B&H · Customer Insights & Customer Experience

Turning customer feedback into operational diagnosis.

A score can show that something changed. The useful work begins when feedback reveals where the experience is failing, which customers are affected, and what operational system may be responsible.

A change in vantage point

From designing employee tools to examining the customer experience.

The work changed from shaping individual interfaces to building a system for listening, comparing, diagnosing, and communicating what customers were experiencing across the business.

Enterprise UX

Designed operational tools for sales, service, cashiers, buyers, and managers—developing direct familiarity with the systems behind the customer experience.

Customer Insights

Worked on a two-person team designing and maintaining surveys, analyzing feedback, identifying experience drivers, and reporting issues to the CMO and executive leadership.

Customer Experience & Medallia

Helped select and implement the CX platform, then built and maintained its surveys, permissions, dashboards, topic rules, validation, and data connections.

Customer feedback became another view into the operating system.

Earlier enterprise knowledge made it possible to ask not only what customers disliked, but which process, handoff, rule, or technical system might be producing the experience.

01 · From score to cause

Measurement was the beginning—not the conclusion.

Likelihood-to-recommend and overall-satisfaction scores established direction. Open-text feedback and experience-driver analysis supplied the clues needed to find specific, actionable causes.

01ListenCollect structured ratings and open-text feedback.
02GroupOrganize comments into recurring topics and issues.
03CompareExamine experience, time period, customer group, and related measures.
04InvestigateConnect the customer signal with operational and technical context.
05TraceIdentify the likely process or system behind the pattern.

Operational example

Emerging fulfillment complaints did not remain a reporting category.

A pattern noticed in customer feedback prompted investigation beyond the survey data. The team helped trace the complaints to a malfunction in an automated warehouse system.

Why it mattersThe finding connected customer language to an operational cause.

02 · A measurement program

Ongoing listening and focused research served different purposes.

The program combined continuously maintained experience surveys with shorter investigations designed around a specific question. Ad hoc findings could then influence recurring studies and the questions added to ongoing surveys.

20+

Ongoing surveys

Continuously maintained across multiple customer experiences, with parameters, conditional questions, and routing logic.

40+

Ad hoc surveys

Focused, time-limited studies created to investigate a specific issue, audience, or decision.

One program

Learning moved in both directions

Recurring surveys exposed questions worth deeper study; focused research supplied findings and questions that could become part of ongoing measurement.

03 · Survey architecture

One invitation could produce many legitimate paths.

Parameters and query strings carried context into Medallia. Conditional questions and routing logic then adapted the survey to the customer’s experience, while enriched data and topic rules supported later analysis.

Context enters

Parameters

Experience, transaction, customer, and invitation context arrive with the response path.

Experience adapts

Conditional questions

Questions and branches change according to what the respondent actually experienced.

Data connects

Enrichment & integration

Technical and Data Warehouse coordination adds business context and resolves cross-system issues.

Meaning emerges

Dashboards & topics

Measures, open text, classifications, and operational context become usable views for analysis.

ParametersQuery stringsConditional questionsRouting logicRole architectureData WarehouseJira coordination

04 · Validate the automation

Automated classification was treated as a claim to test.

Topic classification and sentiment analysis can create the appearance of precision. R and Excel were used to test how well those outputs held up across different slices of the program rather than assuming the platform was correct.

Survey typeCheck whether performance changes across different customer experiences.
Time periodCompare results across changing language, conditions, and sample periods.
Topic hierarchyInspect both broader categories and more specific nested topics.
ClassifierCompare alternative classification rules and model outputs.
OutputTest both topic assignment and sentiment labeling.

05 · Platform stewardship

The work continued after implementation.

Helping evaluate and select Medallia led into implementation, the initial role-and-permission structure, launch, and continuing responsibility for configuration, diagnosis, and improvement.

Selection and implementation

Contributed to evaluating and selecting the CX platform, then to the implementation and initial role-and-permission architecture.

Platform launch

Medallia launched at B&H, moving the work from implementation into ongoing survey, dashboard, topic, permission, and integration stewardship.

Deeper platform capability

Earned the Medallia Super Admin Credential in 2023 and Partner Credential in 2024, supporting more advanced configuration and diagnosis.

What the work left behind

A measurement and diagnosis system.

The evidence is not a gallery of interface screens. It is the structure connecting questions, customer context, analytical methods, platform rules, and operational findings.

Research programOngoing surveysAd hoc studiesExperience driversExecutive reporting
Survey designParametersConditional questionsRouting logicQuestion evolution
Platform structureRolesPermissionsDashboardsTopic rulesData integration
Analytical evidenceR validationExcel validationClassification testsSentiment testsOperational diagnosis
Presentation boundary

This page uses abstract diagrams and documented scope rather than reconstructing confidential dashboards, survey screens, customer data, or internal system views.

No customer dataNo simulated screenshotsSupported claims only

Continue through the chronological portfolio.