Visual editor
Latent variables, indicators, paths via drag & drop. No YAML, no code, no friction.
PLS-SEM · Open-source engine Structural equation modeling.
Structural equation modeling.
Finally modern.
The open platform for structural equation models: visual path editor, the full PLS-SEM pipeline and publication-ready reports. Right in the browser. Or self-hosted, when your data is sensitive.
No setup, no credit card · Open-source engine · Validated against established reference values
Live preview
A real model, fully computed.
This is what openpls-engine returns for a small Quality → Satisfaction → Loyalty chain. Paths, R², HTMT and bootstrap CIs in one view.
ECSI customer-satisfaction sample (n = 250), 5 000 bootstrap resamples. Reproducible in the demo project.
Features
Everything you need for serious PLS-SEM.
Established statistics, modern tooling, open standards.
Complete PLS-SEM pipeline
Loadings, weights, path coefficients, R², GoF, validated against established reference values from the PLS-SEM literature.
Clean reports
PDF, XLSX, LaTeX and a publication JSON bundle, ready to drop into your paper. Tables, charts, methods section, one click each.
Self-host for sensitive data
Docker image for teaching, clinical work and the enterprise. Your data never leaves your network.
Open export formats
Models, weights and reports in open standards (JSON, CSV, LaTeX). Migrate in and out of other PLS tools, no lock-in.
Teams & versioning
Co-authors, reviewers and version comparison for research groups. Without endless email threads.
Advanced analyses New
Beyond the standard pipeline.
Fifteen modern PLS-SEM analyses, one click in the same editor, no second tool.
Quantify uncertainty in your effects and turn the model into out-of-sample predictions.
-
IPMA
Importance-Performance Map: surfaces constructs that are highly important and underperforming at the same time. Management priority at a glance.
-
PLSpredict
k-fold out-of-sample assessment with the full Shmueli panel: RMSE, MAE and MAPE versus a linear-model benchmark, plus an in-sample fit table for context.
-
Specific indirect effects
Every mediation chain in your model with point estimates and bootstrap confidence intervals, t-values and p-values. Stop reporting only the total indirect effect.
-
CVPAT (predictive ability)
Cross-validated predictive ability test after Liengaard et al. 2021. Head-to-head comparison against indicator-average or linear-model benchmarks, paired t-test, k-fold cross-validation.
-
FIMIX-PLS
Finite-mixture segmentation surfaces unobserved subgroups in your sample. Make heterogeneous effects visible.
-
MICOM
Three-step Henseler–Ringle–Sarstedt test for measurement invariance: configural, compositional and scalar. The required prerequisite before any multi-group comparison across countries, segments or time points.
Stress-test the measurement model and the structural equations against the most common biases.
-
PLSc (Consistent PLS)
Dijkstra-Henseler bias correction for reflective measurement. Per-LV ρ_A, dis-attenuated correlations, corrected paths and R².
-
Discriminant validity
HTMT plus the geometric-mean refinement HTMT2 (Roemer et al. 2021) and the Fornell-Larcker check side by side. Three independent perspectives on construct distinctness in one panel.
-
Gaussian copula
Park-Gupta / Hult et al. endogeneity test for structural predictors. Detects bias without instrumental variables, with built-in admissibility check.
-
Nomological validity
Pre-register expected signs for construct correlations and test each hypothesis with one-sided t-tests. Turns the theory chapter into a testable panel with a clean verdict per relationship.
-
Common method bias (Lindell-Whitney)
Marker-variable partial correlation adjustment to detect common method variance. Pick a theoretically unrelated marker, compare adjusted vs. raw correlations, decide whether CMB is a concern.
-
Moderation (Two-Stage)
Test interaction terms in your structural model. Two-stage approach after Henseler & Chin, in one click.
Build richer model topologies and pick the convergence algorithm that fits your data.
-
Higher-Order Constructs (HOC)
Disjoint two-stage workflow for hierarchical models. All four canonical types (R-R, R-F, F-R, F-F) plus nested HOCs, without rewiring the editor structure.
-
Newton scheme
Quasi-Newton inner weighting for more stable and faster convergence on complex models.
-
PCA scheme
Lohmöller’s PCA inner-weighting scheme as an alternative to centroid and path. Algorithmic choice per model.
Methods, visualised
What the outputs actually look like.
Four examples of what OpenPLS produces for your analyses. Straight from the in-app explainer panels.
01
Bootstrap distribution
5,000 resamples, point estimate and 95 % CI. This is what the uncertainty around a path coefficient looks like.
02
Importance-performance map
Which constructs matter and underperform? A four-quadrant view for management prioritisation.
03
FIMIX segmentation
Find unobserved subgroups. Surface heterogeneous effects before they bias your results.
04
Simple slopes
How does a path coefficient change at low, mean and high moderator? Three lines say more than a single p-value.
Workflow
Four steps from data to model.
- 01
Upload data
CSV, XLSX, SPSS or Stata: we read whatever you have.
- 02
Draw the model
Constructs and paths via drag & drop. Live preview of the model structure.
- 03
Compute
One click. plspm computes: loadings, paths, bootstrap confidences.
- 04
Export
Report as PDF, tables as XLSX, model as JSON or LaTeX snippet, all in open formats.
For whom
Concrete applications from the field.
PLS-SEM is used wherever relationships between latent, non-directly-measurable constructs need to be estimated, from PhD dissertations to the market research department of a Fortune 500 company.
Research & teaching
Empirical research without licensing hurdles
Dissertations, journal submissions and seminars: no commercial licence, no shared lab account. Open engine, documented methodology, reproducible reports.
- Acceptance models (TAM / UTAUT) for health-IT adoption
- Service quality & loyalty (ACSI / ECSI) in hospitality research
- Reputation and trust models (Corporate Reputation Model)
Marketing & CX
What really drives your customer KPIs?
Instead of correlating individual touchpoints, OpenPLS estimates the whole system simultaneously. You see which lever actually moves loyalty and NPS, and which is just noise.
- NPS and CSAT driver models from survey data
- Customer-journey impact models (Awareness → Trust → Conversion)
- Brand-equity studies with formative indices
Industry & operations
Make latent drivers of complex processes measurable
OpenPLS is particularly strong with small N and many indicators, the typical setup for employee, supplier and innovation surveys.
- Employee engagement and safety culture (safety-climate models)
- Supplier performance and supply-chain resilience
- Innovation capability (absorptive capacity, dynamic capabilities)
Health & public sector
Sensitive data, clear methodology
Self-host as a Docker container keeps patient and employee data inside your own network. Methodologically identical to the cloud version, documented for ethics committees.
- Patient-reported-outcome models in rehab & care
- Adoption of digital health offerings (DTx, telemedicine)
- Citizen-trust models for e-government initiatives
Case Studies
Reproducible in 60 seconds.
Four published PLS-SEM models, available as one-click clones in your OpenPLS workspace. Each ships with synthetic original data, the complete path definition, and reproducible key metrics.
ECSI
European Customer Satisfaction Index
Customer Experience
Six reflective constructs (Image → Expectations → Quality → Value → Satisfaction → Loyalty), the canonical model from service marketing research.
- N
- 250
- LVs
- 6
- R²(SAT)
- 0.71
Fornell et al. (1996)
Open case study UTAUT2 + Trust
E-commerce acceptance with trust
Marketing / E-commerce
Performance expectancy, effort expectancy, social influence, and trust explain purchase intention and actual use. Includes demographics for multi-group analysis.
- N
- 400
- LVs
- 6
- MGA
- Age · Device
Venkatesh et al. (2012)
Open case study JD-R
Employee engagement (Job Demands-Resources)
HR / Organizational Behavior
Job demands and job resources affect satisfaction and performance via engagement. A classic mediator model with five constructs.
- N
- 350
- LVs
- 5
- R²(JP)
- 0.40
Bakker & Demerouti (2017)
Open case study TAM + Privacy
Adoption of digital health apps
Healthcare / mHealth
PEOU, usefulness, health consciousness, and privacy risk drive attitude and behavioral intention. A negative privacy path makes for a great f² effect-size example.
- N
- 380
- LVs
- 6
- Effect
- PR → BI negative
Davis (1989); Sun et al. (2013)
Open case studyCloning requires a free OpenPLS account.
Open Source
An engine you are allowed to inspect.
Reviewers and dissertation committees want to know exactly how your numbers were produced. With OpenPLS, the answer is simple: here is the code.
Engine is Open Source
The computation engine (loadings, paths, bootstrap, fit) is open: GPL-3.0, reproducible, auditable. No vendor lock-in for your methods section.
Continuously extended
IPMA, PLSpredict, moderation, FIMIX segmentation, plus Newton and PCA inner-weighting schemes are shipped. Every extension lands in the engine first, then in the app.
Validated against references
A growing matrix of example models is checked against established reference values from the PLS-SEM literature: loadings, paths, R², SRMR. Per-case status is public.
Pricing
Open today. Open forever.
The web app is free to use. The engine is open-source under GPL-3.0 and always will be.
Cloud
Free
Everything you need for your research. Hosted by us.
- Unlimited models and datasets
- All reports & exports
- Visual path editor
- Version comparison
Self-host
$0 GPL-3.0
Engine as a Docker container and Python library. On your server, inside your network.
- Full engine functionality offline
- CLI and Python API
- Data sovereignty for clinic / industry
- Community support on GitHub
Consulting
On request Custom
Custom adaptations, training and methodological advice for research groups.
- Methodological coaching
- Custom models / plugins
- On-site workshops
- Priority support