How to Improve Performance When Querying Apple Health Samples in Shortcuts

My home page has a couple of sections which display my health and activity data. This is powered by a Shortcut that runs on my iPhone every day at 9am. I query data from Apple Health, and then trigger a deployment via GitHub actions.

This was working well, until I started exercising a lot more during the summer months. The Shortcut was taking longer and longer to run, and before long started hitting many problems. Over time, it got slower and slower, sometimes running from 9am and into the afternoon. This was destroying my phone's battery and making the device hot for long periods. Eventually, it ended up throwing an error most days.

Initially when I built the shortcut it would run in about 10 minutes. This seemed acceptable, so I didn't put any effort into optimising it. My assumption was that there were so many samples being queried, it was just taking a long time to return any data. But so long as it finished once in a while, I was happy, so I moved on from the problem.

My shortcut is broken down into several individual ones to gather each piece of data, then the main shortcut pipes this data to my website. The general format for each of the shortcuts is to gather samples over the last 30 days and then sum the result. (e.g Get steps in the last 30 days, get distance cycled in the last 30 days, etc)

Earlier this month, a conversation started in the office, prompted by Robb knowing several people who offer services for optimising shortcuts. I mentioned my issue, and we briefly had a look into what my shortcut was doing when it was running. My assumption about the query for health samples taking a long time was wrong, this finished within a few seconds. Oddly the time-consuming part was summing the samples.

This is because samples are snapshots of activity within a small timeframe. A single day could see hundreds of samples added. So looping over these samples to sum them understandably takes time.

The fix to this is simple, reduce the amount of results coming from the Apple Health query. To do this you can group results, I chose to group this by day, reducing my results to 30. This improved the performance dramatically, and is now under 5 seconds

Now collecting my health data and deploying my site takes under 1 minute. My phone's battery is much happier, and I can deploy my site whenever I want.