Air quality matters but don’t trust Foobot on it

I never paid much attention to my mom telling me to open the windows and get some “fresh air” when I lived at home. Sounded like advice from the same pot as “don’t swim after you eat”. But I was wrong.

Just how wrong has taken a few decades to realize. But as our family moved into a new house, I got forcefully sent down the rabbit hole of quantifying air quality when my wife got symptoms of formaldehyde poisoning after just a few weeks.

That’s when I got forced to learn everything about ventilation design, air exchanges, CFM rates, heat-recovery units, and baselines. Through that education, I read numerous scientific papers and industry standards on how to quantify exactly what “the air smells stale” and “I have a headache after staying in here for an hour” really meant.

All that research led to focusing on two big indicators of air quality: volatile organic compounds (or VOCs) and CO2 concentration. VOCs come in many forms, but the one that we had particular problems with was the formaldehyde (HCHO) off gassing from a closet system. So we chased HCHO with a battery of lab tests and professional sensor systems.

Collecting samples and sending them to labs for testing is not just expensive, but it’s also slow. That’s where the IoT Foobot for air quality measurements came in. Slick marketing, check. HCHO measured as part of a total VOC count, check. App and push notifications, check. 4/5 star rating on Amazon, check!

So I bought three units and placed them throughout the house. The results confirmed what we had found with the lab tests. VOCs and CO2 concentrations were rising without open doors/windows, so we were on the right path chasing mechanical ventilation solutions. But that’s when the good parts stop.

Because to call the Foobots imprecise would be an understatement. After seeing some oddly high values, I gathered all three devices in a single room to gauge the calibration. It was terrible! The VOC measures would swing by 50%, the fine particle counts by 400%, and the derived CO2 values were off the chart. Here’s how it looked:

Thankfully I had also ordered three CO2 measuring units from In that same room, here’s the readings:

So all within a 1% of variance on the main CO2 measurement. And, more importantly, a reading (440ppm) that was just one fourth of what the worst Foobot read (1814ppm). That’s the difference between “~perfect air quality re: CO2” and “holy shit, you’re going to have symptoms, DO SOMETHING!”.

To be fair, the Foobots do seem to some times agree more than what was showed here. But there’s no rhythm or rhyme to it. I’ve gotten “pollution event” push notifications from one Foobot while another in the same room was saying everything was “Good”.

Foobot do say in their fine print that the CO2 levels are “calculated according to VOC levels thanks to complex algorithms”, but there’s nothing in the app to project that lack of confidence. All the values are very precisely presented. 1814ppm. 1115ppm. 1250ppm. That’s just shady.

So is the fact that between three units, I was getting such widely different results at times. Yet when presented with this data, the response from Foobot’s support department was this:

As Zach already mentioned, your devices, all three of them, operate within regular operational parameters. The discrepancies you mention with regards to the readings are indeed normal and are consequence of both manufacturing tolerances and variance as well as environmental factors present in your home.

Further, Foobot justifies this wild variance with the following:

It is not a measuring instrument and in fact could not be given the price point of the device itself. Your may appreciate the fact that measuring instruments are indeed very expensive and indeed would be outside the reach of most families.

So in summary: There’s nothing wrong, and even if there was, it would be too expensive to do it right!

First of all, that’s just crazy. You can’t sell a product that purports to measure air quality to the level of precision presented in Foobot’s app, and then say, oh well, it might be off by 50%. Or 400%. That’s just how it is!

Second of all, it’s false. The variance on that set of CO2 sensors was below 1% on the CO2 reading! And those units are $129, while the Foobot’s are $199. Now, to be fair, the co2 sensors do not measure VOC or fine particles. But come on.

What irks me the most about this is how similar the playbook of Foobot comes to that of the archetypical Silicon Valley story. Like Theranos proclaiming they have a revolutionary process for doing blood tests, with lots of venture capital and fanfare, and then fail miserably when actually measured in the wild.

You can’t escape terms like medtech and fintech at the moment, but finance and medicine is not social media. Running fast and breaking things can actually bankrupt people or make them sick in these domains. Surely both industries need an influx of fresh ideas, but having well-funded startups swoop down with the same fake-it-till-you-make-it attitude is reckless.

Uber taught a generation of Silicon Valley aspirants that ethics can take the backseat as long as the growth is good. When you apply those lessons to social media, you get inflated user counts and fraudulent ad-play rates, but when you apply to medtech, you get dangerously false data that can make people sick or paranoid. This is not right.

Foobot should go back to the drawing board. Recall their uncalibrated, inaccurate, and misleading devices. Stop pretending to have precision that they have no backing to project. Then come back when their issues are fixed and help us all understand air quality better without the huckster spin.

But I wouldn’t hold my breath (ha). Foobot isn’t likely to do a damn thing. In the mean time, I recommend getting your CO2 readings from, opening your windows, and read these eye-opening papers about the effects of CO2 contraction on high-order cognitive functions.