How to really quantify the economic value of customer information in corporate databases

The study addresses the challenge firms face in quantifying the economic value of customer information generated and managed via CRM and related IT tools within a knowledge-based economy. While CRM’s strategic role is accepted, practitioners lack clear, implementable methods to translate information into economic value for customer value management. Existing customer lifetime value (CLV) work often remains theoretical, complex, or omits links to information assets and IT. The research aims to conceptualize and empirically verify a model that quantifies the economic value of customer information and knowledge in B2C settings, rooted in information valuation theory (information as an asset whose value derives from its use). Objectives: (1) propose and test a new model to measure the value of business information in CRM environments (VICRM), (2) compare it with traditional CLV methods, and (3) analyze the sensitivity and influence of each component. The study is motivated by the need for holistic, interdisciplinary approaches integrating marketing, IT, and finance, and by the managerial need (especially in SMEs) for credible ROI assessments of CRM/BI initiatives.

The paper builds on several streams: (a) CLV as a predictor of future profitability and its evolution to include indirect customer contributions (Kumar, 2018); (b) CRM as a strategic bridge linking IT and marketing to create value through better information flows and customer understanding (Harrigan et al., 2020); (c) information/knowledge as an economic asset (Glazer, 1991; Moody and Walsh, 1999); (d) gaps in demonstrating BI/CRM’s tangible business value, especially for customer value management (Trieu, 2017); and (e) inclusion of nonfinancial, attitudinal “mindset metrics” to improve predictive power (Srinivasan et al., 2010). Traditional transactional CLV models often overemphasize margins and retention and ignore network effects and customer engagement/knowledge externalities, leading to overvaluation in some contexts. The literature calls for integrative, practical models unifying marketing, IT, and finance (the triad), capturing direct and indirect value, and empirically validated in real settings.

Design: Dual approach comprising (1) exploratory-conceptual literature review to derive constructs and structure of the model, and (2) descriptive-quantitative empirical case study in a Spanish insurance company (mature, data-rich, medium-sized, high managerial buy-in). Assumptions include that churned customers do not return (consistent with contractual insurance context). Discount factors were derived from the forward risk-free rate curve (average of next 5 years applied across the projection). Model formulation: Traditional transactional CLV is defined as the discounted sum of gross profits (including serving costs) over n periods at discount rate d. The model extends CLV to f(CLV) by adding three mediating constructs that modify a customer’s potential transactional value: Engagement (CEV), Knowledge (CKV), and Social (CSV) values: f(CLV) = f(CLV) + CEV + CKV + CSV. Operationally, the VICRM for a portfolio sums the extended values across customers/segments with multiplicative interactions: VICRM = Σ[ CLV*r*CEV + |CLV|*r*CKV + CLV*CSV*CEV ], recognizing interaction effects among constructs. Constructs and metrics: - Engagement Value (CEV): CEV = (NPS + Ton)/2 * sqrt(Vo + Vi + Pop). Inputs: NPS in [-1,1] or scaled; Ton sentiment/CSAT in [0,1]; Vo = social comment volume; Vi = virality (shares); Pop = likes. Constraint 0 ≤ r*CEV ≤ 1; acts as a mediator of retention r and CLV. - Knowledge Value (CKV): CKV = Q * [(Act + Ki)/2], with Q (information quality/usefulness) in [-1,1], Act (recency/update) in [0,1], Ki (quantity of customer data) in [0,1]. Mediator of CLV using |CLV| to avoid sign paradoxes when both CLV and CKV are negative. - Social Value (CSV): Based on network effects (Metcalfe’s law). CSV = Qm * N^2, with Qm in [-1,1] capturing contact quality/authority/influence and N = size of influential contact network. Mediates CLV and interacts with CEV. Empirical operationalization: Applied to N=60 customers sampled randomly across six predefined archetypes (10 each) representing consumption patterns and socio-demographics: Alternative Lawyers, Complementary Lawyers, Relatives, Youths, Other collectives, Seniors. All selected customers had responded to company surveys; responders showed higher engagement but were otherwise similar in representativeness. Variables were mapped to company data (Table 2): CLV constructed from premiums/balances, product margins, and costs; retention defined as active customers over two-year activity window; sentiment/CSAT, contact volumes, and digital interactions sourced from internal systems; CKV inputs (quality/recency/quantity) from CRM field completeness, update recency, and user-rated usefulness; CSV inputs from declared contacts and social media followers/authority; several variables normalized via Min-Max scaling. Statistical analysis included normality tests (Shapiro-Wilk, D’Agostino-Pearson), nonparametric two-way Scheirer–Ray–Hare tests to assess heterogeneity across constructs and segments, correlation analyses (Pearson, Spearman), dispersion metrics, temporal distribution analysis (annual relative weights 2021–2035), and sensitivity analysis to retention rate changes.

- Descriptive results (N=60): Totals—Engagement value: €65,758.21; Knowledge value: €40,774.33; Social value: €3,872.99; VICRM: €110,405.53. Averages—Engagement: €1,095.97; Knowledge: €679.57; Social: €64.55; VICRM: €1,840.09. Substantial heterogeneity across segments and constructs; CSV contributions were small due to limited recorded relationships. - Heterogeneity confirmed: Scheirer–Ray–Hare test showed significant differences for both factors (segments and components): H with p-values 0.017 (rows/segments) and 3.74E-19 (columns/components); no significant interaction effect. - Construct relationships: Positive associations among constructs and with VICRM. As anticipated, engagement correlates with the amount/quality/recency of information (Spearman rho ≈ 0.468). Social and knowledge components showed near-zero, nonsignificant relationships. - Comparison with traditional CLV: Strong positive correlation between VICRM and traditional CLV (Pearson r ≈ 0.562; Spearman rho ≈ 0.647). Traditional CLV values were more than twice VICRM on average, indicating potential overvaluation in traditional models. For 14/60 customers (23.3%), VICRM exceeded traditional CLV, implying risk of undervaluation under traditional approaches for a nontrivial subset. - Robustness/dispersion: New model exhibited greater homogeneity and robustness to outliers compared to traditional CLV, based on dispersion statistics (e.g., lower interquartile-based dispersion for VICRM; though coefficient of variation was lower for traditional CLV, other dispersion measures favored VICRM). - Temporal distribution: VICRM less regressive; approximately seven years to reach 50% of expected value vs. traditional findings concentrating value in early years. Annual relative weights (illustrative) showed VICRM distributing value more evenly over 2021–2035 than the traditional model. - Retention sensitivity: VICRM substantially reduced sensitivity to retention changes compared to traditional CLV—5% drop in retention led to −32.7% in traditional CLV vs. −9.2% in VICRM; a 10% drop led to −41.2% vs. −14.1%, respectively. - Segment-level extrapolation (illustrative totals): Alternative Lawyers VICRM ≈ €71.33M, Complementary Lawyers ≈ €47.49M, Relatives ≈ €8.11M, Youths ≈ €20.14M, Other collectives ≈ €3.88M, Seniors ≈ €19.31M; total ≈ €170.25M, evidencing strong heterogeneity and actionable segment differences. - Distributional insight: Results align with a long-tail pattern rather than Pareto, suggesting less concentrated profitability and risk dispersion enabled by IT/CRM-informed management.

The findings demonstrate that treating customer information and knowledge as economic assets and embedding them as mediators of transactional CLV (via engagement, knowledge, and social constructs) yields meaningfully different and more nuanced valuations than traditional models. VICRM addresses the research gap by linking IT/CRM-derived soft metrics (e.g., sentiment, interaction volumes, data quality/recency, network influence) to hard financial outcomes. This approach reduces overreliance on retention rates and early-period cash flows, mitigating traditional models’ regressiveness and sensitivity to retention shocks. The strong correlation with traditional CLV indicates consistency in ranking/value direction, while substantial differences in magnitude and temporal allocation provide managers with new levers for segmentation and resource allocation (e.g., focusing on boosting engagement or data quality to unlock value). The small observed CSV contribution reflects data limitations, not conceptual weakness; improving capture of social ties and influence could unlock additional value. Overall, VICRM offers a more comprehensive, triad-based (marketing–IT–finance) framework to inform customer portfolio management and firm valuation perspectives.

This paper proposes and empirically tests VICRM, a deterministic, practicable model that extends CLV with engagement, knowledge, and social value constructs to quantify the economic value of customer information in CRM environments. The model produces systematically different results from traditional CLV, is less regressive, and is less sensitive to retention, enabling more robust segmentation and investment decisions. Academically, it operationalizes information-as-asset theory with actionable soft and hard metrics, validates predicted construct relationships, and provides an empirical basis for claims about traditional overvaluation. Managerially, it equips B2C firms (notably SMEs) with a transparent, implementable tool to value CRM initiatives, segment customers by information-derived value, and prioritize interventions (e.g., enhancing engagement, data quality/recency, or social influence capture). Future research should refine construct weighting, expand datasets (especially for social networks), test in non-contractual and B2B contexts, and explore stochastic variants and industry-specific calibrations.

- Sample and composition: Small N (60) from a single insurance company; respondents more engaged than non-respondents, potentially biasing engagement-related metrics; extrapolations to the full portfolio should be cautious. - Data coverage: Limited recording of social ties reduced CSV’s observed impact; broader, higher-quality social/influence data may materially change results. - Model choices: Multiplicative structure and equal weighting in some subcomponents (e.g., CKV’s recency vs. quantity) are partly heuristic pending broader empirical validation; construct ranges and normalizations may influence outcomes. - Context specificity: Results from a contractual, insurance context may not generalize to non-contractual settings without adaptation; discount and retention assumptions affect magnitudes. - Potential measurement error: User-rated information quality (Q) and sentiment measures (Ton) may introduce subjectivity; normalization choices (Min–Max scaling) can affect comparability.