The Experience Gap Audit: Where Strong Brands Quietly Lose Trust

Strong brands lose trust in the joins

A brand can have a clear strategy, polished identity, good products, and a customer experience that quietly contradicts all three. The losses often occur between owned surfaces: advert to landing page, sales to delivery, website to support, automated answer to human resolution, promise to returns policy.

Surveys may show generally positive sentiment while customers abandon silently, work around a broken process, or decide not to return. Analytics may show a completed purchase while fulfilment creates repeated contact. Support may close a ticket while the customer repeats the issue on another channel.

An Experience Gap audit looks at those joins. It combines the intended promise with observed behaviour, operational traces, customer language, recovery evidence, and consequence. The output is not a journey-map poster. It is a prioritised claim about where delivery separates from the promise and which repair should be tested.

Define the promise in observable terms

Begin with one high-consequence journey and the promise relevant to it. “Effortless,” “personal,” and “trusted” are not observable enough. Translate them into conditions.

If the promise is clarity, a customer should understand eligibility, price, process, and next action without contradictory explanations. If it is personal expertise, the service should recognise context and provide qualified judgement at the consequential moment. If it is speed, state which stage, under what normal conditions, and how exceptions are communicated.

Record where the promise appears: advertising, website, sales material, contract, onboarding, product interface, delivery message, support script, and policy. Differences may be legitimate, but contradictions need an owner.

The GOV.UK Service Standard asks teams to solve a whole problem for users, across organisational boundaries where necessary. It is public-service guidance rather than a commercial audit instrument. Its useful challenge is that customers should not need to understand the organisation chart to complete their task.

Map the whole service, not only the interface

The GOV.UK guidance on mapping a user’s whole problem distinguishes the user’s experience from the service landscape of systems, teams, processes, and evidence. Use both views.

On the customer line, map triggers, decisions, actions, channels, waiting, uncertainty, emotional load, and outcomes. Below it, map owners, platforms, handoffs, rules, data, queues, suppliers, and controls. Mark where the customer has to carry information between parts.

Do not map every possible branch at first. Choose a bounded task such as buying and returning a product, qualifying and starting a service, resolving a billing error, or changing an account after a life event. Include the route before the formal start and after the nominal completion.

Validate the map through observation. Workshops reveal organisational understanding and disagreement, but they do not prove what customers experience. Watch the task, inspect actual records, and follow live or recent cases with appropriate consent and privacy.

The Gap Evidence Grid

The Gap Evidence Grid uses six streams: promise, behaviour, operation, language, recovery, and consequence. A robust finding connects several streams without pretending they are interchangeable.

Promise

Collect the statements that set expectation: claims, service levels, policies, sales explanations, interface labels, and agent scripts. Mark which are contractual or regulated, which are marketing propositions, and which are informal habits.

Check recency and provenance. A promise may persist in search, email automation, partner pages, or old sales decks after the operating model changes. The gap can begin before the current website.

Behaviour

Observe customers attempting the task. Look for hesitation, backtracking, repeated entry, channel switching, abandonment, help seeking, and workarounds. Use task completion, progression, errors, and field performance as supporting evidence.

The UK service manual’s performance-data guidance connects completion and other measures to improvement and follow-up research. Its mandatory public-sector metrics do not transfer directly to every commercial organisation, but the method matters: use performance data to locate a question, then research why.

Avoid replay voyeurism. If session replay is used, apply consent and privacy controls, redact sensitive data, define a question, and restrict access. Watching many anonymous sessions without a decision protocol creates volume, not evidence.

Operation

Inspect the records produced when the service works and fails: queue time, inventory, payment state, dispatch, exceptions, rework, transfers, contact, refund, cancellation, and cost. Compare system timestamps with the customer-facing promise.

Follow a small set of cases end to end. Aggregate averages hide loops and tails. A mean response time can look acceptable while a consequential minority wait far longer. Use distributions and case review where data permits, while protecting identity.

Map manual workarounds. Staff often preserve the customer experience through private spreadsheets, personal memory, or unauthorised discretion. Those behaviours are evidence of both commitment and system fragility.

Language

Customer words reveal the task and expectation. Review search queries, chat and call reasons, email, complaints, reviews, cancellation notes, sales objections, and research interviews.

Code language around decisions: “I cannot tell,” “I thought,” “nobody told me,” “I already sent,” “I am waiting,” or “I had to.” Preserve examples without presenting a few vivid comments as prevalence. Quantitative coding requires a defined corpus and method.

Compare internal language. If teams use different terms for the same status, customers receive translation work. If one label covers several operational states, the interface may promise certainty the system cannot provide.

Recovery

Inspect what happens after a failure: recognition, access to help, context preservation, authority, remedy, timing, closure, and learning. The public abstract for ISO 10002:2018 describes open, effective complaints handling, analysis, audit, and improvement. The full standard is paid; citing its abstract is not a claim of certification or complete conformance.

A complaint count alone is ambiguous. Low volume can mean high satisfaction, silent exit, an inaccessible route, or a customer who expects no useful response. Test the route yourself and observe recovery cases.

WCAG 2.2 helps inspect whether digital recovery controls are perceivable and operable. Technical conformance remains only one part of whether a person can obtain an effective remedy.

Consequence

Connect the gap to customer, commercial, operational, and risk outcomes. A confusing status may create anxiety and repeat contact. A hidden delivery constraint may create cancellation and rework. A lost handoff may expose sensitive data or deny an entitled action.

State consequence as a reasoned hypothesis when causality is not established. “This gap coincides with repeat contacts and customers describe the same uncertainty” is defensible. “This wording costs €500,000” requires a credible design and data.

Find silent exits

The customers who complain are visible. Those who leave without explaining are not. Look for interrupted tasks, no-result searches, abandoned support queues, unopened resolutions, one-time buyers, quote withdrawal, failed authentication, and channel switches.

Then recruit around the behaviour. Ask former customers or non-completers to reconstruct the decision and context. Memory is imperfect, so combine interviews with available records and observed tasks. Do not imply that every departure is a service failure; price, timing, fit, and changed need can make exit rational.

Employees near the boundary also see silent exits. Sales knows objections that never become lost-reason data. Operations sees substitutions. Support sees policies customers stop challenging. Invite those observations, then verify them.

Score gaps by consequence and recoverability

Avoid a simple “pain-point count.” Score each substantiated gap using:

• customer consequence: effort, exclusion, financial loss, uncertainty, or inability to complete;
• frequency and reach: how often and across which groups or channels it appears;
• promise distance: how directly it contradicts a material brand commitment;
• recoverability: whether the customer can recognise, reverse, or obtain remedy;
• evidence confidence: the number and quality of converging streams;
• system leverage: whether one repair affects many instances.

Legal, safety, accessibility, and privacy obligations can override a composite score. Document judgement and uncertainty.

Prioritise the first consequential gap in the journey when later issues depend on it. A better apology cannot compensate for a preventable account lockout. A faster support response cannot make an unavailable contact method accessible.

Test one repair as a system

Select the smallest repair capable of changing the mechanism. If customers repeat information because sales notes do not reach delivery, changing the confirmation email is not enough. The repair may require a shared field, ownership, consent rule, handoff summary, and agent view.

Define the expected chain: the workflow change improves context availability; staff stop asking for repeated facts; customer effort and handling time decline; resolution becomes more reliable. Measure each available link rather than jumping from feature release to loyalty.

Prototype with staff and customers. Include exception states. Run a limited release where safe, establish guardrails, and schedule review. Preserve a rollback path for changes involving automation or consequential policy.

Report evidence, not theatre

The audit report should fit the decision. Include:

1. Journey and promise in scope.
2. Methods, sample, dates, systems, and limitations.
3. Evidence grid for each material gap.
4. Service landscape and ownership behind the gap.
5. Priority with rationale and legal or risk boundary.
6. Repair hypothesis, owner, evidence, and review date.
7. Unresolved questions and data-quality problems.

Avoid a maturity score that implies more precision than the method supports. Keep source records securely and minimise personal data in the presentation.

Observe the joins repeatedly

An experience audit is a baseline, not a certification. Offers, people, platforms, policies, and customer expectations change. Add a light operating rhythm: monthly case review for one consequential journey, quarterly cross-channel evidence review, and a trigger review after material platform, policy, supplier, or automation changes.

Watch whether the repair moves work elsewhere. Shorter checkout can increase support if essential product information disappears. Faster automation can increase difficult human cases. Lower complaint volume can mean a route has become harder to access.

Strong brands lose trust quietly when their parts stop behaving as one service. Follow the customer’s whole problem, inspect the operation beneath it, and repair the first evidence-backed gap. The brand promise becomes credible where the joins hold.