When I first started traveling for work in 2009, hotel Wi-Fi was generally pretty slow. From 2010 through 2013, I started running speedtests and recording the results in a spreadsheet. Though it’s still not perfect, internet quality in hotels has vastly improved nowadays. For the twelve months from July 2022 through June 2023, I once again collected hotel internet speeds. Unlike last time’s manual entry into a spreadsheet, this time I wrote some Python scripts to automatically run a speedtest once per hour, and save the results into a CSV file.
The bars on this chart show a histogram of internet speeds—how many results I observed at each particular 1 Mbps window. Brown bars are download speed, and blue bars are upload speed. I also added a probability density plot for this data–approximately speaking, this smooths the histogram into a curve that shows how likely it is that any given speed will be measured (the higher the line, the more likely a given speed is).
Last decade, hotel bandwidth was generally low enough that internet speeds varied a lot depending on how many people were using it. Nowadays, it’s often simply capped at a particular speed. Many of my hotel stays are at Hilton brands, and they generally seem to have a cap around 30 Mbps, leading to that large spike on the chart; similarly, there are spikes around the round numbers of 10 Mbps and 20 Mbps.
However, there were a few hotels that didn’t cap and just allowed the speed to be whatever was available. These speeds generally seemed to be in the 60–115 Mbps range, with the most likely uncapped speed being around 85–90 Mbps.
Last Decade vs. This Decade
Since I still had the spreadsheet of data from 2010–2013, I was able to compare how much hotel internet speeds had improved over the last decade.
For download speeds, I never saw a speed above about 35 Mbps last decade, and the most likely download speed was about 20 Mbps. This past year, my most likely download speed was 30 Mbps (due to speed caps at many hotels), but I had a good chance of getting much higher download speeds as well.
Upload speeds are a much more dramatic difference. Ten years ago, every single one of my speedtests had an upload speed of less than 1 Mbps; these speeds are so likely that they tremendously stretch the y-axis of the chart. However, looking closely, it is possible to see that upload speeds are now generally pretty close to download speeds; last year’s most likely upload speed was about 20 Mbps, with a long tail of higher possible speeds.
Speed by Time of Day
In theory, hotel bandwidth available should be dependent on how many people are at the hotel, meaning I would expect lower internet speeds in the evening when many people are in their hotel rooms. Ten years ago that was relatively true, but since the limiting factor is the per-device speed cap at most hotels, that’s less true today:
Each hour bar shows every observation with that hour in its timestamp. For example, the bar at 11 represents every time from 11:00:00 to 11:59:59. Timestamps are in 24-hour time, so hours less than 12 are a.m., and hours 12 or greater are p.m.
The median speed does seem a little higher around 18:00 and 19:00 hours (6 p.m. and 7 p.m.)—presumably, when many guests are out at dinner. However, I’m surprised that it’s higher than some of the early hours of the morning, when presumably most guests are asleep.
Overall, hotel internet speeds are generally good enough for most routine tasks that they are rarely worth worrying about anymore. That was certainly not true ten years ago, so this is a big improvement. Of course, cellular speeds have also improved enough that hotel internet speeds matter less; if I do end up in a hotel with bad internet, at least I now have the option to get decent speeds via my phone’s data connection.