At North London Foundation Trust, we’ve been using Statistical Process Control (SPC) charts to better understand performance and guide improvement. With initial guidance from the Making Data Count team, our reports were redesigned to use SPC, and dashboards were built to highlight signals of non-random variation. Suddenly, those orange and blue icons began to shift how we talked about data.

Over a year on, we wanted to go deeper. What are the intricacies of applying SPC rules? Where does the science end and the art begin? Here’s what we learned from our latest conversation with Adam Smith and Karen Hayllar from the Making Data Count team.

SPC in Practice: Setting Limits, Spotting Signals

One common worry is whether recalculating baselines or moving control limits changes the story. The team’s advice: don’t overcomplicate it — the signals matter most.

• Four rules are enough — Anything more risks false positives.
• Focus on patterns, not tiny shifts — Shifts and trends hold up, even if the mean moves.
• Context is everything — An “outlier” could be Christmas, a missing piece of kit, or something serious.
• Baselines have a purpose — In improvement projects, they matter. In monitoring, chunking into 2–3 years keeps things relevant.
• Stay focused on the now — Long historic charts are rarely helpful if nobody recalls what happened five years ago.

Karen reminded us that SPC is there to spark the right conversations, not to provide all the answers. 

Recalculations: When the Data Tells You It’s Time

So, when should you recalculate your mean and limits? The frustrating truth: there’s no formula. It’s a judgement call on keeping the story meaningful. The team shared some of these top tips with us.

• Look for big shifts — If the chart no longer reflects reality, it’s time.
• Wait for stabilisation — Give the system chance to settle (8+ points is a good rule of thumb).
• Show history — Always keep the “before” and “after” visible to tell the full story.
• Be pragmatic — 15–20 points is ideal, but with monthly data you may need to move sooner.
• Stay consistent — Whatever rules you adopt, apply them consistently.

More information on process limit recalculations can be found here:  https://future.nhs.uk/connect.ti/MDC/view?objectId=166795429 

Practicalities: Power BI, Data Points & Rolling Averages

The conversation also turned to everyday challenges of working with SPC in practice:

Power BI tools — The team shared some great new developments within Power BI to automate recalculations. It’s on GitHub — and worth exploring if you want features like multiple targets. https://github.com/AUS-DOH-Safety-and-Quality/PowerBI-SPC

• How many points is too many? — More than 60 can be unwieldy. For monthly data, 2–3 years is usually the sweet spot. For daily data, compare Mondays with Mondays (and so on) to avoid noise.
• Rolling averages — Great for benchmarking, poor for decision-making. They smooth away the very variation we need to see. Use in-month data for managing services.
• Shifting targets — National metrics often change yearly. Excel can handle it; Power BI needs development. The advice: raise requests on GitHub and keep pushing for flexibility.

Top Tip: Always come back to usefulness: if a chart isn’t helping you make better decisions today, it needs rethinking.

What became clear from our conversation is that SPC isn’t a rigid science. Yes, there are rules, but the real skill lies in applying them thoughtfully. As the Making Data Count team put it:

“SPC is more art than science — it’s about using data to tell the story.”

Thankyou to Karen Hayllar and Adam Smith for their insights, and to Eddie Short and Ram Kuttykrishnan for asking the questions that grounded the discussion in real practice. 

The Making Data Count Team run regular training sessions to help you understand their approach with practical examples and application of SPC tools. Find out more about their training sessions here: https://future.nhs.uk/MDC/groupHome