Build Sensitivity Analyses with AI
Sensitivity and scenario analysis in minutes, not hours. Point Compound at your model and get data tables, tornado charts, and scenario cases built on real Excel formulas you can trace and audit.
Trusted by 1,000+ finance professionals across PE, VC, growth equity, real estate, and hedge funds.
How AI Sensitivity Analysis Works
Upload your model
Drop in your existing Excel model. Compound reads the structure, finds the drivers and outputs, and works with your real formulas.
Choose the variables and outputs
Tell Compound which inputs to flex and which outputs to track. Ask for a one-way or two-way data table, a scenario set, or a tornado chart across the key drivers.
Get auditable tables in Excel
Receive data tables, scenario cases, and tornado charts built on real Excel formulas, wired to your model so you can trace and extend every number.
Why Sensitivity Analysis Is Tedious to Build
The mechanics of data tables and scenarios are fiddly, and one wrong reference can quietly break the whole grid.
Data tables are fiddly to wire
Setting up a two-way data table means arranging inputs and outputs just right and pointing the table at the correct driver cells. A single misplaced reference produces a grid of wrong numbers that looks fine.
Every new cut is a rebuild
Want to swap the axes, add a third variable, or track a different output? In a manual workflow that means rebuilding the table, relinking cells, and reformatting from scratch.
Tornado charts take real work
Ranking drivers by impact requires flexing each one independently, capturing the output swing, sorting the results, and then building the chart. Doing it by hand for a dozen drivers is slow.
Hard to trust under pressure
Sensitivity output often lands in an IC deck or a pitch. If the tables were assembled by hand at the last minute, it is hard to be sure every cell traces back to the model correctly.
Why Compound Excels at Sensitivity Analysis
Real formulas, wired to your model
Compound builds data tables and scenario cases with real Excel formulas linked to your model's drivers, not pasted values. Change an assumption and the analysis updates the way it should.
Any cut in plain English
Ask for a two-way table, a new variable, a different output, or a tornado across the top drivers. Reframe the analysis in conversation instead of rebuilding the grid each time.
Auditable end to end
Every cell traces back through real formulas to the input that moved it, so the tables hold up when they reach an investment committee or a client.
From your model to a full sensitivity view
What you can upload
- Existing Excel models (LBO, DCF, operating)
- The drivers and assumptions to flex
- The outputs to track (IRR, EPS, valuation)
- Scenario definitions (bull, base, bear)
What Compound produces
- One-way and two-way data tables
- Scenario cases side by side
- Tornado charts ranking driver impact
- Real formulas wired to your model
Why teams switch from manual data tables
For sensitivity work, the difference is being able to re-cut the analysis freely while every number still ties back to the model.
Works on your model
- Manual Excel data tables
- You wire every reference
- Generic AI tools
- Ignores model structure
- Compound
- Reads your model and its drivers
Formulas vs values
- Manual Excel data tables
- Formula-based but manual
- Generic AI tools
- Static pasted values
- Compound
- Real formulas linked to your model
Re-cutting the analysis
- Manual Excel data tables
- Rebuild the table each time
- Generic AI tools
- Regenerates from scratch
- Compound
- Swap variables in conversation
Tornado charts
- Manual Excel data tables
- Flex and sort each driver by hand
- Generic AI tools
- Not supported reliably
- Compound
- Ranked driver impact and chart
Auditability
- Manual Excel data tables
- Prone to broken references
- Generic AI tools
- Opaque numbers
- Compound
- Every cell traces to its driver
Built for teams that stress-test every model
Private Equity
Sensitize entry and exit multiples, leverage, and growth on your LBO. See how IRR and MOIC move across cases without rebuilding the model by hand.
Investment Banking
Produce football fields and valuation sensitivity tables for pitch and deal materials, with data tables built on the model's own formulas.
Equity Research
Flex margin, growth, and multiple assumptions to frame bull, base, and bear price targets, with every case traceable back to the driver that moved it.
Corporate Development
Test deal assumptions and synergy cases on your model, and see accretion, dilution, and returns react across a full grid of inputs.
Upload a model and get sensitivity tables in minutes
Frequently Asked Questions
Yes. Upload your Excel model and Compound reads its structure, identifies the drivers and outputs, and builds the sensitivity analysis on your real formulas rather than a rebuilt copy.
Yes. Ask for a one-way or two-way data table across the variables and output you choose, and Compound wires it to the correct driver cells so the grid updates when your assumptions change.
Yes. Compound flexes each driver, captures the output swing, ranks the drivers by impact, and builds a tornado chart so you can see which assumptions move the result the most.
Real formulas. Every data table, scenario case, and calculation is built with live Excel formulas linked to your model, so you can trace each cell and extend the analysis yourself.
Yes. Define the assumption sets for each scenario and Compound lays the cases out side by side, showing how your key outputs respond across them.
Yes. Swap an axis, add a third variable, track a different output, or widen the ranges, all in the same conversation. Each request builds on the existing analysis instead of starting over.
Compound handles LBO, DCF, operating, merger, and other financial models. As long as the drivers flow through to the outputs, it can sensitize the inputs you care about.
Yes. Because everything is built on real formulas wired to your model, every number in the tables traces back through the calculation to the input that moved it.
Compound is SOC 2 Type II certified with AES-256 encryption at rest and in transit. Your models are never used for model training, and VPC deployments are available for teams with the highest security requirements.