The Customer Experience Benchmark Index (CXBI) is Tortoise & Hare’s proprietary benchmark of how effectively brands meet customer-experience (CX) expectations in market.
Unlike traditional CX metrics that rely on surveys or prompted feedback, CXBI is built from continuously collected, publicly available, unprompted customer commentary. This includes reviews, social discussion and media references where customers choose to express their experience in their own words.
CXBI transforms this real-world feedback into a standardised, comparable signal that shows whether a brand’s customer experience is improving or deteriorating over time, and how that performance compares with competitors and industry benchmarks.
If survey-based metrics provide a snapshot of sentiment at a moment in time, CXBI provides an always-on view of CX performance as it is experienced and discussed in the wild.
CXBI is designed to measure:
Relative CX performance
How a brand’s customer experience compares to the broader market and to direct competitors.
Direction and momentum
Whether CX performance is improving, declining or stabilising over time.
Drivers of experience
The underlying CX Drivers — Quality, Service, Value, Convenience and Loyalty — that are lifting or dragging overall performance.
Market-level signals
How customers collectively respond to experiences, rather than individual customer outcomes.
CXBI is optimised for trend detection, benchmarking and prioritisation, not for transactional diagnosis.
CXBI does not attempt to measure:
Absolute or objective “CX quality”
Individual customer satisfaction or intent
Causal attribution between CX changes and commercial outcomes
Internal operational performance
Scores and movements should be interpreted as signals, not verdicts. CXBI indicates where to look and what is changing, not why something happened in isolation.
This deliberate boundary is what allows CXBI to remain stable, comparable and defensible across industries and time.
CXBI uses only content already available in the public domain. This includes publicly accessible customer reviews, social commentary and media references that discuss customer experience.
No personal, private or sensitive customer information is collected or inferred.
Key principles include:
Unprompted data only
CXBI excludes surveys, incentives or structured prompts to reduce response bias.
Privacy-first design
No personally identifiable information (PII) is captured, stored or reported.
Ethical aggregation
Individual customer voices are never surfaced; CXBI operates exclusively on aggregated signals.
For participating organisations, internal Voice-of-Customer (VoC) data may be integrated into private CXBI dashboards using strict privacy and security controls. This proprietary data is never included in public CXBI benchmarks or reports.
CXBI converts large volumes of real-world customer commentary into a single, comparable benchmark signal through a structured, multi-stage pipeline.
At a high level, the process works as follows:
Continuous collection
Publicly available customer-experience mentions are collected on an ongoing basis across multiple sources.
Cleaning and bias removal
Non-numerical text is processed to preserve customer meaning while reducing distortions introduced by platform-specific rating systems, duplicate content and atypical posting behaviour.
Aspect-based sentiment analysis
Each mention is analysed using aspect-based sentiment analysis, allowing sentiment to be assigned to the specific elements of the experience being discussed, rather than to the text as a whole.
Theme identification and clustering
Related experience signals are grouped into themes based on semantic similarity and recurrence in customer language. These themes surface the specific issues, moments or expectations shaping customer experience within each CX Driver.
CX Driver classification
Identified aspects and themes are mapped to the five CX Drivers — Quality, Service, Value, Convenience and Loyalty — to reveal which parts of the experience are lifting or dragging overall performance.
Normalisation to a historical baseline
Scores are standardised against a multi-year historical baseline so that different brands, industries and sources can be compared on a common scale.
De-seasonalisation and trend isolation
Known weekly and annual cycles are removed to isolate the underlying CX trend rather than routine fluctuations.
Aggregation into the CX Benchmark Index
The resulting signal is aggregated into a single CXBI score that reflects relative market performance and directional movement over time.
Each stage of this pipeline is designed to improve comparability, stability and interpretability, ensuring CXBI reflects meaningful shifts in customer experience rather than short-term noise.
Theme identification occurs after aspect-level sentiment is established, ensuring themes reflect both what customers are discussing and how they feel about it.
Themes provide a diagnostic layer that explains movement in CX Driver scores but are not themselves used as benchmark indices.
Normalisation and trend extraction are applied only after theme and driver signals are aggregated, preserving consistency across brands and time periods.
CXBI breaks overall customer-experience performance into five core CX Drivers. These drivers represent the primary dimensions through which customers evaluate their experience, regardless of industry or channel.
The five CX Drivers are:
Quality
The reliability, effectiveness and consistency of a product or service, including performance, stability and delivery against expectations.
Service
The support and assistance provided to customers, including responsiveness, resolution quality, staff interactions and availability of help.
Value
Perceived worth in relation to cost, pricing fairness, fees, and return on investment.
Convenience
Ease of access, use and interaction across channels and touchpoints, including speed, effort and friction.
Loyalty
Factors that drive repeat behaviour, advocacy and emotional connection, including trust, rewards and long-term relationship signals.
The CX Drivers are deliberately:
Stable — they do not change over time
Industry-agnostic — applicable across sectors
Mutually exclusive — reducing overlap and double-counting
Collectively exhaustive — capturing the dominant dimensions of CX
This stability allows CXBI to benchmark performance consistently across industries and time, while still remaining interpretable and actionable.
Driver-level analysis reveals not only whether CX performance is changing, but which parts of the experience are responsible for that movement, enabling focused prioritisation rather than broad, unfocused action.
While CX Drivers explain which part of the experience is influencing overall performance, theme analysis explains what customers are actually talking about within those drivers.
Themes represent recurring topics, issues or moments that emerge from customer commentary — such as pricing transparency, wait times, account access or claim delays — and provide the contextual detail behind changes in CXBI and CX Driver scores.
CXBI identifies themes using semantic embedding and clustering techniques applied to customer commentary.
In practice, this means:
Customer text is converted into semantic representations that capture meaning rather than keywords.
Similar expressions of experience are grouped together based on semantic similarity.
Clusters are reviewed and refined to ensure they represent coherent experience themes.
This approach allows themes to emerge naturally from customer language, rather than being constrained by predefined taxonomies or survey-based categories.
Themes are therefore:
Derived directly from how customers describe their experiences
Dynamic and emergent, evolving as customer expectations and market conditions change
Specific to industries and brands, rather than fixed, universal constructs
Within the CXBI pipeline:
Aspect-based sentiment analysis determines how customers feel about specific elements of the experience.
Theme clustering groups related experience signals into meaningful topics.
CX Drivers provide the stable framework that organises themes into benchmarkable dimensions.
This separation ensures that CXBI remains statistically stable at the index and driver level, while themes remain flexible and sensitive to real-world change.
Theme analysis is used to:
Explain why CXBI and CX Driver scores are moving.
Identify the specific experience moments lifting or dragging performance.
Surface emerging risks and opportunities earlier than traditional CX metrics.
Themes provide the diagnostic layer that connects benchmark movement to real customer experiences and informs targeted investigation and action.
Themes are not intended to be benchmarked in isolation in the same way as CXBI or CX Driver scores.
Because themes reflect evolving customer language and priorities, their value lies in:
Relative importance within an industry or brand.
Directional change in share of conversation and sentiment.
Their relationship to CX Driver movement over time.
This distinction ensures CXBI remains stable and comparable, while themes remain contextual, explanatory and responsive to real-world customer experience.
Customer sentiment varies widely by industry, channel, brand visibility and overall market conditions. Raw sentiment levels alone are therefore not suitable for fair comparison.
To address this, CXBI uses normalisation against a historical baseline to place all results on a common, comparable scale.
A shared reference point
CXBI scores are standardised so that zero (0) represents the long-term market average at an industry level.
Relative, not absolute, meaning
A positive CXBI (>0) indicates above-market performance, while a negative CXBI (<0) indicates below-market performance. Zero does not imply neutral sentiment; it represents relative position versus the historic industry baseline.
Consistent comparability over time
Because all scores are calculated against the same historical reference, month-on-month, year-on-year and cross-industry comparisons remain valid.
Using a multi-year baseline allows CXBI to:
Smooth short-term volatility
Account for structural differences between industries
Detect meaningful change rather than background noise
This approach ensures CXBI reflects true movement in customer experience performance, not fluctuations driven by source mix, seasonality or changes in overall sentiment levels.
Scores are standardised using a consistent historical window to preserve longitudinal comparability.
Because CXBI operates on a standardised scale, the magnitude of gaps and movements can be interpreted directionally and comparatively across brands and periods.
Customer-experience signals exhibit strong and predictable weekly and annual patterns. For example, conversation volume often peaks mid-week and softens on weekends, while sentiment can shift systematically across the year due to seasonal demand, campaigns or external events.
If left unadjusted, these recurring patterns can obscure genuine changes in customer experience.
To address this, CXBI applies seasonal decomposition and smoothing techniques to isolate the underlying CX trend from routine cyclical effects.
Identification of recurring cycles
CXBI identifies consistent weekly and annual cycles present in customer-experience data, recognising that these patterns are expected and repeatable.
Removal of seasonal effects
Once identified, seasonal components are removed so that benchmark movement reflects changes beyond normal calendar-driven behaviour.
Preservation of underlying signal
The remaining trend represents sustained shifts in customer experience rather than short-term spikes or dips.
By smoothing daily variation and stripping out known seasonal effects, CXBI is able to:
Reduce the impact of short-term noise
Avoid over-reacting to campaign- or event-driven spikes
Highlight meaningful directional change earlier and more reliably
This ensures CXBI focuses on what has changed.
Seasonal and trend components are separated using time-series decomposition techniques designed to handle multiple seasonal patterns.
Smoothing prioritises sustained movement over transient volatility, supporting reliable month-on-month and year-on-year comparison.
Not all customer-experience signals are equally strong. Differences in data volume, consistency and agreement across sources can affect how much weight should be placed on any individual CXBI result.
To account for this, CXBI includes a confidence score alongside benchmark outputs.
The confidence score reflects the strength and reliability of the underlying signal, not the quality of the customer experience itself.
It is influenced by four contributing factors:
Volume
The amount of relevant customer commentary available for a brand or industry.
Stability
The consistency of the signal over time, including the absence of erratic swings driven by isolated events.
Cross-source agreement
The degree to which sentiment and themes align across different public data sources.
Trend alignment
The coherence between short-term movement and longer-term trends.
Together, these factors help distinguish between robust signals and signals that require greater contextual interpretation.
Higher confidence indicates a stronger, more stable signal that can be interpreted with greater assurance.
Lower confidence indicates higher uncertainty and the need to interpret results cautiously and in context.
A low confidence score does not imply poor CX performance. It simply reflects limitations in signal strength, such as lower data volume or greater variability.
Confidence scores are designed to support responsible use of CXBI, helping prevent over-interpretation of weak or volatile signals while preserving sensitivity to meaningful change.
CXBI is designed to support benchmarking, prioritisation and early signal detection, not to act as a definitive measure of customer experience in isolation.
To interpret CXBI responsibly, results should be considered in terms of direction, magnitude and context, rather than individual point values alone.
Focus on movement, not absolutes
Changes in CXBI over time are more meaningful than any single score in isolation.
Interpret gaps relatively
Differences between brands or against industry benchmarks indicate relative performance, not absolute CX quality.
Use driver-level insight
Changes in overall CXBI should be interpreted alongside CX Driver movement to understand what is lifting or dragging performance.
Consider confidence alongside scores
Confidence scores provide essential context on how much weight to place on any given result.
CXBI is most effective when used in combination with other CX and business measures, such as:
Internal Voice-of-Customer data
Operational performance metrics
Customer-effort and satisfaction measures
Commercial and behavioural indicators
Used this way, CXBI helps organisations see where experience is changing, and help explain why and what to do next.
While CXBI is designed to provide a robust, real-world view of customer experience, it has inherent limitations that should be understood when interpreting results.
Key considerations include:
Digital visibility bias
CXBI relies on publicly available customer commentary and may under-represent experiences that occur primarily offline or are less likely to be discussed publicly.
Brand coverage variability
Brands with lower public visibility or fewer customer interactions may exhibit lower confidence or more volatile signals.
Campaign and event effects
Marketing campaigns, service incidents or media coverage can temporarily influence customer commentary without reflecting sustained CX change.
Relative, not absolute truth
CXBI reflects relative market position and momentum, not an objective or complete assessment of CX quality.
These limitations are not flaws but characteristics of market-level CX measurement. When interpreted with appropriate context, CXBI provides a valuable lens on how customer experience is evolving in the real world.
CXBI is designed to complement, not replace, existing customer-experience metrics.
Traditional CX measures such as NPS, CSAT, CES and internal Voice-of-Customer programs provide valuable insight into specific journeys, moments and customer cohorts. However, they are often episodic, prompted, and limited to an organisation’s existing customer base.
CXBI adds value by providing:
An external market view
CXBI reflects how customers discuss experience in the real world, beyond surveys or owned channels.
Continuous benchmarking
Always-on measurement enables early detection of change between survey cycles.
Competitive and cross-industry context
CXBI allows performance to be understood relative to peers and broader market expectations.
Used together:
Internal CX metrics explain what customers experienced and why.
CXBI highlights where experience is shifting and how that shift compares to the market.
This combination enables more confident prioritisation, sharper diagnosis and better-aligned CX decision-making.
CXBI is governed as a long-lived benchmark rather than a static report.
Public CXBI benchmarks and reports are refreshed on a regular, scheduled basis.
Scores reflect the best available information at the time of publication.
Because CXBI operates on a historical baseline and trend-based approach:
Past values may be fully revised as new data improves trend estimation.
Revisions prioritise accuracy and comparability over preserving provisional point estimates.
This ensures the CXBI trend remains coherent and analytically sound over time.
CXBI methodology may evolve to reflect improvements in data quality, analytical techniques or market coverage.
When changes occur:
Core benchmark principles remain stable.
Changes are applied consistently across historical data where appropriate.
Material updates are documented to preserve interpretability and trust.
This governance approach ensures CXBI remains credible, transparent and fit for purpose as a market benchmark.
The CX Benchmark Index (CXBI) and associated methodology are published by Tortoise & Hare CX Agency for informational purposes only.
CXBI is derived from aggregated, publicly available data and analysed using proprietary techniques. While care is taken to ensure consistency and robustness, no representation or warranty is made regarding the accuracy, completeness or suitability of the information for any particular purpose.
CXBI scores and insights reflect relative market signals at the time of publication. They should not be interpreted as factual statements about any individual brand, nor as endorsement, criticism or recommendation.
CXBI outputs are intended to support benchmarking, prioritisation and strategic interpretation. They should be used in conjunction with other sources of information, including internal data and professional judgement.
CXBI content may be referenced or cited for internal analysis, commentary or research purposes, provided appropriate attribution is given to Tortoise & Hare CX Agency.
Re-use, redistribution or publication of CXBI data, charts or findings outside the original reports or this website requires prior written permission from Tortoise & Hare CX Agency.
© 2025 Tortoise & Hare CX Agency. All rights reserved.
