The DOI matches its Crossref record; 29 of the 29 references it cites were found in a registry or an index; 2 of the 57 sources its claims cite were checked for support.
Forecasting the next likely purchase events of insurance customers: A case study on the value of data-rich multichannel environments
Open accessAbstract
Purpose: The purpose of this paper is to demonstrate the value of enriched customer data for analytical customer relationship management (CRM) in the insurance sector. In this study, online quotes from an insurer’s website are evaluated in terms of serving as a trigger event to predict churn, retention, and cross-selling. Design/methodology/approach: For this purpose, the records of online quotes from a Swiss insurer are linked to records of existing customers from 2012 to 2015. Based on the data from automobile and home insurance policyholders, random forest prediction models for classification are fitted. Findings: Enhancing traditional customer data with such additional information substantially boosts the accuracy for predicting future purchases. The models identify customers who have a high probability of adjusting their insurance coverage. Research limitations/implications: The findings of the study imply that enriching traditional customer data with online quotes yields a valuable approach to predicting purchase behavior. Moreover, the quote data provide supplementary features that contribute to improving prediction performance. Practical implications: This study highlights the importance of selecting the relevant data sources to target the right customers at the right time and to thus benefit from analytical CRM practices. Originality/value: This paper is one of the first to investigate the potential value of data-rich environments for insurers and their customers. It provides insights on how to identify relevant customers for ensuing marketing activities efficiently and thus avoiding irrelevant offers. Hence, the study creates value for insurers as well as customers.
Overview
CoChat Summary
TL;DR
This case study links anonymous online quote records from a Swiss insurer’s website to existing automobile and household policyholder data and uses random forest models to forecast churn/retention and cross-selling. Enriching traditional CRM data with quote-derived features substantially improves prediction of future purchase events compared with baseline models using only traditional customer data.
Key findings
- For churn and retention, the research-shopper (RS) models outperformed non-RS baselines: automobile accuracy 0.844 vs 0.684 (increase 0.160), and household accuracy 0.859 vs 0.787 (increase 0.072).
- The RS churn/retention models correctly classified churn and retention much better than non-RS models: automobile churn F-score 0.732 vs 0.020 and retention F-score 0.792 vs 0.130; household churn F-score 0.652 vs 0.055 and retention F-score 0.750 vs 0.195.
- For cross-selling, non-RS models had high accuracy but failed on the positive class: Automobile→Household RS accuracy 0.893 vs non-RS 0.994, but F-score 0.818 vs 0.025; Household→Automobile RS accuracy 0.879 vs non-RS 0.998, but F-score 0.786 vs 0.000.
- Online quote features showed high variable importance, e.g., vehicle age (onl) VI = 1075.88 in the automobile churn/retention RS model, household value insured (onl) VI = 427.46 in the household churn/retention RS model, and vehicle age (onl) VI = 894.30 in the Automobile→Household cross-selling RS model.
Methods
- Case study using data from a Swiss top-three non-life insurer, 2012–2015. Original data included automobile and household policies (approx. 2.5 million and 3 million observations) and anonymous online quotes (approx. 275,000 automobile and 90,000 household).
- Active customer records were linked to online quotes via joint covariates (e.g., date of birth, postal code, gender, driver’s license issue date, vehicle model for automobile; date of birth, postal code, family status, home ownership status, household value insured for household). RS samples contained linked quotes; non-RS samples used a random observation date without a quote. Future activity was observed for six months, with response variables for churn/retention/no action or cross-selling/no action.
- Random forest classification models were fitted with class weights for imbalance, using a holdout split (2/3 training, 1/3 validation; Automobile→Household used 80/20). Variable importance was based on Gini index, parameters B and m were optimized with 10-fold cross-validation, and performance was evaluated with accuracy, F-score, ROC curves, and AUC in R.
Limitations
- Stated by the paper: only two non-life insurance products are included, so results may differ for life insurance; cross-selling could not be analyzed as a next-product-to-buy approach because online quote data for additional products were unavailable; data come from one insurer only and may be biased by organizational structure and marketing strategy; the database lacks online quote data from other insurers, so false positives in cross-selling cannot be distinguished between no shopping and purchase from a competitor; cross-selling results may be affected by a customer data quality issue where customers re-registered under new customer IDs, causing missed cross-selling cases.
- Evident from the text: the non-RS cross-selling models achieve very high accuracy mainly by correctly classifying the dominant no-action class, while their F-scores for actual cross-selling are near zero (0.025 and 0.000), indicating poor positive-class prediction.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures
Verification
Claims the paper makes
2 Valid · 0 Not valid · 44 Not checked · 62 Own results and interpretations
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“We use a period of six months based on insurance-specific empirical evidence of previous research (see e.g. Guelman et al., 2015; Mau et al., 2015).”
What the sources say
- Not checked — no full text of the sourcecites Guelman et al., 2015 source
- Partly supportedcites Mau et al., 2015 sourceExperimental — accuracy not yet measured
Mau et al. provide insurance-specific empirical evidence about the time from online research to policy purchase (Section 4.2, p. 12). They report an average of 19.74 workdays, but our copy does not establish six months as an evidence-based period. The insurance-specific evidence is supported; the six-month choice is not. (LLM, conf=0.84)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The models identify customers who have a short-termed demand to adapt their insurance coverage with a high probability.”
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Moreover, the quote data provide supplementary features that contribute to improving the prediction performance.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Thus, the presented approach is able to separate customers who currently shop for insurance and would be receptive to offerings from those who are not.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Enhancing traditional customer data with such additional information boosts the accuracy for predicting future purchases substantially.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
47 more claims
Create a free account to see all 108 claims and every source verdict.
- Probably not a substantial claim. 34
- It may state two separate claims as one. 29
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Original online quotes (OQ)
“| 1678912 | 2 | 10.01.14 | 31.12.99 | R | AM | 02.06.12 | AM | NOA |”
§ Forecasting next likely purchase events
“Thus, we choose a random observation date during the policy tenure as substitution and observe future customer activities a posteriori within the next six months.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 4.2 Cross-Selling Models
“For the non-research-shopper sample, the datasets include 66,667 (training) and 3,333 (test) observations.”
§ Prediction performance of the final models
“| Sample Size | 902 | 225 | 73 334 | 36 666 | 9 894 | 4 946 | 66 667 | 33 333 |”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 4.1 Churn and Retention Models
“age (obs) (VI = 1047 . 61)”
§ Forecasting next likely purchase events
“| Age (inc) | 1047.61 | | | Household value insured (inc) | 247.41 | | | | Age (obs) | 1009.99 | | | Period since last claim (obs) | 204.17 | | |”
Citations
Cites
An introduction to ROC analysis
Fawcett, T. (2006) · Pattern Recognition Letters · doi:10.1016/j.patrec.2005.10.010 · Cited by 21,879
Variable selection using random forests
Genuer, R., Poggi, J.-M., & Tuleau-Malot, C. (2010) · Pattern Recognition Letters · doi:10.1016/j.patrec.2010.03.014 · Cited by 2,619
Challenges and Opportunities in Multichannel Customer Management
Neslin, S. A., Grewal, D., Leghorn, R., Shankar, V., Teerling, M. L., Thomas, J. S., & Verhoef, P. C. (2006) · Journal of Service Research · doi:10.1177/1094670506293559 · Cited by 1,154
Multichannel customer management: Understanding the research-shopper phenomenon.
Verhoef, P. C., Neslin, S. A., & Vroomen, B. (2007) · International Journal of Research in Marketing · doi:10.1016/j.ijresmar.2006.11.002 · Cited by 999
Bagging and Boosting Classification Trees to Predict Churn
Lemmens, A., & Croux, C. (2006) · Journal of Marketing Research · doi:10.1509/jmkr.43.2.276 · Cited by 411
The role of multichannel integration in customer relationship management
Payne, A., & Frow, P. (2004) · Industrial Marketing Management · doi:10.1016/j.indmarman.2004.02.002 · Cited by 367
Customer Relationship Management: A Databased Approach
Kumar, V., & Reinartz, W. (2005) · Cited by 294
Managing Marketing Communications with Multichannel Customers
Thomas, J. S., & Sullivan, U. Y. (2005) · Journal of Marketing · doi:10.1509/jmkg.2005.69.4.239 · Cited by 286
Probability Models for Customer-Base Analysis
Fader, P. S., & Hardie, B. G. S. (2009) · Journal of Interactive Marketing · doi:10.1016/j.intmar.2008.11.003 · Cited by 204
Instant Customer Base Analysis: Managerial Heuristics Often “Get It Right”
Wübben, M., & von Wangenheim, F. (2008) · Journal of Marketing · doi:10.1509/jmkg.72.3.82 · Cited by 198
Research works printed but not linked: 4
- 2011, Data Mining: Concepts and Techniques
- 2009, The Elements of Statistical Learning Data Mining, Inference, and Prediction
- MSI (2014), “Research priorities 2014-2016”, Marketing Research Institute (MSI), Cambridge, available at: www.msi.org/uploads/_les/MSI RP14-16.pdf (accessed April 24, 2017).
- Swiss Re, 2014, Digital distribution in insurance: a quiet revolution, Sigma
Cited by
Customer relationship management and its impact on entrepreneurial marketing: a literature review
Guerola-Navarro, V., Gil-Gómez, H., Oltra-Badenes, R., & Soto‐Acosta, P. (2022) · International Entrepreneurship and Management Journal · doi:10.1007/s11365-022-00800-x · Cited by 211
Machine Learning Approaches for Auto Insurance Big Data
Hanafy, M., & Ming, R. (2021) · Risks · doi:10.3390/risks9020042 · Cited by 102
Meena, P., & Sahu, P. (2021) · Vision The Journal of Business Perspective · doi:10.1177/0972262920984550 · Cited by 120
Pan, H. X., & Zhou, H. (2020) · Electronic Commerce Research · doi:10.1007/s10660-020-09409-0 · Cited by 88
Biochar implications in cleaner agricultural production and environmental sustainability
Babu, S., Singh, R., Kumar, S., Rathore, S. S., Yadav, D., Yadav, S. K., Yadav, V., Ansari, M. A., Das, A., Rajanna, G. A., et al. (2023) · Environmental Science Advances · doi:10.1039/d2va00324d · Cited by 63
Metadata
- Authors
- Mau, S., Pletikosa, I., & Wagner, J.
- Affiliations
- ETH Zurich (CH)
- Swiss Finance Institute (CH)
- ORCID
- Wagner, Joël
- Venue
- International Journal of Bank Marketing
- Publisher
- Emerald
- Published
- 2018-06-05
- Volume
- 36
- Issue
- 6
- Pages
- 1125-1144
- ISSN
- 0265-2323 (ISSN-L)
- OpenAlex
- W2806167717
- Cited by
- 23
- References
- 29
- Citations by year
- 2026: 1 · 2025: 2 · 2024: 3 · 2023: 4 · 2022: 2 · 2021: 4 · 2020: 4 · 2019: 3
- FWCI
- 4.92
- Type
- journal article
- Language
- en
- Open access
- green
- serval.unil.ch
- Licence
- other-oa
- www.emerald.com/insight/site-policies (tdm)
- Locations
- International Journal of Bank Marketing (publishedVersion)
- SERVAL (Université de Lausanne) (acceptedVersion, other-oa)
- IRIS (submittedVersion)
- Alexandria (UniSG) (University of St.Gallen) (submittedVersion)
- Topics
- Social Sciences › Business, Management and Accounting › Marketing › Customer churn and segmentation
- Social Sciences › Business, Management and Accounting › Marketing › Customer Service Quality and Loyalty
- Social Sciences › Economics, Econometrics and Finance › Economics and Econometrics › Insurance and Financial Risk Management
- Keywords
- insurance customers, purchase prediction, customer churn, customer retention, cross-selling, data enrichment, automobile insurance, household insurance, random forest classification, targeted marketing
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-09-30, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-09-30 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Forecasting the next likely purchase events of insurance customers: A case study on the value of data-rich multichannel environments. https://app.cochat.ai/papers/10.1108/ijbm-11-2016-0180
- The paper
- Mau, S., Pletikosa, I., & Wagner, J. (2018). Forecasting the next likely purchase events of insurance customers: A case study on the value of data-rich multichannel environments. International Journal of Bank Marketing. https://doi.org/10.1108/ijbm-11-2016-0180
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.

