Get a number you can defend.
Ask in plain words and get the answer, the odds of hitting your goal, and a live model a skeptic can test. Every simulation run has a receipt anyone can re-run. Right inside Google Sheets™.
Two permissions · no Drive access · your values stay in the sheet unless you send them
PROFIT MODEL · D10WHAT THE SHEET SAYS
One cell. No range. No odds.
Pretending to be certain.
Median about $156,000. Plan on $114,000 to be right 80 times in 100; 1 run in 10 lands below $92,700. Price moves it most.
One number is not a forecast.
A forecast should come with its odds.
The answer should be readable by whoever opens the file.
A tool should not cost you your client’s data.
Nothing should be capped by your model’s size.
Every number should be reproducible by seed.
The profit model’s one cell says $162,000. Run its three estimates 10,000 times and 1 run in 10 lands below $92,700. Nothing in that cell says so.
The profit model the add-on ships with: 10,000 runs, seed 42.
Hold us to it. The rest of this page is the evidence.
Sortia checks the math,
whoever built the model.
Building a model got fast. Knowing whether to trust it did not. Sortia is the step between the workbook and the decision.
Generate
You, a template or an AI assistant builds the workbook. Sortia does not need to be the one that writes it.
Compute uncertainty
Estimates become ranges, and the model runs thousands of times on a fixed seed.
Verify
Re-run to the digit, see what drove the answer, and keep a receipt.
Share and govern
Publish the decision and its receipt as a read-only link, never the workbook.
Move an estimate and watch the odds change.
This is the profit model the add-on ships with, running here on the same seeded engine. Drag a guess and the odds re-run.
WILL PROFIT CLEAR $120,000?
76in 100 runs clear
$120,000
Median about $156,000. Plan on $114,000 to be right 80 times in 100; 1 run in 10 lands below $92,700.
10,000 runs, seed 42. Cost stays $28 to $43, most likely $34. The number here is the number the add-on gives for the same estimates.
Every simulation run has a receipt anyone can re-run.
RECEIPT · PROFIT MODEL
- seed
- 42
- runs
- 10,000
- drawn
- spread evenly across each range
- answer
- 76 in 100 clear $120,000
The fingerprints Verify compares
- estimates
- 5b04…719c
- result
- 4d4b…6c5a
Verify re-runs the model in your browser on /proof and compares the result hash. No spreadsheet data is involved.
Start from your question.
All 379 templates ›Nobody searches for a method. They search for the thing they have to decide by Thursday.
One click writes a working model into your sheet, with real numbers already in it. 194 of the 379 run start to finish on free tools.
Your odds, what moves them, what to tighten first.
Every answer explains itself. Free gives you the tools. Pro reads the answer with you.
Treat $156,000 as a coin flip, not a promise.
What this means
Half of the runs put Profit (D10) at or below $156,000, so treat $156,000 as a coin flip, not a promise. Eight runs in ten land between $92,700 and $229,000; plan against whichever end of that range would hurt, not the middle. Price (B3) moves the result most (rank correlation +0.68: higher price, higher profit). Tightening that estimate narrows the range faster than anything else.
Engine reading. Pro adds the model’s read: the same figures in plain words, with the driver and tail weighed against each other.
{"kind":"simulation","trials":10000,"outputs":[{"ref":"D10","p10r":0.595,"p90r":1.471,"meanr":1.019}],"drivers":[{"ref":"B3","rho":0.68},{"ref":"B4","rho":-0.52},{"ref":"B5","rho":0.44}]}
Ratios, shares, counts and cell references only. No cell values, no labels, no names, no formulas.
Read by the model. Only the shape of the result was sent, never your cells.
Every simulation, optimization, decision tree, schedule and critical chain result carries a reading, and so do most of the 38 data-analysis tools. Type the figure you are planning against into a Risk run and the reading leads with the odds of clearing it, counted from the runs. On the free tier the engine writes it, from its own figures, and nothing is sent to a model. AI readings are a Pro feature.
That is the whole payload, on every Pro reading, one click from the card. The words come back with placeholders and your figures are put in here, so a reply that quotes a number the model was never given is thrown away and the engine reading stands. “Written readings of results” turns off for you in one click, and an administrator can turn it off for a whole organization.
What free covers, and what Pro covers.
Every tool in the Data Analysis, What-If and Optimization panels is free, with no run counter and no size cap that Pro would lift. Five engines are Pro: risk simulation, decision tools, project tools, Critical Chain, and optimization under uncertainty. Every free install gets five full-quality runs to spend on those, one allowance shared across all five rather than five for each, and a run only counts once it has produced a report. On the sixth, the run does not start: Sortia names the engine, says the five are used, and offers Pro at $199 a year or a $9 Week Pass for seven days. Everything else keeps working. So Risk Analysis is a Pro engine, and you can run it five times for nothing before you decide.
Report written to “Risk Analysis Report”.
| A | B | C | D | E | |
|---|---|---|---|---|---|
| 1 | Risk Simulation: D10 | ||||
| 2 | Runs | 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 76,097 and 252,053. | ||||
| 8 | |||||
| 9 | Simulation Statistics: D10 | ||||
| 10 | Runs (numeric) | 10000 | |||
| 11 | Mean | 158,651.625959 | |||
| 12 | Median | 155,733.146587 | |||
| 13 | Std Deviation | 53,215.229828 | |||
Distribution: D10
The report Sortia writes back into the same file: the run, the plain-language row, the statistics and the chart, on their own tab.
Every field explains itself in plain language, and every tool opens with a worked example inside its guide. Optimization is free on all three engines, up to 10,000 decision cells on Simplex LP, 100 on Evolutionary and 60 on Nonlinear, and the panel tells you the number before you run. Where an engine has a ceiling it is the same ceiling on both plans: we charge for the depth of an engine, never for the size of your model.
Six panels, one sidebar.
- Data Analysis38 tools: statistics, tests, forecasting, machine learning, matching.
- OptimizationMaximize or minimize with constraints: linear, nonlinear, evolutionary.
- Risk AnalysisEstimates in, odds out, with tornado charts and a decision optimizer.
- What-If AnalysisScenarios, Goal Seek, Data Tables.
- DecisionsTrees, weighted scoring, tradeoffs, 2×2 games.
- Project PlanningCritical path, schedule risk, resource load, earned value.
The data toolbox, 38 tools.
Statistics, forecasting and machine learning, and every result is reproducible: same data, same seed, same engine build, same answer, on any machine. Each one writes a full report back into your file, charts included, on its own tab.
- Describe
- Descriptive StatisticsHistogramRank & PercentileOutlier DetectionDistribution Fitting
- Compare
- ANOVA: Single FactorANOVA: Two-Factor With ReplicationANOVA: Two-Factor Without Replicationt-Test: Pairedt-Test: Equal Variancest-Test: Unequal Variancesz-Test: Two SampleF-Test: Two-SampleChi-Square: IndependenceChi-Square: Goodness of FitMann-Whitney UWilcoxon Signed-RankKruskal-Wallis
- Relate
- CorrelationCovarianceRegression
- Match
- Fuzzy Lookup
- Forecast
- Moving AverageExponential SmoothingHolt-WintersAutocorrelation (ACF)ETS & ARIMA Forecast
- Machine learning
- Neural NetworkLogistic RegressionClassification TreeK-Means ClusteringAssociation Rules
- A/B testing
- Two-Proportion z-TestA/B Test Sample SizeSample Size (General)
- Generate
- Random Number GenerationSampling
- Transform
- Fourier Analysis
New to one of these? Read the plain-English guides to Monte Carlo simulation, decision trees, schedule risk analysis, critical chain planning and optimization under uncertainty.
Or start from the method: statistics in Google Sheets, machine learning, forecasting, optimization and what-if analysis.
Answers to the questions people ask before they pick a method:
- Which statistical test should I use? Which test fits the columns you actually have in front of you.
- t-test in Google Sheets. Which of the three t-tests is yours, and what T.TEST leaves out.
- ANOVA in Google Sheets. Three groups or more, and then which pair is responsible for the gap.
- Is my data normally distributed? Whether the bell-curve assumption matters for the test you are about to run.
- Histogram in Google Sheets. What shape your numbers are in, and what the automatic bins decide for you.
- Data analysis in Google Sheets. The whole Data Analysis bench, one row per analysis, and what each returns.
- Monte Carlo without an add-on. How to build a simulation out of formulas alone, and where that runs out.
Two permissions, and the two Google adds.
Sortia asks for two: the spreadsheet you opened it in, and permission to draw a sidebar. It has no Drive access at all, and your spreadsheet data stays in your sheet unless you send it: there is one Pro export, described on the privacy page, that you have to press a button to reach.
Google adds two more to every Workspace Marketplace install, your primary Google account email address and your basic profile information, so the consent screen and the listing’s Permissions tab show four. Sortia does not use them. Its own code asks Google who you are and gets nothing back, which is why activating a site key has to ask you to type your address instead of reading it.
Under the hood, the odds come from a seeded simulation engine that runs in your browser, and so does the optimizer. The data tools run in Apps Script, attached to your file. Either way, none of your numbers are computed on a Sortia server.
Seed 42 reproduces the run at the top of this page to the last digit, on any machine running the same engine build, which the receipt under every simulation result names. More than 10,000 automated tests run on every commit, checking the engines against published textbook values, published critical values and independent implementations that derive the answer a different way. The privacy policy lists the reading payload field by field.
// sortia · asked for by the add-on (app/appsscript.json) 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 // added by google to every marketplace install See your primary Google Account email address not read by sortia See your personal info not read by sortia drive access ··········· none cell values, formulas ·· never sent where it computes ······ your browser, your file usage events ··········· anonymous, see privacy
Next: a record of how
good your estimates were.
Three waves, each shipped as an honest experiment first. The status on each is the real one.
Live estimate cells
Type an estimate into a cell and the odds formula over it moves when the estimate does. Off by default.
=EST(28, 10.5) · =ODDS(…)
Receipts, Verify, model files
Every simulation run carries a receipt anyone can re-run, and a model file you can open or check on this site.
seed 42 · result 4d4b…6c5a
Describe your decision
Answer four plain questions and Sortia drafts the model, shown cell by cell before anything is written.
Will the office move stay under 40,000?
Buy it for yourself, or for everyone.
The next number you report can carry its odds.
Buying for a group? See team and organization pricing.
Barely used Google Sheets before? Start here, from an empty browser tab to your first answer.
5 Pro engines · 38 statistics, forecasting and machine-learning tools · 379 ready-made models · 2 permissions Sortia asks for · more than 10,000 automated tests
Estimates in, odds out.