[majos]

Really? I find even Yelp reviews low quality in general, so I use older methods (food forums like Chowhound or eGullet, some bloggers I'm familiar with, sometimes even newspaper critics) to find restaurants.

Too many Yelp reviewers take off points for feeling disrespected in odd ways or don't know much about food to begin with. I don't really trust a completely random person's opinion on food, so I prefer going to the "hobbyists" (forums). These people have their own biases, but the base level of knowledge and care is much higher.

Plus Yelp has a bad history (edit: disputed below) of extorting money from businesses by controlling which of their reviews show up.

[CommieBobDole]

I've been able to get some value out of Yelp reviews by using the the following guidelines:

First and of course, look at the overall number of reviews and score as a rough guideline. You're probably not going to get burned going to a place with 700 reviews averaging 4.5 stars. Might not be as good as its reviews, but it's probably not crap.

Second, ignore any review with the word "groupon" or "scoutmob" in it. For whatever reason these always seem to be nitpicks about service specifically related to the offer. And if I had a dollar for every "Groupon was for item A, I wanted to apply it to item B, they wouldn't, one star" I've seen, I'd be a wealthy man.

Third, ignore most of the five star and one star reviews. Mostly undeserved gushing, and angry ranting, respectively.

Fourth, generally avoid anything from a user that has "Yelp Elite" next to their names. While some of these are good reviews from informed food enthusiasts, a lot of them are just long-winded twee crap from folks who like the idea of being "elite" at something.

Finally (and most importantly), pay special attention to the two and three star reviews. This is the meat of Yelp. They will say moderately bad things about the restaurant, usually with reasons to back it up. Often you'll start to see a theme. Whatever issues the restaurant has, this is where you'll find them. This is where you'll find out the popular place for some cuisine is actually kind of mediocre and everyone familiar with that cuisine goes somewhere else that's less popular but has better food, or whatever.

Then you ask yourself, "Is this something I care about?" If I want the best Ćevapi in town, do I really care about the dozen two and three star reviews that complain that the waiters only speak Croatian? Or will I show up with my handy Croatian phrasebook?

Yelp is a pit of crap, but there's gold in that crap. Usually.

[ghaff]

>pay special attention to the two and three star reviews.

I find that's often the case on Amazon as well. On your typical book review, say, there are a lot of 4s or 5s that are gushing praise because it's someone's favored author or genre or whatever. 1s tend to be "It sucked. Read 10 pages and threw it out."

2s and 3s as you say are more likely to be along the lines of "Really tried to like it because it had good ideas and I've liked author X in the past but this book was just too disjoined and confusing to recommend it."

[z0r]

I've found 1 star reviews to be hugely useful on Amazon. Often have seen detailed contrarian 1 star reviews for books that have nearly a 5 star average and found very interesting viewpoints. This works especially well when you're suspicious of the product being a scam - less the case for books, but very useful for other consumer products. The 1 star reviews are where people who received a cheap knock off or something that broke or otherwise failed quickly complain. I like to read the 2 and 3 star reviews as well but there's a similar amount of filtering necessary in my experience.

[L_Rahman]

'Sometimes even newspaper critics' - oh man this is harsh. I don't like the biases that newspaper critics tend to have but Pete Wells and Ruth Reichl at the Times, Sietsema (previously at the Village Voice, now at Eater) all do excellent reporting.

But I agree that it all pales in comparison to amateur enthusiast communities and specialist blogs.

[ghaff]

I put Yelp in the category of often (maybe even usually) better than sticking a finger on the map. Or just finding some minimally acceptable place. But, I agree that it's not great and, for me at least, a lot of reviewers seem to have very different priorities than I do and there often do seem to be a lot of weird complaints.

[arkades]

I find Yelp a very reliable source of tourist traps.

If you’re not in an area where that’s a risk, it’s not a good way to find -good- food, but it’s a good way to find the best local mainstream food. If the local mainstream best is mediocre... yeah, it’ll be 5 stars on yelp.

[ghaff]

I don't really disagree with that. It's also a reasonably good way (most of the time) to avoid really bad choices. I certainly don't love Yelp but it's often better than alternatives when traveling. It's not like I find asking for recommendations at a hotel necessarily yields great results either.

[drb91]

You can’t go by the calculated review number; you should read the reviews (which it seems like you already do on other sites.) I would focus on the individual reviews and ignore the aggregate entirely.

It would also be nice if you could rate for, like, food, service, and atmosphere separately. But you can’t, so it’s meaningless.

[ghaff]

>rate for, like, food, service, and atmosphere separately

Which is what Zagat was like initially.

[timr]

"Plus Yelp has a bad history of extorting money from businesses by controlling which of their reviews show up."

No, they don't. This is a rumor that is often repeated, but is never backed with evidence that goes beyond hearsay.

Also, I worked at Yelp, on the systems in question, and I can tell you that it's false.

[majos]

The evidence against Yelp seems to amount to a (from what I can tell sizeable) population of small business owners who claim that Yelp 1. is extremely aggressive about selling its advertising services (perhaps annoying, but not criminal) and 2. responds to failure to buy these services by filtering or deleting positive reviews.

Now, this evidence exists mostly in hearsay-type form: blog posts, forum posts, huge reddit threads. You can find similar stuff for BBB and Angie's List. There are also a couple thousand FTC complaints against Yelp, at least.

Yelp has turned these attacks away in court, but the (apparent) last ruling on this seemed to come down to an inability on the plaintiffs' side to prove Yelp actually changes review visibility based on advertising [1].

It seems silly to believe random people on the internet over tossed lawsuits and a dropped FTC investigation. But I see this as the likely result of Yelp being cagey about how reviews and advertising interact, and being able to hide behind a policy that promises a scrupulous approach and proprietary algorithms. It's not like I'm using Reddit posts to support anti-vax arguments, small business owners can definitely reason about how Yelp operates from their interactions with it.

[1] https://www.wired.com/2015/11/people-keep-suing-yelp-over-it...

[timr]

"Now, this evidence exists mostly in hearsay-type form: blog posts, forum posts, huge reddit threads. You can find similar stuff for BBB and Angie's List. There are also a couple thousand FTC complaints against Yelp, at least."

No, the "evidence" exists entirely in hearsay form. All you ever see are stories about parents, friends of friends, distant cousins, and so on. There's almost never so much as a link to the business page. Why? Because it's trivial to debunk these claims when you can see the reviews.

"It seems silly to believe random people on the internet over tossed lawsuits and a dropped FTC investigation."

Not just silly, but absurd. It's internet conspiracy theory, and people are just thoughtlessly repeating it as fact.

"But I see this as the likely result of Yelp being cagey about how reviews and advertising interact, and being able to hide behind a policy that promises a scrupulous approach and proprietary algorithms."

If Yelp revealed how reviews are filtered, the filters would be rendered useless overnight. It would be the equivalent of Google publicly documenting their search algorithm. And while I think Yelp does many things badly when it comes to this stuff, they're 100% clear about how reviews and ads interact: they don't.

[paulgb]

> I worked at Yelp, on the systems in question, and I can tell you that it's false.

I wouldn't expect there to be a literal code switch that deletes bad reviews if someone pays for ads. But I can imagine a system developing naturally (even by accident!) where it's easier to get a human at Yelp to intervene when non-customers or bots left bad reviews.

Did Yelp have systems in place (e.g. a "Chinese wall") to prevent someone's status as a customer from impacting their ability to resolve a complaint about abusive negative reviews?

[timr]

Yes.

[aphextron]

>Too many Yelp reviewers take off points for feeling disrespected in odd ways or don't know much about food to begin with. Plus Yelp has a bad history of extorting money from businesses by controlling which of their reviews show up.

Agreed, the individual reviews are usually useless. But their aggregate star rating is great. A 4.5 star place with hundreds of reviews is guaranteed to be amazing without fail. 4 star is always decent. But anything 3.5 and less is guaranteed to be mediocre or bad. I eat out almost every day in SF/east bay and this has yet to fail me.

[azinman2]

Disagree. Take El Faro on 1st [1], it’s actually pretty good (their taco dorado al pastor is really good), yet it has a 2.5 rating. It’s also been in business for years, so clearly something else is keeping them alive than their yelp rating. And if you look at individual reviews, there are a bunch of 4 and 5 stars, yet they disappear when scrolling a search list.

I’ve also found many 4+ star restaurants entirely overrated and quite mediocre.

It’s all about the category and how it fits into the local milieux, which might place emphasis on things you don’t care about or expect (the ambiance sucks at el faro, for example, but it’s a lunchtime tacqueria I could care less). People take away stars from it because it’s cash only. Sorry, but what exactly is it that we’re rating here?

That lack of common consensus and the wide disparity between expectations means yelp aggregates aren’t always so useful. I wish they did a better job of showing me the reviews and stars of the things that matter to me, not to the shapeless aggregate. Or at least give me a sense of the distribution to know if there’s more/less contention than normal.

[1] https://m.yelp.com/biz/el-faro-san-francisco-3