Evans, Dean (n.d.). Proofreading test: my wife vs. Grammarly vs. Ginger vs. After The Deadline vs. Microsoft Word 2010. GoodContentCo. Available 2-16 from: http://goodcontentcompany.com/proofreading-test-roundup.
There are things I like about this and things I don't. One is his simple approach: line them up, apply a certain block of text to each, and see what happens. This is what I do, knowing full well that ESL students' problems are to some degree different from the average native speaker writer, and that the free boxes that each offer as teasers often provide the best grammatical advice available on the market, only for 600-word texts, i.e. it's limited to a little bit at a time.
So he applies his text, with three spelling errors, three grammatical errors, and two factual errors. I don't see how you can expect a computer to pick up a factual error; for example, was it called Apple Co. or Apple Corps? Who would know or even look it up? But both Grammarly and Ginger failed to pick up an its/it's error, and I think this is crucial - now I know from the comments that this article was written before 2013, but I also know that even today the its/it's problem is considered one of the biggest out there. You have to get that one, and it's not that hard.
Well, it's hard enough that I myself couldn't program it into a computer, but I could program something in that told the writer: if you are using it's, are you sure you mean it is or it has?
One reason I mention this is that my wife is an academic, a sociology professor, with very high standards for her (fairly high-level, graduate, mostly native-speaker) writers. And she has, in a fit of frustration, just made an assignment that essentially told her writers, no more of this its/it's crap. No more cupertinos. This is formal academic English. Quit with the lousy writing.
Now these are students who are basically on the market for these programs. The internationals get a kind of break in this picture but, basically, they have some of the same problems, often a few more, but at the same time (compared to the native speakers) they have been learning for many years, and if you give them an opportunity to learn they will. The problem is that these programs don't teach. In some cases they un-teach. Usually they just imperiously pronounce. But if they can't even pick up its/it's, we're in big trouble.
The other common ones are there/their/they're and too/to/two which I figured native speakers never got wrong, boy was I wrong about that. The increased frequency of these kinds of errors is alarming, in my book. What, does Autocorrect just remove that extra "o" (making too into to) every time, just for fun?
Back to this guy's methodical comparison. I'm especially impressed by the comments, and may print them out just to see what people have said. They've had at least three years to pipe up; his article has finally reached the point of prominence where if you mention similar grammar programs in one breath, Google will at least pick up on it and put it up there. But three or four years is a long time; a lot has happened since then.