Beta reading is where a writer gives you a copy of their work, and asks for feedback. It's different from reviewing, because your comments are intended for the writer, rather than for potential readers.
And it's not always easy: I've both given and received such feedback, and on the one hand it's usually difficult to know what to write, while on the other, practically no-one tells me what I really wanted to know, and quite often I get feedback which is almost entirely unhelpful.
So here's a few tips:
Praise: Say something nice. Even if it's terrible. Don't lie, but there will almost certainly be some good point.
There's no point going on and on about the good points if you actually thought it wasn't very good, but to leave them out entirely is just mean. If you really do think that there's nothing at all good you can say, then probably the best thing to do is to gently break the news to the author that it might be best to start again.
Criticism: Likewise, try to mention something that could be improved. Even if it's brilliant, there will be something that could be even better. What an author wants from a beta reader is not just to know how good it is, but how it could be made better.
Try to balance praise and criticism according to how good it was. Don't write paragraphs on the good points and one line about the bad if you thought it was rubbish, or loads of criticism and a brief sentence of praise if you loved it.
Be Specific: 'I really liked it' is not helpful feedback. It's better than nothing, and a mention of how good it was overall is good to include, but it doesn't tell the writer very much at all. Try to mention what you liked about it, and what it was that could have been better. That's sometimes easier said than done; some points you may want to consider are:
Characters: are they varied, rounded, identifiable? Do you have clear pictures of them in your mind? How well do they fit in their roles? Does it seem like the author has projected their own values onto them? Is there any character in particular that you like/dislike? Is there a particular reason?
Plot: does it flow well, or does it feel forced? Are there any holes? Is there anything you thought was particularly clever, creative or effective? Anything you thought didn't work? Most importantly, is the story interesting?
Pacing: do you think it's moving too fast to build suspense? Too slowly to keep you interested? Or just right?
Writing: Does it read nicely or are there a lot of jarring notes? Any particular bits you want to call attention to as being especially good (or bad)? Do you think there's too much description, or could there be more? What about dialogue? Does the writer 'show not tell', or dump too much info? Do they do it effectively or is it hard to work out what's going on?
You don't necessarily have to mention all of these things, nor are they the only things, but there's an idea of the sort of things you could mention to be more useful than 'it was very good'.
Mistakes: Don't think that you're being obnoxious by pointing out typos, bad grammar or inconsistencies (I recall an instance in a published book where a character grew from five years old to seven over the course of a few months). You're not pointing out petty flaws in a finished work, you're spotting things the writer can fix. Don't suppose that your catches will be fixed anyway in proof-reading or editing: probably it will, but typos do slip through.
Don't be a grammar Nazi though: in fiction, it is perfectly reasonable to bend or flat out shatter the rules of grammar if it works better that way and is clear; don't bother pointing out sentence fragments, split infinitives, or the starting of a sentence with a connective.
If you think it's probably a mistake, point it out: I assure you the author will be very grateful.
Your Suspicions: One thing which a writer absolutely cannot do without your help is to know how much the reader will guess. You can take all the pains you like to throw out hints, throw out misleading hints, keep facts covered up, but it's not possible to be sure how much of it will be picked up on because as the writer you know what's going to happen.
A reader's guesses from one point in the story about the as yet unread portion are feedback gold. It's easiest to do this with a serial story, but if you're given the whole thing at once, and expect to be giving feedback, try to make a note of your ideas as you read it.