Levels Health Review: I Tested It Again Two Years Later

I first tried Levels Health about two and a half years ago. Spent two months wearing continuous glucose monitors, logging every meal, and learning which foods my body liked and which ones wrecked me. I got some useful data, wrote a review, and moved on.

I always planned to come back to it. And the timing finally made sense. I moved south to a new state, which means new grocery stores, new restaurants, and new brands. I also have two kids now, so my eating schedule and routines look nothing like they did during my first test. It felt like a perfect time to recalibrate.

After another two months of testing, my take on Levels hasn't changed much. It's a useful tool if you treat it like a periodic check-up rather than a lifestyle. The pricing takes some explaining, you now have two CGM options to choose from, and the app has gotten better in some areas while staying frustrating in others.

TL;DR — Is it worth it?

7/10

Levels Health

Try Levels For Yourself

Sign up if:

• You want to see how specific foods impact your body
• You want to optimize exercise and meal timing
• You like using data to improve your health

Skip it if:

• You're uncomfortable with a sensor under your skin
• You're not ready to change your current routine
• You don't want tech taking over your life
• You're an avid snacker who eats 4-5 times a day

What Levels Actually Is (and Isn't)

A lot of people think Levels makes the sensors. They don't. Levels doesn't make any hardware. The CGM sensors come from Dexcom. Levels is purely a software company. They take the raw glucose data from your sensor and run it through their own algorithm to give you meal scores, stability scores, trends, and actionable insights. That's what you're paying for with the $200 per year subscription. The CGM gives you the data, Levels makes the data useful.

This means you're running two apps. The Dexcom or Stelo app collects data from the sensor and exports it to Levels for analysis. It's clunky, and I'd love for Levels to bring sensor setup into their own app, but that would require a Dexcom partnership that doesn't exist yet. One bonus of the Stelo sensors: they don't have the emergency glucose alerts that the G7 has. Those are designed for diabetics and will blare at you in the middle of the night if you roll onto your sensor. Stelo is designed for general wellness, so you don't deal with that.

Pricing, Hardware, and App: What You Need to Know

Two CGM options. Levels now offers the Dexcom Stelo alongside the medical-grade G7 I used last time. The Stelo is over-the-counter (no prescription needed), lasts 15 days instead of 10, and costs $100 per month through Levels versus $200 for the G7. I used the Stelo this time and the data felt just as accurate for my purposes.

New pricing structure. Levels used to have a monthly subscription for around $40 plus the CGM cost. That's gone. Now it's $200 per year just for app access, plus $100 per month for the Stelo or $200 for the G7. A single month of testing runs about $300. Two months is about $400.

Better AI food logging. You can now take a photo of your entire plate and the AI identifies what you're eating. It's surprisingly accurate and makes meal logging way faster. The catch: it also tries to estimate calories and macros, and the portion sizes it guesses are usually about half of what's actually there. Trust the food identification, not the calorie counts.

Is the Levels App Worth $200 Over the Free Stelo App?

A Stelo CGM is $99 for a month supply and uses Dexcom's free Stelo app to track your glucose without paying Levels anything.

I spent some time in the Stelo app and it still feels built for diabetics. The glucose scale goes from 70 to 250 mg/dL, so if you're a healthy person hovering around 80-120, your readings are crammed into the bottom of the graph. You can log food, but there are no meal scores, no algorithm analyzing your response, and no way to compare how your body handled the same meal on different days.

Levels is where the data becomes interesting. The food scores, the trend analysis, the ability to search your history and see that pizza spiked you to 180 last Tuesday but only 140 the time before when you went for a walk after. If you just want to watch a line graph go up and down, the Stelo app works. If you want to actually learn something, Levels is worth it.

That said, Levels now requires an annual subscription, and the app doesn't do much during the months you're not wearing a sensor. Most people don't need a CGM year-round. If Levels could nail the AI portion sizes and calorie estimates in their food logging, it'd basically become a competitor to dedicated tracking apps like Lifesum and would justify the $200 even during off months. Right now, you're paying for 12 months to use it for one or two.

Wearing a CGM Still Sucks

I still hate wearing these things. The sensor sits under your skin on your upper arm, and even though it's not painful most of the time, you're always aware it's there. I lean on it, I bump it during workouts, and taking it off still feels like ripping off the world's stickiest bandaid. A little shaving and some Aquaphor help, but it's never pleasant. Don't be surprised if you see a little blood.

Pressure on the sensor doesn't just feel bad, it gives you bad data. I'm always thinking about my arm positioning on the couch or in bed. And I once ripped a sensor clean out of my arm taking my shirt off. That didn't feel great.

Data accuracy can be shaky for the first 24 hours with readings sometimes off by up to 20%. After a day or so, things stabilize. If the sensor ends up slightly toward the muscle rather than between your triceps and delts, you'll feel it more during weight training.

One unexpected positive: every time I wore it, people asked about it. Random people, friends, family, all curious about what was on the back of my arm. It's a cool conversation starter if you're into tech.

I really hope non-invasive glucose tracking comes to devices like the Apple Watch or Whoop at some point. That would make the Levels subscription a much better value because you could check in on your glucose whenever you wanted throughout the year without dealing with sensors. Right now, the hardware is what makes long-term use impractical.

How the Levels App Works

The home screen shows your average glucose, stability score, and spike duration at a glance. The goal is keeping your stability score above 85, which means avoiding spikes (any time your glucose rises at least 30 mg/dL and exceeds 110 mg/dL). Exercise-related spikes don't count against you if logged or imported from a device like an Apple Watch.

Levels sets three daily goals: 12 hours of glucose stability, one stable meal, and one healthy habit like exercise or good sleep. It's a clean system that keeps you focused without overwhelming you.

The meal scoring is still my favorite feature. Every meal gets a score from 1 to 10 based on how your glucose responds. You can search your history and compare how your body reacted to pizza last Tuesday versus pizza two weeks ago. It gamifies eating in a way that actually teaches you something.

The Food Scoring Window Needs Work

Levels analyzes your glucose response within a two-hour window after eating. For breakfast or lunch, that usually works fine. But for dinner, especially bigger meals, my glucose spike often didn't show up until 3 hours later.

I ate pizza for dinner and got a meal score of 6. Gentle rise. Looks totally fine. But three hours later, my glucose spiked 46 points from baseline. I also had a chicken and white rice meal that scored a 9. Amazing score. Except that one spiked 38 points three hours later too.

Levels does catch these late spikes. I'd get a notification saying "Glucose Spike Detected" with the timestamp. But that spike never connects back to the meal that caused it. The meal just sits there with its original score like nothing happened.
So I tried extending the scoring window to three hours, figuring the scores would drop. They didn't. After more testing, the meal scores never seem to change when you edit the window. The algorithm measures total glucose exposure over time, not just the peak. So two flat hours followed by one bad hour still averages out okay. The science might actually back that up.

But here's my problem. Meal scores are the main reason to pay for Levels over the free Stelo app. If a meal spikes me 46 points and still scores a 6, I'm going to keep eating it thinking it's fine. I don't know how common delayed spikes are for other people, but Levels clearly knows it happens because they built the window extension feature. I just wish the scores reflected it. When meal scores work, they're the best part of this app. They just don't always work for dinner.

Watch Out for Mega Zones

Every time you log an activity, whether it's a meal, a snack, or exercise, Levels creates a two-hour glucose response window. If those windows overlap, they merge into one giant scoring zone. So if you exercise, take a 30-minute break, eat lunch, rest for an hour, have a snack, then have another snack, you could end up with a single six or seven-hour mega zone. At that point, the score is basically useless because it's averaging your glucose response across your entire afternoon.

If you're a frequent snacker or someone who exercises multiple times a day, this is going to be a problem. The data works best when your meals are spaced out with clear gaps between them. I usually eat two bigger meals with maybe a snack in between, so my scores stayed clean. But if you're eating four or five times a day, I don't think the data would be all that useful.

What I Learned About My Body

Across both testing periods, here are the biggest takeaways.

Eating Habits That Made a Difference

Walking after meals is the easiest win. Even five minutes makes a real difference. A bad meal might go from a score of 1 to a 3, and an average meal can jump from a 5 to an 8 just by taking a short walk.

Eat protein and fat before carbs. If I had rice with a meal, eating the protein first noticeably reduced the spike compared to diving straight into the carbs.

Earlier meals score better. The same meal that scored well at lunch consistently scored worse at dinner. This pushed me to eat earlier and rethink my fasting routine. Turns out breakfast really might be the most important meal of the day.

Eating out is rough. Restaurant meals are carb-heavy, low on protein, and almost always cause a spike. Chinese food and pizza were the worst offenders. My favorite Chinese meal caused a 70-point spike that lasted more than two hours.

Specific Foods I Tested

Equip Prime Protein bars (new this round). Prime Bars have 15 grams of sugar from dates and honey, plus 20 grams of beef protein isolate. I consistently got meal scores around 7 as a pre-workout breakfast. Proof that ingredient quality matters. Not all sugar is created equal.

Sourdough bread (new this round). As long as I paired it with a protein shake, I could eat sourdough without tanking my score. Not all flour is bad, especially combined with protein.

Siete tortilla wraps. Made from cassava or almond flour. Four wraps stuffed with meat or eggs gave me an 8. That's 48 grams of carbs proving that carbs aren't the enemy. It's the refined flour doing the damage.

Chickpea pasta (Banza). 50% more protein and fewer carbs than regular pasta. Brought my scores closer to the 6-8 range. Traditional pasta spiked me no matter how much meat I loaded on top. (Both of my kids love Goodles Chickpea Mac & Cheese, by the way.)

Cauliflower rice. Great glucose substitute for white rice (5 carbs per cup versus 37), but tough if you're trying to bulk.

Cauliflower crust pizza. My body loved it. The taste? Terrible. No crispness, kind of soggy. There's no good substitute for real pizza.

Who is this for?

Levels is for people who want to understand how their body reacts to food and are willing to deal with the discomfort of wearing a CGM for a month or two. If you want clear, organized data to act on, Levels does that.

Here's my honest recommendation: subscribe to Levels for the year, then buy a month of sensors and test everything you normally eat. A few months later, buy another month of sensors and do a second round to see how any changes you've made are working. That gives you two solid data sets for $400 total.

A full year of continuous monitoring doesn't make sense. It costs over $1,400, it's mentally taxing, and the CGMs are physically uncomfortable. There's also a real risk of over-analysis. If you're not careful, you can get obsessed with optimizing every meal and it starts controlling how you think about food. Use it to learn, adjust what you eat, then step away.

Someday non-invasive glucose tracking will come to wearables and the Levels subscription will make a lot more sense. Until then, a couple month-long sessions per year is the sweet spot.

Affiliate Disclosure

I purchased my Levels subscription and CGMs with my own money. After my first review, I reached out about an affiliate link for my readers. If you use any links in this post to sign up, I'll receive a commission. My link isn't unique and can be found elsewhere, but using it supports my content and I appreciate it.