A Comprehensive Food Rating System

Alan Liang
7 min readAug 26, 2018


A Little Rant on Current Food Rating Apps

One problem I encounter frequently — more often than I’d like to admit — is on deciding where to eat. Although food review apps like Yelp store vast amounts of information about each restaurant, they ultimately reduce a restaurant into a single number out of 5, as if it’s the sole indicator of how good a restaurant is.

In fact, scientifically speaking, this “number” isn’t even quantitative data as so much it is categorical: the rating values only go up in half increments out of 5, so that there are at most 9 categories from 1 star to 5 stars. What’s worse is that most (say, 95%?) restaurants are huddled around the 3 to 4.5 star categories, such that there are practically only 4 bins to compare.

Yelp’s Display Star Bins

I’ve found these bins to be vast over-generalizations. Yelp, for example, simply rounds the true average rating to the nearest half star, so that a restaurant with a 3.74 average rating would have a 3.5 star display rating, while a restaurant with a 3.75 rating would have a 4 star display rating. Theoretically, these restaurants shouldn’t be too different, but they become victims of this sweeping classification.

In fact, a restaurant’s Yelp display rating has a considerable impact on its popularity. In their paper Learning from the Cloud: Regression Discontinuity Estimates of the Effects of an Online Review Database, Berkeley economists (go bears) Michael Andersen and Jeremy Magruder found that controlling for all else, restaurants sell out up to 34% more frequently when they see a half star increase on Yelp.

Each Point Here Represents a Restaurant; Cutoff at 3.75 Represented by Dotted Line

Google Maps does this slightly better, going up in 0.1 increments for a score out of 5. This, along with its natural location based search, is why I’ve switched mainly to Google Maps for finding restaurants.

But still, this doesn’t solve the problem at heart. The bigger issue here is that ultimately, reviewers are only able to rate their dining experience with one of five categories: 1 to 5 stars. This is far too coarse and almost laughably subjective; for example, a user can rate a restaurant 5 stars because they had the most amazing meal, or perhaps they had a truly memorable experience there nothing to do with the food. On the other end, too many times I’ve seen 1 star reviews just because one small thing (like a mixup in orders) didn’t go well with the easily-upset reviewer.

Rant over.

The question arises: how can we make our reviews more scientific?

I know, food rating isn’t a science; whether something is good is completely subjective. But can we at least try to quantify this subjectiveness?

That’s what I’ve tried to do with my rating system. I acknowledge and completely respect that food is a personal experience, so what I’ve tried to do is enumerate and quantify the aspects that go into a holistic dining experience. Your rating in every category could very well be completely different from mine for the same restaurant, but the bottom line is that now at least we have the same scales to judge.

The Rating System

The 80–20 Rule

Ok, this is a pun, sort of. Essentially, the idea is that the overall rating score is divided into 2 sections: the basics section (out of 80), and the extras section (out of 20). Together, they sum to 100, which is the theoretical maximum score a restaurant can attain.

The idea is that even if a restaurant cooks perfect food, if it is not special in anyway, the restaurant will not score any extra points and hence can only achieve a maximum 80 out of 100. I like to think of the extras section as a measure of specialness and memorability from the dining experience, and no longer about the food itself. In practice, I’ve noticed that most restaurants get some extra points one way or another.

Another notable difference between the extras and basics section is that for all categories in the extra section, the default score is 0. In fact, one category may even go negative for particularly bad experiences. This is not true for the basic categories, where the ‘average’ is 6/10.

The Categories

With the big picture framed, let’s go over the rating categories.

In the basics section, there are 3 categories that sum to 80.

Food: Flavor and Texture (60 points)

Does the food taste good? Does the dish come together?

It goes without saying that this category is the biggest; how the food tastes makes or breaks the place. Notably, I also chose 60 because it is a very divisible number. You can divide this category into chunks for each dish and assign different weightings. For example, if you got an entrée and an appetizer, you can allocate the appetizer 20 of the 60 and the entree the other 40. Or if you got 5 tapas, then 12 points for each, I suppose. Normally, I tend to score dishes out of 10 for my own sanity, then scale that score to the allocated weighting.

Food: Quality and/or Skillfulness (10 points)

How are the quality and freshness of the ingredients? Does the chef’s skillfulness and effort shine through?

On paper, the idea is that you can taste ingredients that are good quality, and be able to discern food prepared with the highest level of skill and effort. But truthfully, this category exists mainly because I think that some junk food or fast food tastes delicious, but I just cannot say that’s objectively good food.

I’ve merged quality and skillfulness into one category because good food doesn’t necessarily have to involve a ton of skill — think a salad or KBBQ. In those cases, quality is the driving factor. Combined, you can think of the ultimate question as how the chef leverages their skill to showcase the quality of the food.

Value (10 points)

Is the meal good value, given the context?

I know some people do not care about price at all when reviewing food, while others’ experiences of dining are completely determined based on the price of the food. Isn’t there a saying like free food is the best food? For me, I’ve given value only 10 points: it’s my way of saying I acknowledge the price, but it doesn’t dictate my food experience.

In the extras section, there are 4 categories that sum to 20. Remember: think memorability and uniqueness.

Service and Setting (-5 to 5 points)

Thoughtful service? Cool spot? Good locale or atmosphere?

I personally don’t care too much about service or setting. In fact, one of my favorite restaurants, a hole-in-the-wall place in Shanghai where you often have to share tables and order from a Shanghainese ayi with an attitude, would probably score a -3 in this category (check out my article, A Local Shanghainese Food Tour). Speaking of negatives, a restaurant can receive a negative score in this category for below average service or setting. Interestingly, a negative score also implies a memorable experience, just in a bad way.

What a -3 Looks Like: It Could Only Get Worse if the AC Broke in the Humid Shanghai Heat

Presentation (0 to 5 points)

Is the food Instagram-worthy? Does it look delicious?

This is where culinary skill sort of becomes an art. Like a good artist, does the chef make their creation look beautiful, or leave you in awe? Also, is the food presented in a way that makes it look appetizing and delicious?

Legitimacy or Creativity (0 to 5 points)

Is the food authentic with potentially significant history? Or is it unique and creative that it pioneers the field?

This category encapsulates the legitimacy or creativity of the restaurant, as well as how much thought was put into each dish.

Misc / Memorability (0 to 5 points)

Special place? Special dish? Special memory created here?

This last category is probably the most subjective — it completely differs based on person and the context of the restaurant experience. I think of this criterion as how memorable my overall dining experience was.

Score Correspondence

After you’ve given a score for each category, sum them up to get the final score. Generally, I’ve found that scores roughly correspond to the following:

  • 85+ Next level restaurant; will go if possible. Worth at least 2–3 hour lines.
  • 80+ Worth at least 1–2 hour lines. I will actively come back. These are the places I’ll evangelize nonstop to my friends.
  • 70+ Very tasty food; I can have this for 2 meals in a row. I will come back if in the area, and I’ll recommend you to try it out too if you’re in the area.
  • 60+ Good food. I don’t regret coming here.
  • 50+ Decent. Not incentivized to come back
  • 40+ Not bad. Will not come back.
  • 30+ Could be worse, but I’ll probably tell you to avoid.
  • 30> Worse than my own cooking; not worth the time and money to come again.

As you can see, I’m not a good cook; I’d probably give my own cooking a 30/100.

Correspondence to Stars

From my rating distributions, I’ve concluded the following score mapping to star ratings:


I know that my focuses and consequent weightings for each category only reflect my personal preferences. Feel free to adjust the weightings, or even add or remove categories from my system. My goal is to at least encourage you to think about what aspects make a good dining experience? Only with a more comprehensive and consistent scale can we compare, apples to apples, our own dining experiences.

Want to see my rating system in action? Check out my instagram at food.footprints.