Stop chasing
metrics that lie.
Upload your data, pick the metric you care about. Causal runs causal discovery to surface the actual drivers—the latent factors correlation analysis will never show you.
Discover what's driving your metric →Upload a CSV. Results in minutes. No data science PhD required.
Why teams switch from correlation dashboards
73%
of clinical findings don't replicate
because the cause was misidentified
4–6 wks
average time analysts spend
forming and testing a single hypothesis
~60%
of reported metric shifts trace to
confounders, not the variable changed
Minutes
to get a causal DAG from
your data with Causal
How it works
From messy CSV to
causal clarity
Drop in your dataset. Pick your metric.
CSV, spreadsheet export, or direct warehouse connection. Select the target metric—patient dropout, response rate, adverse event frequency, treatment effect size. Causal handles mixed data types, missing values, and messy real-world schemas automatically.
product_events_q3.csv
84,201 rows · 23 columns
Target metric
The causal DAG, automatically inferred.
Causal runs PC, FCI, and LiNGAM algorithms against your data—pruning spurious edges, flagging confounders, and surfacing latent variables you never knew to look for. Not a correlation heatmap. An actual directed acyclic graph with statistical confidence bounds.
Individual-level impact, not just aggregate averages.
The aggregate effect hides the real story. Causal breaks out causal impact by segment—plan tier, cohort, geography, device—so you can see which populations respond 4× stronger and where your intervention will actually move the number.
Driver: Dose timing — Causal effect on dropout
Built for real messy data
Everything correlation analysis gets wrong,
Causal gets right.
Latent variable detection
Surfaces hidden confounders—variables driving your metric that don't appear in your dataset at all.
Confounder flags
Every correlated variable is tested for confounding. Stop acting on signals that vanish when you control for the real cause.
Segment-level causal effects
Aggregate averages hide heterogeneity. See which segments are most causally sensitive—and focus your intervention there.
Intervention ranking
Not just "what causes what"—ranked by estimated causal effect size so your team knows exactly where to experiment next.
No setup, no SQL
Upload a flat file. Results in under 5 minutes. No pipeline, no warehouse query, no data engineer required for a first analysis.
Confidence bounds included
Every causal edge comes with statistical confidence. Know when the data is uncertain so you don't over-rotate on weak signals.
Who it's for
Built for the analyst who knows
correlation isn't enough.
Causal is for biostatisticians and data science leads at pharma and biotech organizations who need to know—with rigor—what's actually driving their clinical outcomes.
Clinical Trials
Identify the causal drivers of patient dropout, response variability, and adverse events—not just the variables that correlate with outcomes.
Pharma Analytics
Surface heterogeneous treatment effects across patient subgroups. Find the segments where your intervention actually works—and where it doesn't.
Biostatistics
Generate regulatory-ready evidence with sensitivity analyses, robustness testing, and FDA/EMA-formatted reports that withstand scrutiny.
87%
of analysts report their top "correlated" variable was not a true causal driver after rigorous testing
3–8×
faster to first-priority hypothesis compared to manual EDA and feature importance workflows
Zero
data engineering required. If you can export a spreadsheet, you can run a causal analysis.
Ready to stop guessing?
Find the real lever.
Not the one that just
looks like it.
Upload your dataset. Pick your metric. In minutes you'll have a causal graph, ranked drivers, and segment breakdowns your team can act on today.
Discover what's driving your metric →Works with CSVs from any stack · Results in minutes