Monte Carlo simulation, optimization, decision trees, and a full statistics bench: the serious analyst’s toolkit, native to Google Sheets and running entirely in your browser.
Behind this text, a live Monte Carlo simulation of a small profit model is running: ten thousand seeded trials accumulate into an outcome distribution with brass markers at the 10th, 50th and 90th percentiles.
Sortia is an add-on for Google Sheets™ that brings professional decision analysis into your spreadsheet: Monte Carlo risk simulation, optimization with three solver engines, decision trees, forecasting, and 35+ statistical tools. Point a tool at your data and Sortia writes the full report, charts included, back into the same file as new tabs.
Private by design. Sortia can only access the spreadsheet you open it in, computes in your browser, and sends your data nowhere. It is free to install and use; a Pro subscription unlocks advanced features. See pricing.
Ready-made models that answer a real question — open one in Sheets, change the assumptions, run it. Every figure is computed by the actual tool.
Monte Carlo simulation. Tornado sensitivity. Constrained optimization. Decision trees. The instruments that turn a guess into a defensible decision have existed for decades.
They lived in desktop-era risk suites: Windows-only installs with a license key to babysit, licensed per seat at thousands of dollars a year, tied to one machine for one analyst. Nothing you could just open on whatever laptop was in front of you.
Meanwhile, the people who need those answers moved. Founders, analysts, product managers, operators and finance teams do their modeling in Google Sheets: shared, in the browser, on whatever machine is nearby.
And in Sheets, this entire class of tooling simply never existed. You could leave the browser and lose the collaboration, or stay and go without the rigor. Most teams went without.
Sortia is the missing layer — the analyst’s full toolkit, grown natively inside the sheet.
The exact model from the sky above — the sidebar that runs it, and the sheet it writes back.
The sky above runs this exact engine — same Latin Hypercube draws, same samplers, same seed. Its opening cast (seed 42) converges to the report’s digits, to the sixth decimal. Every field explains itself in plain language; the ⓘ talks like a colleague, not a manual.
| A | B | C | D | E | |
|---|---|---|---|---|---|
| 1 | Risk Simulation — D10 | ||||
| 2 | Trials | 10000 | |||
| 3 | Sampling | Latin Hypercube | |||
| 4 | Seed | 42 | |||
| 5 | Engine time | 0.03 s | |||
| 6 | Inputs | 3 | |||
| 7 | There’s a 90% chance D10 lands between 75,385 and 251,150. | ||||
| 8 | |||||
| 9 | Simulation Statistics — D10 | ||||
| 10 | Trials (numeric) | 10000 | |||
| 11 | Mean | 158,640.422253 | |||
| 12 | Median | 155,805.038185 | |||
| 13 | Std Deviation | 53,492.450297 | |||
Distribution — D10
Analyze, model, decide. One place. No hopping between add-ons that don’t know about each other; the simulation, the solver and the statistics bench share one home and one vocabulary.
Results grow into the sheet. Output lands on its own tab, annotated in plain language, so anyone you share the file with can read the answer, not just the analyst.
Every tool ships with a worked example. One click loads it, so the first run never starts from a blank form.
Mark the cells you’re unsure about, give each a shape, and simulate the whole model thousands of times. Read the odds off the distribution, rank what drives them with a tornado chart, or let the optimizer search for the best decision under uncertainty, with rank correlations honored between inputs.
Constrained optimization against your live sheet: linear, nonlinear, or evolutionary when the surface gets ugly, with integer constraints for when reality only ships whole trucks.
Lay out the choice, attach probabilities and payoffs, and roll the tree back to the policy that maximizes expected value, with the risk profile of every branch, not just the winner.
The questions you ask a model before you trust it: seek the input that hits a target, compare named scenarios side by side, sweep one or two inputs across a table and watch the answer move.
Fit the curve real projects actually follow and forecast completion honestly, including the tail that always takes longer than the plan says it will.
Every analysis your team used to leave the browser for, indexed and ready, each with a guide and a one-click example.
These instruments come from the field of decision science and the professional risk suites that banks, pharmaceutical firms, energy traders, and consultancies still pay thousands per seat to license. What that world learned is that almost every quantitative decision comes down to a handful of questions, and the questions don’t much care what business you’re in.
Investment & credit risk · project cost and schedule · insurance losses · demand and revenue ranges.
Production & supply chain · portfolios · staffing and scheduling · routing and logistics.
R&D go / no-go · competitive bids · drilling and exploration · litigation and M&A.
Sales and demand · pricing · budgets · economic indicators and trends.
Quality and process · experiments and A/B tests · survey and research data.
Break-even · scenario planning · tornado sensitivity · goal seeking.
Where it shines
Price a new product · size a project’s cost contingency · choose a warehouse location · rebalance an investment portfolio · staff a call center · forecast quarterly demand · decide whether to drill or wait · set safety-stock levels · value a sealed bid · plan pension funding · route a delivery fleet · A/B-test a pricing change · schedule a fleet of nurses · plan reservoir yield under drought.
If you can lay it out in a spreadsheet, you can model it here. The toolkit is general; the field is yours.
A sample of the decisions Sortia is built to answer. Each is one of its engines, and every tool ships with a worked example one click away in the sidebar.
Two products draw on the same scarce resources, labor and wood. How many of each should the workshop make to maximize profit without exceeding either?
Result: the exact mix, both resources fully used. Loads and solves in one click.
Sixteen courses, eight time slots, professors with blackout times and students with overlapping wish-lists. Fit them together.
Result: a timetable that minimizes professor and student conflicts at once, from a rugged, non-smooth search space.
Spread capital across assets with different returns and correlations. How much of each, to get the most return for a risk ceiling?
Result: the optimal weights on the efficient frontier, given your constraints.
The classic newsvendor: stock out and you lose the sale; over-order and the leftovers cost you. Demand is a distribution, not a number.
Result: the order quantity and the full profit distribution, with percentiles and the chance of a loss.
Every task duration is uncertain and they compound. When will it really finish, and what should the cost buffer be?
Result: a finish-date and budget distribution with P10 / P50 / P90, and the tasks that drive the risk.
A sequence of choices tangled with chance: run a market test, then decide to launch or abandon, each branch with its own odds and payoffs.
Result: the policy that maximizes expected value, rolled back branch by branch, with the risk profile of each path.
Browse three calm groups in the sidebar, or ask the advisor: two quick questions point you to the right one.
One click fills the form with worked sample data. Risk Analysis is a great first run.
Thousands of trials or solver evaluations, computed in your browser. No queue, no upload, no waiting on a server.
Results land on a new sheet, annotated in plain language, legible to everyone in the file, not just the modeler.
The speed is the quiet trick: Sortia reads your model once, then re-evaluates it with its own spreadsheet-faithful formula engine: around a hundred thousand trials a second, where driving the sheet cell-by-cell manages about five. When a formula isn’t supported, it falls back to the tested sheet path rather than guessing.
in-browser engine ~100,000 trials/sec · through the sheet ~5/sec · same math, same answers, ~20,000× the throughput
There is no Sortia server. The simulation engine, the solver, the statistics bench: all of it runs in your browser, against the spreadsheet you already have open. Nothing is uploaded or “sent for processing,” because there is nowhere to send it.
And the math is checkable, not vibes: 344 automated tests run on every release, cross-checked against scipy and statsmodels to machine precision. Deterministic, seedable, verifiable — that’s the point.
// alea — requested permissions (the entire list) https://www.googleapis.com/auth/spreadsheets.currentonly → read & write this spreadsheet, and only this one https://www.googleapis.com/auth/script.container.ui → draw the sidebar you saw above external servers ···· none data uploaded ······· 0 bytes where it computes ··· your browser
alea iacta est — the die is cast.
Sortia is coming to the Google Workspace Marketplace: the analyst’s full instrument, in the sidebar of the sheet your team already shares.
5 engines · 37 tools · 344 tests · 2 permissions · 0 bytes leave the sheet