Custom visual

Control Charts Pro Coming to AppSource

Statistical process control across every line, ward, site or product at once — one faceted grid instead of forty copy-pasted charts.

Get in touch See the features Limitations

A two-by-two grid of individuals control charts, one per coating line, each with its own centre line and limits: one stable line, one with a single point far above its limit, one with a step change and one drifting upwards. A key names the red and amber markers. — Four production lines, four different stories, one frame — and control limits recalculated inside every panel. Red is beyond a limit, amber is a pattern rule; the key sits at the end of the caption.

Why this exists

A control chart answers the question every operational meeting actually asks: is this change real, or is it noise? Power BI has no control chart. The free ones handle a single series, so monitoring twenty wards means twenty charts, twenty sets of limits, and twenty places to make a mistake. Native small multiples has not changed since November 2022 — bar, column, line and area only, capped at thirty-six panels, one facet field, and the service still forces a shared Y axis.

Features

●Small multiples, done properly. One chart per panel value, with limits recalculated inside each. Auto or fixed columns, panels that scroll past the viewport, and four Y-axis modes: free per panel, shared across all, shared within a row, or shared within a column. Free-per-panel compares behaviour; shared compares level. Most tools force one and call it a feature.
●Every chart type that matters. Individuals (X), XmR with the moving-range chart underneath, run charts, and the attribute charts — P, C and U — with limits that correctly vary point by point as the denominator changes.
●Sigma your way. Estimated by MR̄/1.128, median-MR/0.954 or sample SD — your choice, named on the chart so a reviewer can reproduce it.
●Signals you can explain. Nelson rules 1–8 with configurable run lengths, or one click for the NHS Making Data Count seven-point rules, or WECO, or your own combination. Hover a flagged point and the tooltip names the rule in plain English.
●Rebasing that tells the truth. Freeze limits after a baseline of N points, or let the visual detect a step change and split the series into phases with separate limits and a dashed boundary. Drop in a Phase field to control it explicitly. Limits calculated from a process that has already shifted are worse than no limits at all.
●Capability built in. Specification limits, target lines, and Cp/Cpk in the panel title.
●Panel sorting. Order panels by name, by mean, by steepest trend, or most-violations-first — so the panel that needs attention is the one you see.
●Verified against an independent implementation. Every centre line, sigma estimate, control limit, rule hit and moving-range limit was checked against a clean-room implementation built from textbook formulas, on datasets with a planted spike and a planted step change.

A single control chart with a tooltip open on a flagged point showing the value, centre line, limits, sigma distance, moving range and the three rules that fired, each by number and name. — Every signal explains itself: the tooltip names each rule that fired, so the person who has to act on it understands why.

Two individuals charts with a moving-range strip beneath each; the second line breaches its moving-range limit on two consecutive points. — XmR adds the moving-range chart beneath each panel. A single spike shows up as two consecutive breaches — the signature that tells you it was one event, not a shift.

The same line twice: above, limits frozen on the first fifteen weeks with fourteen of the next fifteen points above the upper limit; below, limits recalculated after the step, giving two phases each in control. — Above: limits frozen on a fifteen-week baseline, and after the step change fourteen of the next fifteen points sit above the limit — unmissable. Below: the same data with limits recalculated after the step, two phases, each in control.

P charts of infection rates for four wards, where the control limits step up and down month to month as the number of patients at risk changes, with one month flagged well above its limit. — A P chart’s limits must move with the denominator. Here patients at risk change every month, the limits step accordingly, and the outbreak still stands out.

Where it fits

NHS and healthcare quality indicators, manufacturing process control, turnaround and SLA times, infection and incident rates — any measure where you need to stop reacting to noise.

Honest limitations

We would rather you know these before you buy than after.

One measure per visual. Multi-measure panels are planned, not shipped.
The X axis is ordinal. Points are equally spaced rather than placed on a true time scale. This matches standard SPC practice, where the subgroup is the unit — but it means gaps in time are not drawn as gaps.
X labels show first, middle and last per panel so a grid stays readable.
Rules 4, 7 and 8 need long series and will rarely fire on short ones.
Cp/Cpk needs at least one specification limit and an individuals-type chart.

Learn it in fifteen minutes

The Control Charts Pro tutorial explains what a control chart tells you, how the limits are calculated and how to check them by hand, the detection rules, XmR, baselines and rebasing, and P charts — with two sample datasets you can download.

Questions about Control Charts Pro?

Pricing, a feature you need, whether it fits your model — ask, and a person who wrote it answers within two business days. It is in AppSource certification now.

Get in touch Read the tutorial