Everyone needs to understand that the "optimizer" isn't a piece of software that runs on your machine. It's just a testing tool that sense the speeds and latency on your network connection and then tweaks various parameters inside your machine to make it work "better" in the given network conditions at that time between you and that test server.
TCP and protocols over that run over the network work best in high bandwidth/low latency scenarios (like a local LAN) and adapt pretty well to low bandwidth/high latency scenarios (like a dialup). The advent of broadband and the internet brought about high bandwidth / high latency kinds of connections -- a set of circumstances which are notoriously poor scenarios for the TCP/IP protocol. There are parameters (such as window size -- number of packets which can go from server to client before an acknowledge packet must be seen, etc.) which can be tweak to help balance this so that you get a nice flow of data, but not such a long period that when an error or dropped packet occurs that you end up having to backup substantially to get sync'd back up again. It's these parameters that the optimizer "tweaks".
There are tools from various sources which you can use that all do that same thing ... for example:
In addition to the tweak test there, you'll also be point at tools that let you adjust those parameters should it be necessary. Remember -- more is not always better here -- it's really matter of finding the right combination of parameter values that are best for your unique situation.
Also with that said -- wireless is different the wired -- moving between the two or even from machine to machine will result in different settings on a machine and each machine in your house needs to be "tweaked" individually -- this is not a single setting for a FiOS installation.
If you want to accurately test your bandwidth, you need to use a machine (with A/V and firewall temporarily turned off) connected directly on a wired connection to the router which has been optimized with the test run against several different servers (Verizon, Speedtest.net, and DSLreports are the ones I frequently use). Remember also that some test locations actually cap their speeds -- so they may report a lower speed than is actually possible because of limitations at their end. If you are receiving artificially low numbers against a specific test site, try a different on at a location near you.
Also, one of the reasons I mentioned A/V in my previous post is that many products are starting to come with a scanning "proxy" built into them. These scanners hijack your browser traffic come to/from your machine and passes under the A/V microscope between it's delivered. The unfortunate side-effect of these is that they often hold onto large amounts of data (letting a little bit trickle out at a time while it scans the entire data set) before releasing it all at once.
While this is not always perseptable in a general "browsing" environment, it completely invalidates results from testing tools by artificially inflating or deflating the numbers depending on how the tool calculates it's results (for example, if it takes the average rate over the entire transmission of a data set, it probably doesn't affect that number much -- but you get to see your peak rate as a result -- if it takes small samples to get an idea of "peak" rate, it might articifically see that as very slow and then suddenly overly fast as the A/V product releases all the data at once).