A huge mistake I see all the time is reviews that summarize the whole plot instead of giving an opinion. If I wanted a synopsis, I'd read the back cover. The whole point of a review is to tell me what you felt, not what happened. Did the ending feel rushed? Did you hate a character so much it ruined the book? That's the interesting part. Also, avoiding spoilers seems obvious, but some reviewers get so excited discussing a twist they just blurt it out, which is a total disservice to people browsing reviews to decide whether to read it.
Another thing is pretending to be objective. It's impossible. Every review is subjective, and claiming otherwise makes you sound like you're writing a school report. Just own your biases. Say 'I usually hate romances, but this one worked for me because...' That's way more helpful than a generic 'the romance was good.' It gives me a benchmark. If I know I love romances, and a romance-hater liked this one, that's a powerful recommendation. The reviews I trust most are the ones where the writer's personality and preferences shine through, not the ones trying to sound like a universal authority.
Finally, writing in a vacuum without any context. Saying a book is 'bad' because it's slow-paced doesn't help if you don't mention what you compare it to. A slow-burn literary novel is supposed to be slow; calling that a flaw misses the point. It's better to frame it as 'if you're looking for a fast-paced thriller, this isn't it,' which is actually useful information. I've picked up books based on negative reviews that clearly stated the book's style, because that style sounded perfect for my mood that week.