10: How to Read Reviews

In my review bribe post, I mentioned that it's probably best to ignore five-star and one-star reviews on products. I thought it might be good to expand on that because it's actually an interesting question about how to not get misled by stuff on the internet.

So in brief, here's the problem: if you want a new product, you should read the reviews. It's a good way to make sure you aren't getting something that's total crap. The trouble is that if you look at nearly any product, it will give you a roughly 3.5-4 star rating. Sometimes a bit higher, and sometimes a bit lower, but nearly always in that range.

That means it's hard to make decisions based on the overall rating, so it's a good idea to read the reviews themselves. There's another problem in here too. If you look at most products, reviews are going to be overwhelmingly positive. That's the reason for things like the 3.5-4 star rating. Most of these are not going to be people that have had the product for very long.

So if you glance through five star reviews on almost anything, you'll see a lot of comments on how things look and the overall design. You'll also get my personal favorite: the "I bought this for my son, and he loves it!" review that, frankly, tells you absolutely nothing. I wish Amazon would take those out entirely, but hey, nothing's perfect.

All of that is to say that the vast majority of the five-star reviews won't tell you if a product is actually good (and that's even without the review buying I mentioned in my last post). It will tell you if something looks nice, but that's about it really.

Amazon itself contributes to this problem. It doesn't take long after you buy something that you'll get an email asking you for a review. I get those emails in a matter of days. Most of the time, that just isn't enough time to actually get to know if a product is good or not. I get that they want product information from verified purchases, but most of the time, the first impressions of a new product will be great. It skews the overall reviews upward.

Then there are the times where the first impressions are not great, and these lead to one-star reviews. These almost always have the same complaint: quality control. These aren't actually product reviews a lot of the time. Instead, they're reflective of QC mistake from the manufacturer. Now if there's a whole lot of those, it might be best to avoid the product, but usually that's not the case.

More often than not, one-star reviews are flukes. The other problem is that issues with shipping often end up in here as well. This is especially true with more fragile items that can easily be broken in transit. Overall, the problem with one-star reviews and five star reviews is pretty much the same: neither is reflective of the actual product.

The real place to look for information is in the 3-4 star review range. There are significantly fewer of those most of the time, but that is usually where you should go for real information. The other good thing is you'll occasionally see updated reviews in these areas. Some people will post that initial five star review, and then after a little more time, they will come back and edit it once they are more familiar with the product.

In terms of good information, those might be the best ones to look for because they will show the product without the rose-colored, first-impression glasses.

But I led with this being a good way to not be misled on the internet. I think the idea of ignoring overly positive or overly critical information is a great way to make sure your thought process is just a little more skeptical. It works for product reviews. It works just as well for political candidates, op-ed pieces, or anything where some critical decision making is required. If you ignore the extreme ends, you're more likely to come across better information.

Amazon reviews are a bit of a silly example, but it's a good lesson in critical thought, and I am always on the lookout for those.


You'll only receive email when Onset to Coda publishes a new post

More from Onset to Coda