied to a real decision, such as renewing a policy or adding a nominee.
5. Freeze the wording. Change one word and the trend line breaks.

Test the statements on eight to ten customers in conversation before fieldwork. If a phrase confuses people in Hindi or a regional language, it will produce noise in every wave that follows.

SIGNALNewsletter by Yamm Labs

The next piece, straight to your inbox.

New writing on BFSI brands, B2B and influencer marketing, and AI search.

Email *

Subscribe

Subscription form is not available at the moment

One click to unsubscribe. Privacy policy

Phase 2: a survey baseline only counts if the sample matches your book

The baseline survey has to reflect the customers whose trust pays your bills, and the prospects you want to win. A national panel weighted to the general population will tell you how urban, online respondents feel, which may have little to do with a cooperative bank’s depositors or a term insurer’s policyholders.

1. Split the sample into customers and non-customers, and report them separately. Customers judge you on experience; prospects judge you on reputation.
2. Within customers, set quotas for the segments that matter: new to bank, digital-only, branch-led, claimants or borrowers in repayment.
3. Field in the languages your customers use. An English-only questionnaire excludes many of the people a regional NBFC serves.
4. Add the two or three competitors a customer would compare you with, and ask the same statements about each.
5. Set the frequency. Quarterly suits a brand in the middle of a campaign or a rebrand; half-yearly suits a steady incumbent. Both are typical practice; no regulator sets them.

Choose a syndicated tracker if you need peer comparison quickly and can live with the vendor’s wording. Choose a custom study if you need the three-part split, your own segments and the right to see the raw data. Choose an in-app pulse if you want a fast signal from active users, and accept that it will never hear from the customers who stopped opening the app.

Write down the sample design and the margin of error the vendor states. When the CFO asks whether a two-point rise is real, that note is the answer.

Phase 3: behavioural and complaint signals show trust that surveys miss

People say they trust a bank and then move their money, or say they distrust an insurer and renew anyway. To measure brand trust properly, treat behaviour as the check on the survey, and in Indian financial services some of it is published by the regulators.

The RBI’s annual report on its Ombudsman Scheme lists complaints by entity type and category. The report for 2023-24 counted 9,34,355 complaints received across its ombudsman offices and central processing centre, with banks accounting for 82.28 per cent of those handled by the ombudsman offices and loans and advances the largest category at 29.01 per cent. You can read the RBI’s Ombudsman Scheme annual report for 2023-24 and track how your group and category move against the total. For insurers, IRDAI runs the Bima Bharosa grievance portal, where policyholders register and track complaints. SEBI’s SCORES platform does the same job for investors in listed companies and market intermediaries.

Build a behavioural panel from signals you already own or can see:

• Complaints per lakh customers. Your own grievance data, normalised by book size, with a note of what share escalated to a regulator.
• Retention after a bad week. Deposit balances, policy renewals or fund redemptions in the 30 days after an outage, a fraud story or a fee change.
• App store ratings and review text. The rating moves slowly; the words “fraud”, “blocked”, “hidden charges” and “no response” move fast.
• Branded search with doubt attached. Searches for your name with “safe”, “fraud” or “complaint”, read from your search console and keyword tools.
• Claims and service escalations. For an insurer, repeat calls about a single claim are a goodwill signal long before a complaint is filed.

Keep the behavioural panel on the same quarterly calendar as the survey, so the two can be read side by side.

Phase 4: the consent and data plan decides which signals you may use

Every source in the first three phases touches personal data, so the consent plan is part of the measurement design. India’s Digital Personal Data Protection Rules were notified on 14 November 2025 and give firms an eighteen-month period for phased compliance, according to the government’s official backgrounder on the DPDP Rules. The same document describes consent that must be clear and informed, a separate notice explaining the specific purpose, and the right to withdraw consent at any time. Your compliance team will read the trust study against those principles, so design for them now.

These are the decisions we would put in writing before fieldwork:

1. Survey invitations to customers go out only to people whose contact consent covers research, through a service channel, with a one-line purpose statement and an opt-out in the same message.
2. Respondent identity stays with the fieldwork vendor or a single analyst. The trust dataset carries a pseudonymous ID, never a phone number, email, PAN, account number or policy number.
3. Web and app analytics tags that feed the behavioural panel fire only after the visitor accepts them, and the event set strips personal data from URLs, form fields and search terms before anything leaves the device.
4. Raw survey answers and verbatims are deleted on a fixed date set by your retention policy, while aggregated scores are kept for trend analysis. Write the date into the vendor contract.
5. Social listening stays at the level of public posts and aggregate counts. Nobody matches a complaining handle to a customer record for the study.

Expect IT security and compliance to ask the same five questions about any tool: where the data is stored and in which country, who the sub-processors are, whether the vendor can see identities, how a respondent withdraws, and how the vendor reports a breach. Answer them on the data-flow sheet and the tool review moves faster. This is general guidance, not legal advice; your compliance team decides how the rules apply to your firm.

One more decision belongs here. If marketing later wants to say “most trusted insurer” in an advertisement, ASCI’s code expects claims of that kind to be backed by evidence, so keep the method, sample and wording on file from the first wave. For an insurer, IRDAI’s advertising rules also apply to the finished advertisement itself.

Phase 5: the board should see one trust score with its drivers

Report a single trust index, its three parts and the behavioural panel on one slide, with the event log underneath. A number without its drivers invites the question you could not answer in the opening meeting.

Trust index = (ability score + integrity score + goodwill score) ÷ 3, each converted to 0 to 100 from the share of customers who agree or strongly agree.

Take a hypothetical private bank. In the baseline wave, 72 per cent agree with the ability statements, 58 per cent with the integrity statements and 49 per cent with the goodwill statements. The index is (72 + 58 + 49) ÷ 3, which is 59.7. The headline looks acceptable. The parts say the bank is seen as competent and not on the customer’s side, which points at sales conduct and grievance handling, not at the advertising budget.

Three reporting rules keep the score honest:

1. Show the margin of error on the slide and call a movement real only when it clears it.
2. Read the survey next to the behavioural panel. If trust rises and complaints per lakh customers rise too, check the sample before you celebrate.
3. Do not credit a campaign with a trust gain unless you held out a region or segment that did not see it. Our guide to marketing measurement for financial brands covers how to set up that comparison.

Brand trust measurement plan: steps, owners and time

The table sets out a first cycle to measure brand trust. Timings are typical practice for a mid-size bank or insurer, not rules from any regulator, and they stretch when a new vendor needs a full security review.

How long it takes to measure brand trust, and who signs off

A first cycle typically runs ten to twelve weeks from kick-off to board slide, and later waves take four to six weeks because the questionnaire, consent plan and vendor are already approved. The longest single stretch is usually the security review of a new survey or listening tool.

Sign-off runs in a fixed order. The brand head approves the definition and statements, compliance approves the questionnaire and any customer contact, the data protection officer and IT security approve the data flow and the vendor, and the brand head signs the final report. If the trust index will appear in an investor presentation or an annual report, the investor relations or company secretarial team should see the method before it is published.

When a rebrand is part of the plan, run the baseline before the new identity launches, or you lose the before-and-after comparison. Our piece on how to rebrand a bank without losing trust sets out where that baseline fits in the rebrand timeline.

How to tell your brand trust measurement worked

The measurement works when it changes a decision, and the board can say which one. A tracker that produces a number every quarter and changes nothing is a subscription.

Look for four signs after the second wave:

1. The survey and the behavioural panel move in the same direction most quarters, and you can explain the quarters when they do not.
2. A low part of the index has led to a named fix, such as a rewritten charges screen, a new claims update message or a branch sales script, with an owner and a date.
3. The trust index in one quarter tells you something about renewals, balances or complaints in the next. If it never does, the statements are measuring mood.
4. Finance has stopped asking what the number means and started asking why it moved.

If none of these are true after two waves, the problem is usually the definition or the sample, and you should go back to phase 1 or phase 2.

What to do when a step fails

Most failures when you measure brand trust are fixable within one wave if you catch them early. These are the ones we see most often in BFSI work, with the fix for each:

• Response rates are too low among customers. Move the invitation to the channel customers already use for service, such as the app inbox or a WhatsApp business message, shorten the survey to under ten minutes, and keep the purpose line plain.
• Compliance blocks a vendor. Ask which of the five data questions failed. Often it is data storage location or sub-processors, and a vendor already on the approved list solves it.
• The score jumps with no explanation. Check the event log, then the sample mix. A wave that over-sampled digital-only customers will look different from one heavy on branch customers.
• Trust falls after an incident. Report it in full. Track goodwill and integrity weekly through an in-app pulse for six weeks, and fix the operational cause before any reassurance campaign.
• Marketing wants to quote the score in an advertisement. Send the method, sample and wording to compliance first, and treat the line as a claim that needs evidence on file.

For the analytics set-up behind the behavioural panel, including consent-gated tags and server-side event filtering, see our marketing analytics and martech services.

People also ask

What is the best way to measure brand trust for a bank?

Use a survey that splits trust into ability, integrity and goodwill, and read it next to behavioural signals such as complaints per lakh customers, retention after incidents and app review text. A single trust question cannot tell you what to fix.

How often should an insurer measure brand trust?

Quarterly suits an insurer in the middle of a campaign, a rebrand or a claims-service change, and half-yearly suits a steady period. No regulator sets either interval; it is typical practice, so pick one and keep the wording frozen.

Can we use customer data from our CRM for a trust survey?

Only within the consent the customer gave and the purpose you told them about. Keep phone numbers, PAN and policy or account numbers out of the trust dataset, use a pseudonymous ID, and agree a deletion date for raw answers with your data protection officer.

Is likelihood to recommend the same as brand trust?

No. A customer may recommend a bank for its app and still doubt it would treat them fairly in a dispute. Keep the recommend question for comparison with service surveys, and score trust from its three parts.