Shorten time at the gym, maximize Move goal
Problem Definition
For this upcoming quarter, we wanted to focus on improving our workout routine to shorten time in the gym, maximize progress towards daily move goals and reap the health benefits of increased energy needed to tackle a busy schedule.
It’s predicted that time savings from less time at the gym and the associated increase in energy could free up about 5 hours a week of productive time. Leading a freelance digital project or UX workshop that requires a 20 hour investment each month, for example, is estimated to be worth $2,500 in additional revenues per quarter. We wanted to target the following in terms of measuring success of a streamlined workout process:
- Track and compare number of days all Activity rings (Move kcal, Stand hrs and Exercise min) were closed using the old vs new workout processes
- Identify the impact of different workout types on daily calorie goals (Move)
- Gain insights to select more effective workouts that reduce time at the gym while increasing effectiveness of each workout
- Confidently raise our daily Move goal from 350 to 450 kcal based on effectiveness of the new process.
The DMAIC Process
Define
Given an increasingly busy schedule, we wanted to find a way to meet our Move target, setup an ongoing process to reduce workout time and raise the daily Move goal from 350 to 450 kcal.
Define y = f(x)
Our y = f(x) had y as the Move data point where x consisted of several variables measuring exercise time and workout type.
Setup the hypothesis (α = 0.10)
H0: Shorter, higher intensity workouts will have no effect on Move goals when compared to regular full-length cardio workouts. (μ1 = μ2).
Ha: Shorter, higher intensity workouts will have a positive impact on daily Move goals when compared to regular full-length cardio workouts. (μ1 < μ2).
Kickoff the project
Created a problem definition worksheet that outlines the problem, business impact and project timeline.
Measure
We leveraged data from an Apple iWatch worn daily to understand how a shift in the workout process could address our issues. iWatch data was extracted using the Apple Health app.
Collect the data
- 15 min cardio session on an elliptical machine (reduced from 45 min in WP1) 10-15 min of strength training
- 10-15 min modified yoga using seated poses and the sauna when available
Refine and measure
- 45 days of Activity data collected for 8.31 to 10.15 (WP1) and 10.16 to 11.30 (WP2).
- Performance measures: Move goal achieved (kcal = 350); # of Activity rings closed; and, % of days all three Activity rings were closed.
Analyze
Exercise time of 30+ min with any single workout or mix of workouts was the strongest predictor of Move goal success. Strength training and yoga also had a statistically significant impact on the Move goal.
Answer the questions
- Does a shorter, higher intensity workout have an impact on the Move goal? Yes
- Which workout type had the most impact on the Move goal? Strength and Yoga
- Which workout type, or mix, has the most potential for Move = 450 kcal? Elliptical only
Reject H0
Exercise time, strength training and yoga were strongly correlated and had a statistically significant impact on the y variable, Move, at α = 0.10.
Improve
Even though we didn’t have scope to improve the process beyond the WP2 findings, we were able to gain some valuable insights and identify some process optimization scenarios for the future.
Predicting y
For our regression formula y = b0 + b1x1 + b2x2 + b3x3 + b4x4 + b5x5, we used the following values to predict several possible scenarios for the future: y = 391.97 + 7.25×1 -76.84×2 – 68.08×3 – 7.72×4 + 19.10×5
Resetting the bar
We feel confident that this new process allows us to raise the daily Move goal from 350 kcal to 450 kcal given any number of scenarios above, and with Exercise time remaining constant at 30 min or more.
Control
Our goal is to make the data extraction process from the health app more streamlined so we can eventually move from manual data entry and analysis to automated dashboard.
For the Control phase, we will leverage the Xbar/R and IMR charts created during our time series analysis of WP1 and WP2 performance. We would take samples of data throughout the year and run our statistical analysis again.
Time Series Analysis
After we implemented WP2, we observed the data points trending slightly up with most falling above the mean or much closer to it than in WP1.
While the process was mostly under control in W1, there is a data point that exceeds the UCL.
Data also had much larger variations versus WP2 from point to point. While still random, there is less variation from point to point suggesting this new process is more predictable than the original workout process.
The new process displays a trend towards consistently higher Move data points.
This trend suggests we’ve found a process that could enable us to successfully raise our Move goal from 350 kcal to 450 kcal without sacrificing Move success rate.
When using our regression formula to explore several scenarios, all scenarios have the potential to get us to the 450 kcal Move goal.
In Summary
We feel confident that this new process allows us to raise the daily Move goal from 350 kcal to 450 kcal given any number of scenarios described previously, and with Exercise time remaining constant at 30 min or more.
Even though we didn’t have scope to improve the process beyond the findings from WP2, we were able to gain some valuable insights and identify some process optimization scenarios for the future.
For our regression formula y = 391.97 + 7.25×1 – 76.84×2 – 68.08×3 – 7.72×4 + 19.10×5, we used the following values to predict several possible scenarios that can be explored in the Improve phase.
We achieved an overall positive business impact by fixing the current workout process over the course of this project. The Control phase will be key to staying on track for the long term.
Moving forward, we will continue to implement our revamped WP2 process, and continue obtaining daily Activity data. To make the measurement process simple, we could take 30 days of data at the end of Q2 and Q4 to see if the process is continuing as expected, and assess whether one of our test scenarios has a significant enough impact that it warrants consideration as part of a WP3 refined process.
