Zometric Statistical AI is statistical analysis software that runs in a web browser. It covers what quality, process and development engineers use day to day: graphs, control charts, process capability, measurement system analysis, hypothesis tests, ANOVA, regression and designed experiments. It also covers the specialist methods pharma and reliability teams need: ICH Q1E stability studies, bioassay potency, life data analysis and acceptance sampling to ANSI/ASQ Z1.4 and Z1.9.
Each tool works the same way. You paste data into a spreadsheet grid or upload an Excel file, choose the columns and options, and get tables and interactive charts. Tools come with example datasets to try them with. The results download as a PDF, Word or Excel report with your company letterhead.
Two features help the people who are not statisticians. Assistants ask plain questions about the data and the goal, for example whether the data is continuous or counts, and recommend the chart, test or study with the reason. On most tools there is also a box where you can type a question about the results. The AI answers from the result tables, so its explanation always refers back to the numbers.
Six things engineers and quality teams say about statistics at work, and what changes with Statistical AI.
From the question to the report, every step happens in the browser.
Choose · Tool
Pick a tool from the catalog by family, or answer an assistant's questions and let it recommend one. Each tool links to its help page.
Data · Grid
Paste into the spreadsheet grid, upload an Excel file and choose the sheet, or load one of the tool's example datasets. A sheet named options in the file can set the tool's options too.
Configure · Options
Select the response and factor columns, then set the options the method needs, such as specification limits, subgroup size, run-rule tests and their K values, or the distribution to fit.
Analyse · Results
Results appear as tables and interactive charts. Capability reports can add the AIAG-VDA assessment, and DOE tools continue to the next step, such as optimising the factor settings.
Ask · AI
Type a question in the box under the results. The AI answers in the context of that tool's result tables only.
Share · Report
Download a PDF, Word or Excel report with your letterhead, optionally including the dataset and the options used, or copy a chart as an image.
Each capability is tied to a decision your team is already trying to make.
Thirty-five graphs and summaries — Pareto, box plot, multi-vari, individual value, probability and marginal plots, histograms, scatter and matrix plots, contour and 3D surface, empirical CDF, descriptive statistics, graphical summary and a cause-and-effect diagram. The first look at any dataset, before you commit to a test.
Twenty-nine charts — Xbar-R, Xbar-S, I-MR, I-MR-R/S, Z-MR, P, NP, C, U, Laney P′ and U′, CUSUM, EWMA, moving average, zone, rare-event T and G, and multivariate Hotelling T², MEWMA and generalized variance. Eight run-rule tests, each switchable with its own K, plus Box-Cox and Johnson transformations.
Fourteen tools — machine, normal, non-normal, nonparametric, between/within, automated, multiple-variable and multivariate (MCpm) capability, three capability six-packs, Poisson and binomial capability. Capability reports can add an AIAG-VDA assessment with the class target, stability quadrant and recommended action.
Type 1, crossed (including AIAG format), expanded and nested Gage R&R, linearity and bias, EMP and attribute agreement. Twenty-nine hypothesis tests, twelve ANOVA tools up to general linear and mixed-effects models, and fourteen regression tools including logistic, nonlinear, best subsets and an ICH Q1E stability study.
Definitive screening (up to 48 factors), factorial (2–15 factors), mixture (2–7 components), Taguchi and response surface designs. Nineteen reliability tools, fourteen time-series tools, ANSI/ASQ Z1.4 and Z1.9 acceptance sampling, sixteen power and sample-size calculators, and six bioassay methods.
Assistants ask plain questions about your data and goal, then point you to the right chart, test or study. On most tools you can also type a question about your results, such as whether the process is capable, and the AI answers from the result tables.
Capability reports can end with an AIAG-VDA assessment. It compares Pp and Ppk with the targets in the AIAG-VDA SPC manual for the characteristic class, the study type and the sample size. A sample size between two columns of the table uses the smaller one. The control chart's run rules decide stability, and the two together place the process in one of four quadrants. Only an ongoing study that is shown to be stable reports its indices as Cp and Cpk.
| Study and class · n | Pp / Ppk | Target | Stability | Quadrant |
|---|---|---|---|---|
| Ongoing, major · n 125 | Pp / Ppk1.61 / 1.52 | Target1.33 / 1.33 | StabilityNo signals | QuadrantI |
| Ongoing, major · n 125 | Pp / Ppk1.58 / 1.41 | Target1.33 / 1.33 | Stability3 points flagged | QuadrantII |
| Ongoing, critical · n 125 | Pp / Ppk1.70 / 1.52 | Target1.67 / 1.67 | StabilityNo signals | QuadrantIII |
| Preliminary, major · n 110 | Pp / Ppk1.80 / 1.40 | Target1.73 / 1.38 | StabilityNot verifiable | QuadrantII |
Control charts, capability studies with the AIAG-VDA assessment, Gage R&R and acceptance sampling for incoming lots.
Designed experiments from screening to response surface, regression and ANOVA to find which settings matter.
ICH Q1E shelf-life studies, bioassay potency, equivalence tests and 2×2 crossover bioequivalence.
Life data with right or arbitrary censoring, accelerated life tests, warranty prediction and repairable-system growth curves.
Creates users one by one or in bulk. The subscription sets how many named users and concurrent logins the account has.
Full Catalogue
233 tools across 17 categories. Each opens its documentation — the same help your engineers reach from inside the platform.
Shows how a whole divides into categories using proportional circular slices.
Compares categorical or discrete values using bars whose length reflects each quantity.
Ranks problems or causes by frequency to highlight the vital few driving most issues.
Summarizes a distribution through its quartiles, median, spread, and outliers.
Visualizes where data points concentrate across a two-dimensional space using color intensity.
Plots the relationship between two continuous variables as points on a plane.
Plots two variables against each other with a histogram, boxplot, violin or dot plot on each axis, showing their relationship and each variable's distribution.
Quantifies the strength and direction of linear relationships between variables.
Displays data points across three axes to show relationships among three variables.
Transforms data toward normality before capability analysis or control charting.
Converts non-normal data toward normality for analyses that require it.
Tracks subgroup means and ranges to monitor process stability for variable data.
Monitors the average of subgroups over time to detect mean shifts.
Monitors within-subgroup variation by plotting the range of each subgroup.
Monitors subgroup means and standard deviations for larger-sample variable data.
Monitors within-subgroup variation by plotting subgroup standard deviations.
Tracks individual, between-subgroup, and within-subgroup variation together.
Monitors individual measurements and their moving range for single-value processes.
Evaluates a machine's short-term capability against its specifications per ISO 22514-3.
Determines how well process output meets customer specifications when the data is reasonably normal.
Evaluates process capability when the data does not follow a normal distribution.
Assesses process capability without assuming any specific data distribution.
Compares the capability of several process variables side by side when the data is normal.
Computes one joint capability index for two or more correlated characteristics, treating them as a single multivariate process.
Assesses stability, capability and normality together for data collected as individuals or subgroups.
Checks process stability and capability when variation differs between subgroups.
Computes capability indices for a process with systematic variation between subgroups.
Assesses a gauge's inherent repeatability and bias using repeated measurements of one part.
Evaluates the accuracy of a measurement device across its operating range.
Generates a randomized data-collection worksheet for a Gage R&R study.
Quantifies measurement variation from operators and gauges measuring the same parts.
Runs a crossed Gage R&R with an optional third random factor, such as gauge or fixture, using variance components.
Plots every measurement by part, one panel per operator, as a quick visual check before a full Gage R&R.
Assesses measurement-system variation when every operator measures every part, with data captured in the format of the AIAG MSA manual.
Assesses measurement-system variation when each operator cannot measure every part.
Evaluates a measurement process with Wheeler's EMP method, reporting the intraclass correlation and classing the gauge as a first- to fourth-class monitor.
Tests whether a sample mean differs from a known population mean with known variance.
Tests whether a population mean differs from a target, with known standard deviation, from summary statistics.
Compares a sample mean to a target value when the population standard deviation is unknown.
Tests whether a population mean differs from a target, with unknown standard deviation, from summary statistics.
Tests whether the means of two independent groups differ significantly.
Compares two independent group means using summary statistics rather than raw data.
Tests for a significant difference between the means of two related groups.
Runs a paired t test from the number of pairs and the mean and standard deviation of the differences.
Checks whether two related methods produce results close enough to be equivalent.
Tests whether the means of three or more independent groups differ significantly.
Tests whether group means differ when each group's responses sit in their own column.
Checks whether two or more groups share the same amount of variability.
Shows how the average response shifts across the levels of each factor in a study.
Shows whether the effect of one factor on the response depends on the level of another.
Quantifies how much each source of variation contributes to the total variation of a process or sample set.
Compares group means and their confidence intervals across the levels of a factor.
Compares each group mean with the overall mean against decision limits and flags the groups that differ.
Models a response against several categorical factors, covariates and interactions, with Type III ANOVA, least-squares means and post-hoc comparisons.
Fits a straight, quadratic or cubic line through two variables to show the relationship and its fit.
Quantifies how predictors relate to a continuous response variable.
Generates a data collection worksheet with batch, time and response columns for a stability study.
Estimates how long a product stays within its quality specifications over time.
Estimates pharmaceutical shelf life to ICH Q1E standards with batch poolability, temperature projection, and extrapolation assessment.
Models predictors against an ordered categorical response with three or more levels.
Models predictors against an unordered categorical response with three or more outcomes.
Models how predictors relate to a two-outcome categorical response.
Fits a relationship between two variables that both contain measurement error.
Creates and analyses a definitive screening experiment to find the significant factors among up to 48.
Generates a factorial experiment plan and analyzes factor and interaction effects.
Designs and analyzes experiments where factors are ingredient proportions summing to a constant.
Creates and analyses a Taguchi orthogonal-array experiment to find factor settings that are robust to noise.
Creates and analyses central composite, Box-Behnken or D-optimal designs to model and optimise curved responses.
Compares three or more independent groups without assuming a normal distribution.
Computes Walsh averages, Hodges-Lehmann shift estimates or the Theil-Sen slope from all pairs of values.
Compares two independent groups using ranks instead of means.
Compares the medians of two or more groups by counting values above and below the overall median.
Checks whether a sequence of values is arranged in a truly random order.
Compares three or more related groups using a nonparametric ranked test.
Tests whether a sample median equals a specified target value.
Tests whether a sample median equals a target using signed ranks.
Reduces many correlated variables into a smaller set of uncorrelated components.
Explains the correlations among many variables with a few underlying factors, with optional varimax rotation.
Classifies observations into known groups with linear or quadratic discriminant functions and reports cross-validated accuracy.
Groups rows into a dendrogram by their similarity across several variables.
Groups columns into a dendrogram by their correlation or distance.
Splits observations into K clusters that minimise within-cluster variation.
Maps the association between two categorical variables as row and column points in a few dimensions.
Explores the relationships among three or more categorical variables at once.
Measures the internal consistency of a multi-item scale or test with Cronbach's alpha.
Determines test sample size and duration to prove a reliability target at set confidence.
Determines sample size and duration needed to estimate reliability parameters.
Prepares raw warranty field-return data for reliability analysis.
Models how a repairable system's failure rate changes as it accumulates operating time.
Forecasts expected future warranty returns from a fitted reliability model.
Fits life data at each stress level and relates life to stress with Arrhenius, inverse power law or linear models to extrapolate to use conditions.
Simulates the expected precision of a proposed accelerated life test before it is run.
Fits grouped success-and-trial counts at several stress levels and estimates the stress for chosen response percentages, such as ED50.
Estimates the mean cumulative function of repair or recurrent-event data without assuming a model.
Plots a measured variable against time to reveal trends and changes over a period.
Fits a linear, quadratic, exponential growth or S-curve trend and forecasts future values.
Splits a seasonal series into trend, seasonal and residual components, additive or multiplicative.
Smooths a series with a trailing moving average and forecasts future periods.
Smooths and forecasts a series with no clear trend using one smoothing weight.
Smooths and forecasts a trending series with Holt's level and trend weights.
Smooths and forecasts a series with both trend and seasonality.
Computes lagged differences of a series, for example to remove a trend before further analysis.
Creates and plots a lagged copy of a series beside the original.
Builds AQL inspection plans for pass/fail attribute data.
Creates AQL inspection plans for continuous measured quality characteristics.
Builds an acceptance sampling plan for continuous measurement data.
Compares sampling plans side by side using their operating characteristic curves.
Applies an existing variables sampling plan to a sample of measurements and returns an accept or reject decision.
Computes a single sampling plan (n, c) for defective counts from AQL, RQL and the producer's and consumer's risks.
Compares several attribute sampling plans (n, c) on the same OC, AOQ and ATI curves.
Solves for sample size, detectable difference or power, given the other two, for a 1-sample t test.
Finds the sample size needed to estimate a mean or proportion within a target margin of error.
Finds the sample size a normal tolerance interval needs to reach the required confidence and coverage.
Solves for sample size, detectable difference or power, given the other two, for a 1-sample Z test.
Solves for sample size, detectable difference or power, given the other two, for a 2-sample t test.
Solves for sample size, detectable paired difference or power, given the other two, for a paired t test.
Solves for sample size per level, detectable difference or power, given the other two, for a one-way ANOVA.
Solves for sample size, detectable proportion or power, given the other two, for a 1-proportion test.
Solves for sample size per group, detectable proportion or power, given the other two, for a 2-proportions test.
Estimates relative potency by comparing the parallel log dose-response lines of a test and a standard preparation.
Estimates relative potency from the ratio of dose-response slopes when preparations share a common baseline response.
Combines potency estimates from several independent assays into one weighted result with homogeneity testing.
Estimates effective dose and relative potency when biological responses are binary outcomes rather than continuous measurements.
Estimates relative potency by fitting a four-parameter logistic sigmoidal curve, suited to immunoassays such as ELISA.
Estimates antibiotic potency from agar-plate zones of inhibition using a USP 5+1 cylinder-plate design.
Builds a two-way table of two categorical variables with row, column and total percentages and a chi-square test of association.
Counts each value of the selected columns, with percentages and cumulative totals.
Tabulates descriptive statistics for numeric variables, optionally split by up to two categorical variables.
Generates random data from any of 26 distributions, including Normal, Binomial, Poisson, Weibull and multivariate Normal.
| Zometric Statistical AI vs statistics in a spreadsheet | Spreadsheet | Zometric Statistical AI |
|---|---|---|
| Methods | Basic functions plus formulas you build and check yourself | 233 tools, from control charts to mixed-effects models, ARIMA and Cox regression |
| Choosing a method | Up to the person building the sheet | Assistants that recommend the chart, test or study and explain why |
| Control charts | Limits computed by hand, run rules checked by eye | Eight run-rule tests, each switchable with its own K |
| Capability verdict | Cpk against a number someone remembers | AIAG-VDA targets by class and sample size, with quadrant and action |
| Pharma methods | Not available | ICH Q1E stability, bioassay potency and 2×2 crossover bioequivalence |
| Explaining results | Whoever built the sheet | An AI you can ask about the result tables |
| Reports | Copy and paste into a document | PDF, Word or Excel report with your letterhead |
Paste data into the grid or upload an Excel file and choose the sheet. A sheet named options can also fill in the tool's settings.
Every result downloads as a report under your company letterhead, with the dataset and the options used if you choose. Charts copy as images.
The Workspace exports all its tabs in one file. An Excel export can be loaded again to rebuild the worksheets and tabs and recalculate them.
Straight answers to the questions teams ask before a demo.
Zometric Statistical AI is browser-based statistical software with 233 tools in 17 families: graphical analysis, control charts, process capability, MSA, hypothesis tests, ANOVA, regression, DOE, nonparametrics, multivariate, reliability, time series, acceptance sampling, power and sample size, bioassay, tables and random data. Assistants help you pick a tool, and you can ask an AI about the results.
On most tools there is a box under the results where you can type a question, for example whether the process is capable or which factor matters most. The AI answers in the context of that tool's result tables only. It does not run the analysis or change the numbers.
Use an assistant. There are assistants for control charts, capability, MSA, hypothesis tests, equivalence tests, ANOVA, regression, DOE, nonparametrics, multivariate, reliability, time series, sampling, power and sample size, bioassay, tables, graphical analysis, quality tools and random data. Each asks a few plain questions and recommends a tool with the reason.
Eight run-rule tests for special causes, such as one point beyond K standard deviations or K points in a row on one side of the centre line. Each test can be switched on or off and has its own K. Attribute charts use the first four tests.
Yes. Capability reports can add an AIAG-VDA assessment that checks Pp and Ppk, or Cp and Cpk when stability is shown, against the manual's targets for the characteristic class, study type and sample size. It places the result in one of four quadrants and states the recommended action. You can also enter custom targets agreed with the customer.
Yes. The advanced stability study follows ICH Q1E: batch poolability tests, the extrapolation rules, bracketing and matrixing designs, and Arrhenius projection to other temperatures. A worksheet tool sets up the batch, time and response columns.
Attribute AQL plans follow ANSI/ASQ Z1.4 and variables AQL plans follow ANSI/ASQ Z1.9, both also citing IS 2500. You can also create and compare your own attribute and variables plans on OC, AOQ and ATI curves, and get an accept or reject decision for a lot from its measurements.
Paste data into the spreadsheet grid or upload an Excel file, or start from an example dataset. Results download as a PDF, Word or Excel report with your company letterhead, and charts copy as images.
Zometric's modules share a single data backbone — adopting more of them compounds the signal-to-noise advantage.
Catch process deviations before they become defects.
Every gauge calibrated on time, signed off by the right people and proven by MSA.
Plan, document, execute and archive every experiment in one place.
From idea to financial impact — structured CI pipeline.
A 30-minute discovery call. A 2–3 week pilot on real factory data. Then scale at your pace.
