Bayesio: our Bayesian simulator goes online

October 2026 - Socio Data Management's Bayesian simulator gets a makeover and becomes Bayesio.

We already talked about it in January 2025: Bayesian networks answer questions that traditional statistics often struggle with. Our historical simulator, used with our clients for years, has now been fully rebuilt as a web application: Bayesio.

A simple question

"If this aspect of the customer experience improves - or deteriorates - what effect does it have on the overall score?"

That is the question Bayesio answers. It relies on the statistical model built from your satisfaction survey: each survey theme becomes a slider. Noise in the room, booking lead time, the clarity of an offer… Move a slider and immediately see the expected effect on your overall indicator: NPS, average satisfaction score, share of respondents or volume.

The goal: identify the levers that really matter, rather than spreading action plans across every theme of the survey.

A network, not a list

The strength of a Bayesian network is that it models the influences between themes. Improving a restaurant's setting lowers perceived noise, which in turn improves the perceived speed of service: Bayesio propagates the effect of a lever through the whole network, and shows for each theme what "pulls" it and what it "acts on".

Two simulation modes are available:

• Observation: the model is conditioned on the chosen values, without changing its structure;
• Intervention: Bayesio simulates a true causal intervention on one or several themes, up to ten at once.

Results to decide on

• A gauge continuously compares the initial value with the simulated score;
• A table of ranked effects, with 90% confidence intervals;
• The contribution of each theme and its standardized total effect;
• A reading of the levers as key, plus, basic and secondary elements, to tell apart what to watch, what differentiates, what is expected by default and what weighs little.

A high-impact theme that is still poorly rated? That is your action priority.

Built for teamwork

Filter by segment (country, target group, period), save your scenarios - "noise and booking action plan" —, share them with members of your organization or export them to CSV. Access is secure and partitioned: each client only sees its own studies.

Discover Bayesio

Bayesio is offered to clients of our satisfaction and recommendation studies. To learn more, see the Bayesio page or contact us for a demo on your own study.