Wait a minute i have comcast 105 extreme service which can get speeds up to 105 and i get 50mbps download and 11 upload with this same test!! Your telling me a smartphone can double that? Is comcast this bad, does anyone else have these results i am getting or is it not typical?
TL;DR:
3) verify your local cable hardware is up to date and configured for 105 service
(DOCSIS 3.0/1 fully capable and aligned with your head end modem settings)
2) verify that old sub 105mpbs QoS settings from Comcast aren't in place.
1) verify the SpeedTest Server that is you are using isn't close geographically, but close net topology wise.
0) you realistically can't get more than 50mbps if your running wireless at 2.4ghz bgn mode (assuming you're not doing your comcast test hardwired with a PC), as one device that flips to b or g mode drives the speed down for all of them from 'n' to 'g' mode (54mbps). Run at 5Ghz 'a/n only' mode as your preferred wireless connection, if at all possible.
That last one is critical. 5ghz allows for less congestion/noise (sacrificing distance), and allows you to exploit n and ac bandwidth protocols more effectively. IF you are using a Apple wireless look at your connection quality in Airport Configuration and see what your getting for your devices. I set up a separate 5g (A/N only) network to isolate my remote apple TV, iPad, and MBA to get better speeds for a 35' radius of my AirportExpress/Extreme pair (hard cabled together so I'm not doing wireless extending). I've got a couple 'b/g' devices, so I keep the 2.4ghz running, and it also drives about 150' into the yard.
Then, it's all about where your server is, the saturation of the links between you and it and the quality of your cable modem/router and the head end.
Not exactly the same, but I bought 25/5 service for my local FIOS, and I was getting spotty performance. Found out they mis configured my route (I was a static IP and therefore had a different outbound router because that eliminated
double NAT), and when they fixed that, I got 25/5.
Then they automatically upgraded me to 75/25 (was 100/50 for a month, then they eliminated that price point... and 100/100 was double 75/25, which was now lower than my original 25/5). All that said, my FIOS POP was circa 2004, and while designed for 100mbps in 'burst' mode, could only really handle 50mpbs in steady state, and actually degraded the more load I put on it. In fact, the person told me that the device I had was designed to connect apartment buildings 802.11b wireless systems that were capped at 11mbps (no one person expected more than 10mbps). The upgrade to a 2012 built device sped it right up.
Finally, my fiber provider has a local speed test server, and I still was getting weird results. I determined that my route to that was still QoSed at 25 D (I could get faster uploads than downloads). It appears they don't back up their configs or saved to Nvram, and the head end device rebooted back to 2013 settings. Once they fixed that, I'm running 68+D/22U to the local server, and 65/18 to the 2nd 'nearest' (by traceroute hops) servers.
So... if you have trace route and use something like RobTex.com to determine your AU and peer networks, you may want to see how many hops it is to your 'local' speed test server, (done by geolocation), and instead, look for one on the comcast backbone, which may be fewer hops.