What I Learned After 2 Months With a Levels CGM

Coming out of high school, I had some *extremely* poor eating habits which involved lots of Hot Pockets, Monster energy drinks, and Welch's fruit snacks. My metabolism was through the roof and I could eat whatever I wanted, as much as I wanted and still be relatively skinny.

But that metabolic rate doesn't last forever, and combined with the (definitely excessive) binge-drinking my freshman year (and all the subsequent late-night pizza), I found myself quickly adding on some extra width.

That was when I started to learn about diets, nutrition, and what it means to eat healthy.

I've always had an obsessive and contrarian personality, so the first diet regimen that I tried was full ketosis. I relished in how extreme it was and told everyone I could about it (much to their annoyance I now presume).

The result? I lost 15 pounds in less than a month and felt better than I ever had before. I previously thought that brain fog after meals was the norm and that afternoon naps were something that couldn't be avoided. Both were a thing of the past after I cut out carbs completely.

Eventually I had to discontinue keto since I joined the cycling team and had started to lose weight far too aggressively when it was combined with 100+ miles on the bike every week. I simply couldn't consume enough calories without eating carbs. This also happened to be a boon for my social life as well--turns out people are more likely to invite you out to eat when they don't have to accommodate extreme diets.

But that experiment showed me just how drastic the changes could be when the inputs to one's body are tweaked. Ever since then, I've slowly refined my habits around what I eat, exercise, and recover.

A few months ago, I was introduced to Levels Health when its founder, Josh Clemente, appeared on one of my favorite podcasts, Creator Lab. I'd heard of continuous glucose monitors before, but their use was always caveated with their inconvenience and high cost. So when Josh spoke of how easy it was to get one and use, I signed up immediately.

You may be wondering, "Why do you want to have a needle in your arm at all times measuring your glucose?"

Levels has a far more detailed answer to that in their blog posts, but in short, there is no optimal diet for everyone. There was a seminal study in the journal Cell which found significant inter-personal variation in glucose response to different types of foods both at a macro and micro level. Some people can tolerate high-carb diets, and some do better on low-carb, high fat ones. I was pretty sure I was in the latter category given my experience above. And it is with data from a CGM that this can be deduced definitively.

Differing post-meal glucose responses between people in foods of the same macro group

The idea is that you eat all sorts of different foods while recording your glucose levels with a CGM and then replace the foods that cause excess glucose spikes with those that don't.

Here's one example from my own recent experimentation: on separate days I consumed equal carb portions (50g) of white rice and oatmeal. Here was the result:

Rice vs Oatmeal

Despite both being white carbs, the glucose spike with the rice was 5x worse than with the oatmeal. I also repeated the experiment with various types of bread, potatoes, and fruits. In ascending order of glucose spike level (descending of tolerance): oatmeal, white potato, berries, oranges, apples, bananas, sweet potato, brown rice, white rice.

But the exact food tolerances were just the start of what could be experimented with!

I already knew that intense exercise would have some effect on the glucose response, but I never would have expected just how big that effect would be. After a workout, the score of my meal was never below an 8, even when I consumed foods that normally would make my glucose spike significantly.

Here's one example where I got Indian takeout (which included lots of rice), ate it for dinner and then also for lunch after my workout the next day:

Chicken Makhani after a workout vs not

These were approximately equivalent portions, the post-workout meal might have even been a bit larger. And yet, my glucose response is pretty much as flat as it gets.

But one doesn't need to be doing a super intense workout to see significant effects. Even just a 10 minute walk before and after a meal reduced the glucose spike significantly. Here are two consecutive taco nights, one with walks and one without:

Tacos with walks and without

The start times are a bit off, but you get the idea. I'll caveat this by saying that for the foods that I really don't tolerate well (like rice), a walk alone is not sufficient to keep my glucose down to an acceptable level. However for the vast majority of foods it is and I have now added post-meal walks to my daily routine, to great effect.

Another interesting observation was the magnitude of the effect of poor sleep on glucose response. I use an Oura ring to track my sleep and when the score for a night would drop below around 60, even normally well-tolerated foods started having significant glucose responses.

Here's an example where my sleep score was only 48:

Poor sleep results in horrible glucose responses

The first meal was a carb-only meal (brown rice) that is known to have a poor response, but the second meal was salmon and roast mixed veggies, which is my standard dinner and normally will not exceed 120 mg/dl.

This was another case where I had previously known that this effect existed, but had no idea what the magnitude was.

And not only was sleep a cause for poor glucose responses, but also an effect! One of the things I noticed with my Oura ring was that a lower heart rate throughout the night led to more deep sleep and an overall better sleep score. A cool feature of the Levels app is that it takes this heart rate and displays it under the glucose reading in the graph:

High heart rate at night correlates with high glucose level

This let me notice that there was a direct correlation between when my heart rate was high and when glucose was high. This frequently occured if I ate too close to bedtime or if dinner had a particularly bad response. Just one more reason to seek to keep those responses under control!

But perhaps the biggest lesson that I've gotten from this so far is that nutrition is really, really complicated. Just when you think you'll be able to predict what kind of spike your body will respond with, it throws you a curveball.

As an example, I made protein pancakes two Sundays in a row for brunch, nearly identical meals, no exercise before or after, similar amounts of sleep. But yet this was result:

Why it spiked high one week and stayed completely flat the next, I have no idea. Obviously this means I need to repeat for a third week (for science!)


It's not a common occurrence, which makes the glucose data an effective enough feedback loop for accelerating behavior change. Simply by being able to see the response and get a score, one finds oneself naturally making choices that lead to healthier outcomes.

The CGM's aren't perfect however. There was one that I received which led to consistently elevated readings and greatly exaggerated spikes:

Thankfully, the customer support at Levels is amazing and within a few days I got a new one in the mail free of charge.

In conclusion, I've learned a lot from using a CGM and am very excited about the future of Levels. And it appears that I'm not the only one. If you're able to afford it, I highly recommend trying it out.

You can skip the wait-list by using this referral link: https://levels.link/r/ANAMYVUW