Near the end of July, I bought a walking pad – a miniature treadmill designed purely for walking – and it’s been an absolute game-changer for increasing my step count.

But why?

Over the last few years, I’ve become increasingly sedentary outside the gym. While weight training is fantastic, and I’ll forever shill for Big RT, it won’t make up for an otherwise sedentary lifestyle. And there’s a ton of data at this point showing an association between step count (i.e. movement) and a reduction in the risk of all-cause mortality and cardiovascular disease, and ignoring that would be pretty silly.

And yes, I know what you’re thinking––“Why not just walk outside?”

Well, for one thing, have you ever gone for a walk with a toddler? Because it’s not a walk. It’s a hostage negotiation, and the hostage is your sanity. If Amara’s in the stroller, she’s screaming, “OUT!” If you take her out of the stroller, what should have been a 30-minute walk becomes a three-hour toddler-led forensics session, where she’ll stop every second to inspect every rock, stick, and speck of dirt. “*Oh but isn’t it just magical seeing the world through her eyes?*” Shut up, Susan––it’s a fucking stick, not the Northern Lights.

The walking pad also significantly reduces the friction of getting my steps in. I can jump on the walking pad for 20-30 minutes throughout the day, regardless of how busy I am or how bad the UK weather is (historically, my step count falls off a cliff whenever winter rolls around).

Seeing that I’d also started my mini-cut, I was curious to see how increasing my step count would impact my energy expenditure––would I lose weight faster if I significantly bumped up the number of steps I took each day? I know from working with clients that getting someone to walk a bit more almost always speeds up their progress while allowing them to eat a bit more food. So I knew increasing my steps *would* have an impact, but I wanted to quantify *how much* of an effect it would have.

Additionally, I wanted to test the accuracy of my Apple Watch. I’ve written about this topic previously here and here, but the research is pretty unequivocal: The current generation of smartwatches sucks at estimating the number of calories you burn. Even so, I had the walking pad and my watch, and I could test things in a very controlled environment, so why not?

And that’s what the rest of this article will be about. After walking over 10,000 steps every day for nearly two months, how much did my energy expenditure increase, and how accurate was my Apple Watch in estimating the number of calories I burned?

Let’s find out.

**A quick note before we begin: I’ve intentionally removed my body weight from graphs to avoid people falling into the comparison trap. When I need to mention body weight for illustrative purposes, add context to a statement, or provide an example, the weight numbers are made up. This doesn’t impact the interpretation of the data since the graphs are still informative for the point I’m making. Cool? Let’s get started then.*

## The approach to the problem

When researchers test the accuracy of smartwatches, they’ll compare the smartwatch to a gold standard measurement like indirect calorimetry, where the researchers measure the amount of oxygen consumed and carbon dioxide produced by the participant, which can be used to calculate energy expenditure (since the body uses oxygen to burn calories and produces carbon dioxide as a byproduct, researchers can analyse the gas exchange and apply formulas to work out how many calories a person is burning).

Unfortunately, I didn’t have access to this, so I had to take a different approach: I used changes in my body weight and the number of calories I consumed to get a rough idea of my total daily energy expenditure (TDEE), which was averaged over 14 days, and this was compared to the TDEE data from my Apple Watch.

**Why? **We know changes in body weight are dictated by calories in vs. calories out, so by looking at changes in my body weight alongside the number of calories I consumed, I can estimate the total number of calories I’m burning.

To illustrate: If I consume 2,000 calories per day and lose 1 lb per week, that means I’m in a 500-calorie daily deficit, so my TDEE is 2,500 calories.

As for the Apple Watch, I just added the resting energy expenditure (REE) to my Active Calories. So if my REE was 1800 according to my watch and I burned 700 calories, my TDEE for the day would be:

Note: This isn’t my ‘true’ TDEE since I didn’t account for the calories burned digesting/storing food (i.e. the thermic effect of feeding), which is roughly 10% of total calories consumed. But I didn’t want to skew the Apple data, so I ignored that for these calculations.

## Did increasing my steps increase my TDEE?

As you might expect, doubling my step count positively impacted my TDEE, as shown in the image below. During the first half of the mini-cut (when I didn’t have the walking pad), my TDEE averaged about 2300 calories/day. During the second half (when using the walking pad), my TDEE averaged just under 2500 calories/day.

As for my step count: Pre-walking pad, I averaged ~5,600 steps/day, and post-walking pad, I averaged ~11,100 steps/day.

This was the final result: I went from about 20% body fat at the start of the mini-cut (after gaining for ~14 months) to around 12% by the end.

I also compared the difference in weight loss, caloric intake, and estimated TDEE during this mini-cut to my cut last year, where I walked far less. You can see a comparison between the two diets in the table below:

Year | Steps (avg) | Kcals (avg) | Avg Loss/wk | Avg deficit | Est TDEE |
---|---|---|---|---|---|

2023 | 5938 | 2075 | -0.34 | -170 | 2239 |

2024 | 8320 | 2054 | -0.72 | -360 | 2414 |

Change | 2382 | -21 | –0.38 | -190 | 175 |

Looking at the table, we can see that while I consumed roughly the same number of calories during both diets, my average weekly weight loss more than doubled this time around. Additionally, my estimated TDEE was about 170 calories higher this year than last year.

Before moving on, I want to highlight two things.

- The only variable that changed between these two diets was my step count. My training program did change slightly, but not enough to significantly impact my energy expenditure. So, we can be fairly confident the increase in my TDEE is primarily due to the increase in steps.
- At first glance, it seems like my energy expenditure only increased by 175 calories despite an increase in my step count, which, to be fair, is still a sizeable increase. However, it’s important to remember that I was in a caloric deficit, so my metabolic rate would have been slightly lower due to metabolic adaptation. Looking back on my data during my gaining phase, when my metabolism would be more or less back to baseline, my TDEE averaged about ~2300 calories, while my step count was around ~4500/day. I’ve been at maintenance for just over a week now, and I’m currently maintaining my weight and end-of-cut body composition while eating around 2500 calories and averaging over 10,000 steps/day. So, while not super precise, I feel pretty confident saying that doubling my step count increased my TDEE by roughly 200 calories per day.

Just to make sure I’m not pulling numbers out of my ass, I cross-referenced that 200-calorie estimate with the LCDA (Load Carriage Decision Aid) formula—fancy name, I know, but it’s one of the better tools out there for calculating calories burned while walking since it takes things like weight, speed, and even terrain into account. Plus, it factors in the intensity of your movement, making it a bit more legit for real-world situations. ^{1 2 }

After plugging in my current body weight, walking speed, and duration (60 mins; the amount of time I was walking each day), I get 358 kcals––so walking at a pace of 3.7 mph for an hour at my body weight burns roughly 360 calories.

However, we need to make two adjustments to this figure:

- We need to deduct my resting energy expenditure during the 60-minute walk to isolate the number of calories burned during walking. In this case, my RMR works out to 72 kcals. That is, if I sat around and did nothing for 60 mins, I’d burn 72 kcals due to all the metabolic processes occurring to keep me alive.
- We also need to account for energy compensation since increasing activity levels doesn’t mean your energy expenditure will increase by exactly the same amount, and the compensation is higher when someone is in a deficit. For this example, I’m going with the 30% average found in the research.

After adjusting for my RMR and energy compensation, I get a net caloric burn of 200 kcals, which lines up exactly with the 200-calorie estimate I mentioned above.

**In sum, by comparing my calorie intake to how much weight I lost, on average, per week, I found doubling my step count from ~5,600 steps/day to ~11,100 steps/day increased my TDEE by around 200 calories (from ~2300 calories/day to ~2500 calories/day). When comparing my data to last year’s diet, I ate about the same amount of calories, but my weight loss was more than double this time around. **

With that out of the way, let’s move on to the next part of this experiment: how accurate was my Apple Watch in estimating my calorie burn?

## How Reliable is the Apple Watch for Calorie Tracking?

Before diving into how accurate my Apple Watch was for reporting calorie burn, I should probably explain how I went about testing this. So strap the fuck in, bucko, because we’re about to get nerdy.

Here’s the basic approach: Let’s say my Apple Watch reports that my resting energy expenditure (REE) is 1800 kcal, and I burn an additional 700 calories through activity (i.e. ‘Active Calories’ in Apple lingo).

According to the watch, my total daily energy expenditure (TDEE) is 2500 kcal. If I consume 2000 calories that day, I should be in a 500-calorie deficit, meaning I should lose approximately 1 lb per week or 0.14 lbs per day.

Using this information, I can calculate my predicted weight for each day by subtracting the predicted daily weight loss (in lbs) from the previous day’s weight. Then I repeat this process for the entire diet, comparing the predicted weight––based on Apple Watch data––against my actual scale weight.

Example: Let’s assume that I weighed 179.7 lbs today and consumed 2125 kcals, and my Apple Watch estimates my TDEE to be 2228 calories. Based on this data, I’m in an energy deficit of ~103 calories (2125-2228 = -103), which translates to a daily weight loss of -0.03 lbs.

You can see what this would look like over a week in the table below.

Day | Scale weight (lbs) | Calorie intake | TDEE (Apple) | Deficit/ Surplus (kcals/day) | Weight change (lbs/day) | Predicted weight (lbs) |
---|---|---|---|---|---|---|

1 | 179.7 | 2125 | 2228 | -103 | -0.03 | 179.7 |

2 | 179.3 | 1556 | 2407 | -851 | -0.24 | 179.5 |

3 | 178.8 | 1800 | 2227 | -427 | -0.12 | 179.3 |

4 | 178.4 | 1871 | 2146 | -275 | -0.08 | 179.3 |

5 | 177.9 | 1992 | 2092 | -100 | -0.03 | 179.2 |

6 | 178.8 | 2020 | 2346 | -326 | -0.09 | 179.1 |

7 | 178.8 | 1876 | 2567 | -691 | -0.20 | 178.9 |

So, on Day 2, I weighed 179.3 lbs and consumed 1556 calories. My Apple Watch estimates my TDEE to be 2407 calories, meaning I’m in an 851-calorie deficit (1556 – 2407 = -851).

To predict my weight loss for that day, I divide the deficit by 3500 (since 1 lb of fat equals roughly 3500 calories):

-851/3500 = -0.24 lbs

Now, I add this value (-0.24 lbs) to the previous day’s weight:

179.7 + (-0.24) = 179.5 lbs

I continue this process each day, updating the predicted weight and comparing it with my actual weight (which is averaged over 14 days to account for the daily fluctuations) to see how well the Apple Watch’s predictions align with my actual weight.

## How accurate was my Apple Watch?

Not very.

Take a look at the graph below, where I’ve plotted my scale weight, the 14-day average of my scale weight, and the predicted weight based on data from my Apple Watch.

At first, the Apple Watch’s predictions were fairly close to my 14-day average, with an average error of only 0.2 lbs during the first 20 days. But, like Marvel movies post-*Endgame*, things went downhill fast.

As my step count increased, the average error grew to about 1.6 lbs, with some individual days showing errors as high as 3 lbs. Essentially, the Apple Watch was *underestimating* how many calories I was actually burning, causing its weight predictions to fall behind my actual weight loss. Funnily enough, the most recent study testing the accuracy of smartwatches also found the Apple Watch 6 (the model I have) underestimated energy expenditure during walking. ^{3}

For comparison, I ran this same method using my calculated TDEE data and plotted it on the same graph (purple line):

As you can see, the predictions from my TDEE calculations track perfectly with my scale weight, with an average error of just -0.14 lbs––a stark contrast to the Apple Watch’s 1.1 lbs average error over the course of the experiment.

Also, yes, I’m aware my calculated TDEE data would track better with my scale weight since it’s based on *my* caloric intake and changes in *my* body weight. But this proves why you should use *your body weight* to inform calorie adjustments, not the number your smartwatch spits out. But I’m getting ahead of myself, so let’s get back to where we were.

Ok, so we’ve determined my Apple Watch underestimates the number of calories I burn during my walks. But just how bad was the discrepancy?

Let’s look at that next.

## How inaccurate was my Apple Watch?

In this analysis, I compared the active calories from my TDEE data against the active calories my Apple Watch reported using a Bland-Altman plot.

Don’t worry if you just read that and thought, “What the fu–?” I’ll explain what a Bland-Altman plot is in just a sec. But first, I want to explain the method to my madness.

I already know how many calories my Apple Watch estimates I burn since that data is readily available on my Watch and Health app. So, the next step was to figure out my active calories based on my TDEE. As a quick refresher, the number of calories you burn each day comprises three main factors:

**Resting Energy Expenditure (REE):**The calories your body burns at rest, simply keeping you alive—like if you were sitting around all day doing nothing.**Thermic Effect of Feeding (TEF):**The calories burned digesting and storing the food you eat, which accounts for roughly 10% of your total caloric intake.**Activity Energy Expenditure (AEE):**The calories burned through physical activity, including exercise and non-exercise activity such as cardio, lifting weights, fidgeting, maintaining posture, etc.

For our purposes here, we need to isolate the number of calories I was burning via Activity Energy Expenditure (AEE), which is relatively simple to figure out if we know the first two values.

I used the Mifflin-St Jeor formula to calculate the number of calories I burn at rest (REE). Next, I calculated 10% of my daily calorie intake to estimate how many calories were used up digesting and storing the food I ate (TEF). Finally, I subtracted the sum of these two values from my total daily energy expenditure (TDEE) to get the number of calories I was burning through activity (AEE).

Here’s an example:

- Let’s say I weighed 175 lbs yesterday and consumed 1800 calories.
- My estimated TDEE for the day is 2400 calories.
- Using the Mifflin-St Jeor formula, my REE comes in at 1740 calories.
- Since I ate 1800 calories, 180 calories were burned during digestion (TEF).

To calculate my active calories (AEE), I add my REE and TEF together and subtract that total from my TDEE:

This means I burned 480 calories via AEE.

Got it? Good, let’s keep it moving.

Next, we need to see how well the Apple Watch predicted my active calories using the aforementioned Bland-Altman plot, which shows how closely two measurement methods agree. In this case, we’re comparing my *manually calculated active calories* against the *Apple Watch’s Active Calories.*

Here’s the final plot:

The blue dots represent individual days. The y-axis shows the difference between the two estimates (Apple Watch vs. my manual calculations), and the x-axis is the average of the two estimates. The orange line in the middle is the **mean error**, which is **64 calories**. This means that, on average, the Apple Watch overestimated my calorie burn by about 64 calories.

Doesn’t sound too bad, right?

But here’s the thing: averages don’t tell the whole story. The spread of those blue dots (i.e., the difference between the two methods) tells us much more. The dashed lines at the top and bottom represent the limits of agreement (LOA), which show the range where most of the differences between the two methods fall. In my case, that range is -214 to +342 calories.

**In simple terms—my Apple Watch could underestimate the number of calories I burned by up to 214 calories or overestimate by up to 342. So, if the watch says I burned 500 calories, there’s a 95% chance my actual calorie burn is somewhere between 286 and 842 calories. **

But here’s where it gets spicy.

If you look at the plot again, you’ll see the dots spread out as you move to the right side of the plot (higher averages), indicating that as the total calories burned increase, the differences between the two methods also tend to get larger. So, while the Apple Watch might be somewhat accurate for smaller amounts of activity, it’s more likely to show larger discrepancies when I’m burning more calories.

**But wait! Because this shit is about to get even spicier.**

Remember that study I referenced earlier that looked at the accuracy of different smartwatches using indirect calorimetry? ^{3}

When they looked at the mean absolute percentage error for the Apple Watch during walking, they found an error rate of 24.1 ± 13.1 – i.e. the Apple Watch had an average error of 24%, with the error likely falling between 11% and 37.2% for most users.

Well, I decided to calculate the MAPE for my dataset and got 25.3 ± 17.3––pretty darn close to the study’s findings, which adds more credibility to my method and the results of this experiment. I guess I’m not as crazy as I thought.

In real-world terms, my Apple Watch might be off by 8% on some days, but it could also be off by as much ~43% on others. I’m sure we can agree that such a wide range is practically useless for informing calorie adjustments.

To put this into perspective – let’s say you start a diet and create a 500-calorie deficit, aiming to lose a pound per week, but you’re erroneously ‘eating back’ the calories you burned, according to your smartwatch.

If we use the percentage error from my dataset (25.3 ± 17.3) and assume your smartwatch reports you burned 500 calories, on a day when the error is only 8%, you might have burned closer to 540 calories—which isn’t a huge issue. I mean, 40 calories won’t materially impact your progress.

But what happens if the error is closer to 43%? You might have only burned 285 calories instead of 500. If you’re ‘eating back’ those 500 calories based on the inaccurate estimate, you’re consuming 215 more calories than you should. Now, instead of maintaining a 500-calorie deficit, you’re only in a ~280-calorie deficit, which could seriously slow down your progress.

In short, while the Apple Watch provides a reasonably accurate estimate of my active calorie burn on *average*, it doesn’t consistently capture day-to-day discrepancies. On any given day, I could burn significantly more or fewer calories than what my Apple Watch suggests.

## Why is the Apple Watch so bad at estimating energy expenditure?

The Apple Watch primarily uses heart rate to estimate caloric burn, which is a reasonable assumption since the higher your heart rate, the more likely it is you’re burning more calories.

Here’s the problem, though: Your (and my) heart rate can spike even if you’re not engaging in physical activity. Let’s say you decide to watch a horror movie, and your heart starts racing––your smartwatch can see your heart rate is elevated, but it can’t tell whether this uptick is due to you sitting on your butt watching a movie versus going out for a run, so it’ll just assume you’re doing some form of physical exertion, and track it as ‘calories burned’.

It also can’t account for all the small movements we do each day that don’t neatly fall into the ‘physical activity’ bucket. For example, I tend to bounce my legs whenever Kris and I watch Netflix. I don’t even realise I’m doing it until she points it out. But because I’m not moving off the couch, nor is the movement intense enough to increase my heart rate, the Apple Watch won’t register it as movement, even though *some* calories are being burned if I’m doing this for over an hour.

Additionally, your smartwatch can’t account for energy compensation. Just because you burned 500 calories during your morning run doesn’t mean your daily energy expenditure will increase by 500 calories for the day. This is because the body will compensate for some of the calories you burn.

The latest research suggests this compensation is around 30% on average, but it can be as high as 45%. So you might have burned 500 calories, but the actual effect on your energy expenditure might only be an additional 350 calories. However, this could be higher or lower depending on how physically active you are, whether you’re in a caloric deficit or surplus, or even your body fat levels. Unfortunately, your smartwatch only looks at the *gross* energy expenditure without considering energy compensation.

In fact, there are glimpses of this error in my data. One day, I walked about 16,000 steps, and my Apple Watch claimed I burned 840 ‘active’ calories. However, my calculation of AEE said I burned ~540 calories. To cross-check, I used the LCDA formula mentioned earlier, which estimated around 530 calories for walking 16,000 steps—once again, my data gave far more sane numbers than the Apple Watch.

Finally, and somewhat tied to energy compensation, is exercise efficiency. I haven’t run in a decade. If I lost my mind and decided to go for a run right now, I’d burn more calories covering the same distance than a seasoned runner. Someone who’s been running consistently for years will expend fewer calories thanks to their greater efficiency in running form, gliding along like an elegant gazelle while I’d be flapping around like a beached whale. As with energy compensation, your smartwatch can’t account for this in its calorie burn output.

## What you should do instead

Well, the first thing you *shouldn’t* do is use the calorie burn number your smartwatch is giving you to make caloric adjustments because, hopefully, you can see just how poor they are at estimating energy expenditure.

Now, you might be thinking, *Well, if I can’t use my smartwatch, how do I know when to increase or decrease my calories if my activity levels change?*

Simple: Your body weight. Or, more accurately, *changes* in your body weight *over time*. As I mentioned earlier, your TDEE doesn’t scale 1:1 with increases (or decreases) in activity. For instance, you might shift from being sedentary to moderately active, and let’s say that, on paper, you’re burning an extra 500 calories per day. In theory, that means you’d lose 1 lb per week (assuming your food intake hasn’t changed and you’re tracking accurately).

But in reality, your actual energy expenditure might only increase by 350 calories, meaning you’d lose closer to 0.7 lbs per week. However, since you’re using your body weight to inform any adjustments, they’ll be far more accurate than any questionable number your smartwatch throws at you.

Plus, tracking your body weight captures everything that could affect your daily calorie burn—energy compensation, metabolic adaptation, exercise efficiency, and yes, even calorie misreporting. Your smartwatch can’t account for these variables, but your body weight can.

## TL;DR

This article ended up being way longer than I’d envisioned, so here’s a quick recap of some of the key points.

- Increasing my step count from ~5,600 steps/day to ~11,100 steps/day seems to have increased my energy expenditure by about 200 calories.
- I tested the accuracy of my Apple Watch’s reported Active Calories against my calculated active calories (based on my TDEE), and the mean absolute percentage error was 25.3 ± 17.3. In real-world terms, my Apple Watch might be off by 8%, but it could also be off by as much ~43%. While the Apple Watch might be somewhat accurate for smaller amounts of activity, it’s more likely to show larger discrepancies when I’m burning more calories.
- Smartwatches only report your
*gross*energy expenditure but can’t account for all of the things that could reduce the number of calories you’re*actually*burning, like energy compensation and exercise efficiency. - Instead of using your smartwatch to inform calorie adjustments, you should use changes in your body weight
*over time*since your body weight data will account for all the factors that could impact how many calories you’re*truly*burning.

**Thanks for reading. If you enjoyed this, you’d love the Vitamin**

95% of my new content is only being sent to my email list. One email every Thursday, filled with actionable, evidence-based fitness advice to help you with your goals. If you enjoyed this, you’ll love my emails. You can learn more and subscribe for free here.