CLVTools
github.com
- Free plan
- Windows
- macOS
- Linux
What CLVTools is
CLVTools is a free, open-source R package for modeling customer attrition when purchases are not governed by contracts. It implements Pareto/NBD, BG/NBD, Gamma/Gompertz/NBD, and Gamma/Gamma models, and can estimate individual customers’ future transactions and probability of remaining active. Models accept customer transaction histories with IDs and purchase dates; the package example also uses prices. CLVTools supports time-invariant covariates across its models and time-varying covariates in some models. Additional options include correlated purchase and attrition processes, L2 regularization, and equality constraints for contextual factor parameters. The package describes its models as appropriate for non-contractual settings, commonly applied to cohorts that joined at the same time. Its forecast horizon can be configured, and it supports predictive customer lifetime value work, cohort analysis, and segmentation. The stable version is available from CRAN, with a development version on GitHub installed using devtools. Installation instructions cover Linux, macOS, and Windows. Compiling from source requires the external GNU GSL library. The project provides a walkthrough, reference manual, and vignettes, and welcomes bug reports and contributions through GitHub.
Who it suits
CLVTools suits R users analyzing repeat purchases and attrition in non-contractual businesses, especially when they work with customer cohorts. It can support analysts who need individual forecasts, cohort analysis, or customer segmentation.
Good to know, and what to watch
- Implements four probabilistic customer behavior models
- Predicts individual expected transactions and probability of being alive
- Supports covariates and model extensions such as L2 regularization
- Available for Linux, macOS, and Windows
- Requires R
- Compiling from source requires the external GNU GSL library
Plans
Every plan and priceCLVTools R package
Open source
Getting started
- Open the CLVTools project website
- Install R on Linux, macOS, or Windows
- Install the stable release from CRAN
- Use devtools to install the development version from GitHub
- Install GNU GSL before compiling from source
Limits to check
The package is intended for non-contractual settings, and time-varying covariates are available only for some models. Source installation also depends on the external GNU GSL library.
The Club's review
CLVTools is an R package for analysts modeling customer purchasing and attrition from non-contractual transaction histories. Its strongest case is a reproducible, model-focused workflow; the trade-off is that it is not a ready-made business application.
Overview
Where customers can stop buying without formally cancelling, CLVTools estimates future behavior at the individual level, including expected transactions and the probability that a customer remains active. The package also supports predictive customer lifetime value analysis, cohort analysis, customer segmentation, and a configurable forecast horizon.
Its models are usually applied to cohorts whose customers joined at the same time. That is a useful boundary to keep in mind: CLVTools suits analysts who can prepare appropriately structured customer histories, rather than teams looking for a general-purpose customer platform. The input includes customer IDs and purchase dates; prices also appear in the example.
Key features
Probabilistic model choices
The package implements Pareto/NBD, BG/NBD, Gamma/Gompertz/NBD, and Gamma/Gamma models. Having several established approaches in one R package gives analysts room to choose a model suited to their customer behavior without leaving the same workflow. Its individual predictions can reveal variation that cohort averages obscure.
Covariates and extensions
CLVTools supports time-invariant covariates across its models and time-varying covariates for some of them. Extensions include correlation between purchase and attrition processes, L2 regularization for contextual-factor parameters, and equality constraints between those parameters. These options give technically capable users meaningful control over model specification, though they also make this a specialist package rather than a simplified forecasting tool.
R-based distribution and build
The stable release is installed from CRAN; a development version is available through GitHub using devtools. Source compilation requires the external GNU Scientific Library (GSL), so users building from source need to account for that dependency. Documentation includes a walkthrough, reference manual, and vignettes on modeling, advanced techniques, and package classes. The project accepts bug reports through GitHub Issues and contributions through pull requests.
Pricing
CLVTools R package costs nothing. It is GPL-3 licensed and requires R; compiling from source requires GNU GSL. This is a strong fit for analysts who can work in R and want an open-source modeling package without a subscription. The price does not turn it into a managed service: users still need to prepare transaction data and work within an R-based process.
Platforms
The package supports Linux, macOS, and Windows. Source installation instructions cover Linux, Intel and Apple Silicon Mac, and Windows. Its platform reach is broad for an R package, but it remains dependent on an R environment rather than a standalone business interface.
Who it's for
CLVTools is best for data analysts and researchers handling non-contractual customer histories, especially cohorts with aligned joining dates, who want individual-level forecasts and control over probabilistic model choices. It is less suitable for teams seeking a turnkey interface, automated data refresh, or software that removes the need to manage modeling and installation themselves.
Pros and cons
Pros
- Several established model families: Pareto/NBD, BG/NBD, Gamma/Gompertz/NBD, and Gamma/Gamma are available in one package.
- Individual-level predictions: Expected transactions and probability of being alive support analysis beyond a single cohort estimate.
- Modeling flexibility: Covariates, process correlation, regularization, and equality constraints serve users who need to tailor model specification.
- Open-source cost: The package is free under GPL-3, with installation routes through CRAN and GitHub.
Cons
- Requires an R workflow: Teams wanting a standalone application or turnkey customer platform should look elsewhere.
- Source builds add a dependency: Compiling requires GNU GSL, which is an extra setup consideration.
- Designed for non-contractual behavior: Its stated use and cohort guidance make it a narrower choice than a broad customer analytics platform.
Alternatives
For a wider comparison, see the Customer Lifetime Value Software directory.
- Siftmo is a freemium web option with a free tier capped at 50 monthly orders and paid tiers at $29 a month for up to 500 orders and $69 a month; pick it if a web product and order-based plans suit better than an R package.
- SaaSFlow offers API and web access, with a free tier up to €120k ARR and a Pro plan at €100 a month on annual billing; consider it if those platform interfaces and ARR-based terms fit your use case.
- Pushwoosh is a freemium option across web, desktop, and mobile platforms, with a free plan that includes 1,000 MAU, 5 AI requests per day, 20,000 email sends per month, and unlimited push notifications; choose it if those messaging capabilities and limits are a closer match.
- RevenueCat has a free Pro tier up to $2,500 in monthly tracked revenue, then charges 1% of tracked revenue above that threshold; pick it if that revenue-based arrangement suits your needs better.
- VOZIQ AI Platform offers a solution free of charge with BYOL, though AWS infrastructure costs may apply; consider it if that arrangement is preferable to managing an R package.
- Klaviyo has a web-based free tier capped at 250 active profiles, 500 email sends per month, and $5 of mobile messages per month; choose it if those profile and messaging limits align with your needs.
- WorthSignal is another free option; consider it if you want to compare a different product in this category.
- Reaktion is a paid web and mobile option with a $169 a month Rise plan for 1,000 orders and one store; consider it if that order-and-store plan fits better.
Verdict
Choose CLVTools if you work in R, have non-contractual transaction histories, and want a free package with multiple probabilistic models and substantial control over their specification. Its focused analytical toolkit is the reason to choose it; the need to operate in R, and GNU GSL for source builds, are the reasons to look elsewhere.
The verdict
CLVTools provides a broad set of probabilistic models and extensions for customer behavior analysis in non-contractual settings. It is a fit for R workflows, with source builds requiring GNU GSL.
Questions people ask
Is CLVTools free?
Yes. The CLVTools R package is listed as free and is GPL-3 licensed.
Which models does it include?
It implements Pareto/NBD, BG/NBD, Gamma/Gompertz/NBD, and Gamma/Gamma models.
Which operating systems are supported?
Installation coverage includes Linux, Intel and Apple Silicon Mac, and Windows.
How is it installed?
Install the stable release from CRAN, or install the development version from GitHub with devtools.
What data does it use?
Inputs include customer transaction history with purchase dates and customer IDs; the example also uses prices.
Who maintains CLVTools?
Patrick Bachmann is identified as maintainer and author.
What the maker publishes
- Build dependency
- Compiling CLVTools from source requires the external C++ library GSL. source
- Build dependency
- Compiling CLVTools from source requires the external GNU Scientific Library (GSL). source
- Covariates
- The package supports time-invariant covariates across its models and time-varying covariates for some models. source
- Distribution
- The stable release is installed from CRAN, and a development version is installed from GitHub using devtools. source
- Documentation
- The project links to a walkthrough, a reference manual, and vignettes covering modeling, advanced techniques, and package classes. source
- Input data
- CLVTools takes customer transaction history with purchase dates and customer IDs; the example also uses prices. source
- Input data
- CLVTools requires customer transaction history with a purchase date and customer ID for each transaction record. source
- Installation
- The site documents installation from CRAN and installation of a development version from GitHub. source
- Intended use
- The package describes its models as suited to non-contractual settings and says they are usually applied to customer cohorts that joined at the same time. source
- License
- The package is licensed under GPL-3. source
- Maintainer
- The package authors page identifies Patrick Bachmann as maintainer and author. source
- Model extensions
- Available extensions include purchase and attrition process correlation, L2 regularization for contextual-factor parameters, and equality constraints between contextual-factor parameters. source
- Model options
- Supported extensions include purchase and attrition process correlation, L2 regularization, and equality constraints for contextual factor parameters. source
- Models
- It implements Pareto/NBD, BG/NBD, Gamma/Gompertz/NBD, and Gamma/Gamma models. source
- Operating systems
- The source installation instructions cover Linux, Intel and Apple Silicon Mac, and Windows. source
- Platforms
- The installation instructions cover Linux, Mac with Intel or Apple Silicon, and Windows. source
- Predictions
- The package can predict future customer behavior at the individual level, including expected transactions and probability of being alive. source
- Predictions
- Its predictions can include expected transactions, probability a customer is alive, discounted expected residual transactions, and customer lifetime value when spending data is provided. source
- Purpose
- CLVTools is an R package for probabilistic customer attrition models in non-contractual business settings. source
- Purpose
- CLVTools is an R package for probabilistic customer attrition modeling in non-contractual settings. source
- Support
- The project invites bug reports through GitHub Issues and contributions through pull requests. source
- Support and contributions
- The GitHub project welcomes feedback and contributions, and directs bug reports to GitHub Issues and contributions to pull requests. source
- Supported models
- The package implements Pareto/NBD, BG/NBD, Gamma/Gompertz/NBD, and Gamma/Gamma models. source
Checked against the maker's pages on 9 Oct 2026.