Questions, answered
like questions.

A review platform where every reviewer is a verified human, and every review is weighted by how closely that person's taste matches yours. Instead of one star average meant to describe everybody, you get a predicted score meant to describe you.

Those platforms average strangers. A 4.5 blends the person who wants a quiet room with the person who wants the loudest table in Brooklyn, and the resulting number serves neither of them.

Vouch never averages strangers. It finds the members whose taste actually resembles yours, weights their opinions heavily, and discounts everyone else. It also verifies that the visit happened, which is a claim the incumbents cannot make.

The first is the raw crowd average, included for reference. The second is your predicted score, computed from reviewers weighted by their compatibility with you.

The gap between them is itself the information. A wide gap means the place is polarizing, and Vouch already knows which side of it you land on. A narrow gap means the crowd and your taste twins agree, so you can just go.

Two ways, combined. Stated preference comes from the onboarding quiz, which gives you a usable match on day one. Revealed preference comes from your actual reviews, especially places you and another member have both rated, which sharpens the match over time.

Compatibility is also scoped by domain. You might be 83 percent compatible with someone overall and 99 percent on food specifically.

Because it is the most predictive signal available. A one star review from someone 94 percent compatible with you is worth more than ten five star reviews from people you have nothing in common with.

This cuts both ways, and it is fairer to businesses too. A furious review from someone who was never going to like the place is visibly discounted rather than dragging the average down for everyone.

Because the compatibility math does not work without contributions. If people could read without writing, the graph would be thin, the matches would be vague, and the predictions would be the same mush you get everywhere else.

Three reviews to get in. One contribution a month to keep viewing access.

More than you would expect, on purpose. A full review counts. So does a quick rating, confirming a dish, confirming that a menu or a price is still current, or adding a menu photograph.

This is deliberate. Forcing people to invent experiences on a schedule would fill Vouch with exactly the slop it exists to replace, so the bar is set at something you can honestly clear.

You lose the ability to view other members' recommendations until you catch up. You do not lose your account, your review history, or your position in the compatibility graph.

We punish the lapse, never the graph. Your contributions keep making everyone else's scores sharper whether or not you are currently active.

Layers. One account per phone number and per payment fingerprint, which makes mass account creation expensive. A verified visit signal at the time of writing, either a location check or a receipt. Behavioral rate limits. And graph analysis that catches coordinated manipulation, because paid shills cluster in patterns that honest members do not.

Nobody can make fraud impossible. What you can do is raise the cost of an attack above what the attack is worth, and keep tuning that under real pressure. That work is the company.

No, and it never will. That is a rule we do not get to break later.

AI does the plumbing. It extracts menus from restaurant websites, matches dish names to each other, and helps detect fraud. The experience layer is human only.

Place level data is available to API clients: menus, structured review content, occasion rankings, freshness dates, and aggregate scores. That is information about restaurants, not about you.

A personalized prediction is only ever returned to the agent of the person it concerns, and only after that person has connected it and granted consent. Your agent can ask about you. Nobody else's agent can.

The privacy policy sets this out in full.

No. There is no paid placement on Vouch, and rankings cannot be bought.

If you run a place and something on its page is factually wrong, write to david@diallabs.so and we will look at it. We will not remove an honest review because a business dislikes it.

The thing you get at the end of onboarding. It renders your taste as a shareable card, describing who you are at a table, and gives you a compatibility score with every friend you send it to.

It is not an annual recap. It is the first thing that happens, because comparing taste with a friend is more fun than filling in a profile.

New York City first, a few neighborhoods at a time. Density matters more than coverage here, because compatibility requires that you and other members have actually eaten at some of the same places. A thin presence across five boroughs would produce worse recommendations than a dense one across three neighborhoods.

Free at launch. A paid tier with deeper matching may come later, and the free tier is not going away.

Yes, from your settings, without emailing anyone. Deleting removes your profile, your quiz answers, your reviews, and your photographs. You can export everything first.

The privacy policy explains the two narrow exceptions, which are aggregate figures that do not identify you, and the minimum fraud records the law requires.

Write to david@diallabs.so and a person will answer. If an answer on this page was confusing enough that you had to ask, the page is what needs fixing, and we will fix it.