I randomly got an ad for this TeamViewer press release: https://www.teamviewer.com/en/co2-study/

It basically says that TeamViewer products avoid around 37 megatons of CO2 per year, or 4 tons per subscriber. For reference, this is equivalent to more than nine coal-fired power plants running for an entire year, more than three billion trees binding CO2, or 7000 NY -> Singapore flights of a full A380. Here are the core claims from their study landing page:

A study conducted by TeamViewer together with the DFGE – Institute for Energy, Ecology and Economy showed that using TeamViewer solutions has a significant positive effect on global CO2e emissions. The amount avoided is impressive: 37 megatons of CO2e.

The use of TeamViewer’s digital solutions in the working and private environment – ranging from remote support in the office environment to steering and controlling machines as well as remotely supporting friends and family with IT issues helps, scientifically proven, to avoid CO2e emissions. An average TeamViewer connection can on average avoid 13kg CO2e. An average TeamViewer subscriber avoids up to 4t CO2e per year.

The page also contains the words "Scientifically proven" in big shiny letters. Is it though? Let's look at the study in more detail.

Well, uh, this is where the trouble begins. The study is not linked in the press release. It's not in the landing page. They do mention that the study was realized by an institute called DFGE, which is "ecovadis certified" (read: bullshit), and that institute has a blog post where they mention the study. However, when you click on "see more details on the study", it brings you back to the TeamViewer landing page. By using my mad Google skills ("dfge teamviewer filetype:pdf"), I found the thing that appears to be the Scientific™ study: https://www.handelsblatt.com/downloads/27033886/1/2021_dfge_teamviewer_carbon-emission-avoidance-study.pdf

So, how do I even summarize the problems with this study?

1. Their claim that an average TeamViewer subscriber avoids up to 4t CO2e per year is misleading, because it reverses the causality of subscriptions. Getting a TeamViewer subscription is more likely for people who are power-users, because their high usage makes it worth the cost. Reading their claim, you could think that they causally identified how much CO2 was avoided at the margin once someone got a TeamViewer subscription, but they do no such thing. Instead, they just estimate the average of avoided emissions across all their subscriber base thanks to their products, but it says nothing about how useful it is to get a subscription.

2. They ignore that video call services are relatively fungible. If people weren't on TeamViewer, they wouldn't just take a car to talk to someone IRL if that was impractical, they would find another video call service. This claim is also misleading because it implies that TeamViewer products are the cause of carbon abatement, whereas in reality what they try to identify is how video call technology in general abates emissions.

3. Their entire section about how they perform quantitative analysis shows the full extent of how much of a joke this entire paper is. They take every single TeamViewer call longer than 30 seconds, then compute how much carbon would have been emitted by taking a plane to have this meeting instead. This is obviously ridiculous: video calls massively bring down the cost of having remote meetings. If the technology was not available, people would either a) not have these meetings in the first place b) not build up companies or teams that require as much frequent communications c) have less people in each meeting by sending representatives instead d) group up meetings all at once to make flights less frequent. The effect identification is completely bunk and nonsense, which explains the high estimates.

4. To address this concern, they "qualitatively confirm" the results with a survey. There are multiple problems with this. a) The study is entirely self reported and thus does not necessarily reflect actual choices under budgetary constraints. b) the questions are not detailed, so it could very well be that a bad poll design have led to uninterpretable results. For instance, if a question was phrased "if you hadn't had access to this product, how much would you have to fly for this meeting", it would be too vague for people to enter "0" if they wouldn't have done the meeting in the first place, and they would be lead to enter a value that does not reflect their actual counterfactual behavior. c) Nowhere do they mention how the "qualitative results" were used to adjust the quantitative analysis. They just say that they "test and verify" the assumptions of their quantitative studies, they don't talk about any adjustment factor they would have applied to their initial results to correct for the problems mentioned in 3). We can only assume that they did not make that correction, and thus did not take these problems into account.

This could be funny if they weren't publishing ads about this bullshit, and congratulating themselves in their earnings calls:

Now before we come to the 2021 guidance, I’d like to take the chance to summarize the results of a new carbon emission avoidance study, which were conducted by a leading sustainability institute, which is called DFGE. They cover significant amount of DAX companies as well. And we have released that just last week highlighting the importance of our ESG initiatives. It’s clearly part of our vision that our remote connectivity solutions not only help to save time and money, but also they have a significant positive impact on the environment and, in fact, combating climate change. Therefore, we have commissioned DFGE to quantify the impact that the use of TeamViewer solutions have on the environment in a scientific study. And not surprisingly, we have found out that TeamViewer solutions help to significantly avoid carbon emissions. And in fact, it helps towards 37 megatons of carbon emissions per year based on the data collected by DFGE. This is equivalent to a fully booked A380. Obviously, nowadays they are not flying so much anymore. But this is equivalent to such a plane flying 7,000 times non-stop from Singapore to New York or equivalent to the emissions of 11 million average cars in one single year. So very substantive carbon emission reduction as part – as a result of the usage of our product. So a single TeamViewer subscriber actually can avoid on average around four tons of carbon footprint or carbon emissions per year. I think this proves clearly that our solutions are playing a critical role in helping organizations globally to avoid their carbon emissions

https://ir.teamviewer.com/download/companies/teamviewer/Transcript/20210209_TeamViewer_Q4-FY_2020_Transcript.pdf

This shit is the worst, corporate greenwashing is a plague and TeamViewer and the DFGE should be ashamed of themselves for this.