Skip to main content

Key Concepts

Moveo One is the prediction layer between your data and what happens next. It captures how people actually use your product, turns that behavior into forward-looking signals, and gives you four ways to act on them.

Raw data in, predictions outDirect link to Raw data in, predictions out

The model is simple:

  1. Capture — the SDKs (or an integration like Segment or PostHog) record real user interactions — taps, scrolls, clicks — and group them into sessions.
  2. Understand — after each session, Moveo One computes cognitive and behavioral features (hesitation, rhythm, navigation effort) on top of the raw events.
  3. Predict & act — those features feed analytics and real-time predictive models, scored live (typically within ~100 ms) so you can react while the user is still active.

It sits on top of your existing analytics (PostHog, Amplitude, Mixpanel, …) — no schema migration and no data-science team required.

Want the full vocabulary — contexts, sessions, semantic groups, events, metadata? See Data Model & Tracking.

One platform, four pillarsDirect link to One platform, four pillars

🔮 PredictDirect link to 🔮 Predict

Continuous ML scoring gives you a calibrated probability per user for the outcomes you care about — conversion, churn, activation — updated in real time as the session unfolds.

Predict — prediction insights with probability distribution and volume

🧠 ExplainDirect link to 🧠 Explain

Predictions aren't a black box. Moveo One surfaces the reason behind every prediction — which attributes and behaviors move the outcome most (timezone, platform, referrer, UX friction), in plain terms.

Explain — ranked metadata importance behind a prediction

🧪 ExperimentDirect link to 🧪 Experiment

Prove impact, don't assume it. Built-in A/B testing runs a control group without the model and a test group with it, then compares the target-reached rate between them.

Experiment — A/B test results comparing control and test groups

🌀 SimulateDirect link to 🌀 Simulate

Test changes before you ship. Quantum runs forward simulations against a calibrated population of synthetic users that behave like your real ones — measuring conversion, churn, or any custom outcome against your baseline.

Simulate — Quantum cohort runs with average and best conversion

Each run streams a live agent feed: every synthetic user navigates your product step by step and ends converted or bounced, showing exactly where it got stuck.

Simulate — per-agent run detail with action timeline

Where to go nextDirect link to Where to go next