People see the number — 25,000+ venues — and they think "oh, they scraped some API." They picture a script running overnight and a CSV appearing in the morning.
That is not what happened.
Every single venue in our database has been touched by a human. Not reviewed. Not visited. But looked at, categorized, tagged, and cross-referenced. Because bad data is worse than no data. If someone takes our quiz and we match them to a restaurant that closed six months ago, or a bar that’s actually a liquor store, or a "cafe" that’s a Dunkin’ inside a gas station — we’ve lost that person forever.
The raw data we started with was a mess. Multiple sources. Conflicting categories. A place listed as a "restaurant" in one database and "nightclub" in another. Duplicate entries everywhere — the same bar showing up three times because Google, Foursquare, and Yelp all have slightly different names for it. "Joe’s Bar." "Joe’s Bar & Grill." "Joes Bar NYC." Same place. Three entries.
Here’s what the process actually looks like:
Step one: deduplication. We wrote scripts to fuzzy-match venue names within the same neighborhood. Then we manually reviewed every flagged pair. Thousands of them. Some were obvious dupes. Some were legitimately different places with similar names. We got it wrong sometimes. We went back and fixed it.
Step two: categorization. Every venue needs a primary category — restaurant, bar, cafe, entertainment, or activities. Sounds simple until you hit a place that’s a restaurant until 10pm and a nightclub after midnight. Or a coffee shop that becomes a wine bar at 5pm. Or a "lounge" that could reasonably be filed under three different categories. We made judgment calls. Thousands of them.
Step three: subcategorization. A restaurant isn’t just a restaurant. It’s Italian, or Japanese, or a pizza spot, or a steakhouse, or a diner. A bar isn’t just a bar. It’s a cocktail bar, or a dive bar, or a speakeasy, or a rooftop, or a sports bar. We tag every venue with the most specific subcategory we can. Some venues have multiple. A Japanese restaurant with a cocktail bar in the back? Both get tagged.
Step four: neighborhood mapping. Manhattan has 43 neighborhoods on our platform, and the boundaries are not straightforward. Is that spot on Houston Street in the East Village or NoHo? Is the place on 96th and Broadway in the Upper West Side or Manhattan Valley? We used GeoJSON polygons and then manually corrected hundreds of edge cases where the polygon disagreed with local knowledge.
Step five: mood tagging. This is the hard part. This is what makes Moodap different from every other directory. Every venue gets tagged with vibe data: is it romantic? Lively? Chill? Trendy? Cozy? What’s the crowd like? Is it good for groups? For dates? For solo? What time does it peak? What’s the noise level? We derive this from reviews, photos, descriptions, and sometimes just from knowing the place.
Step six: quality checks. Is the place still open? Is the address correct? Do the hours look right? Is the price level accurate? We cross-reference multiple sources and flag anything suspicious for manual review.
And then you do it again.
Because venues close. New ones open. Hours change. A restaurant rebrands. A bar moves locations. A cafe adds a liquor license and becomes a wine bar. Manhattan is a living thing and the data has to breathe with it.
We’re at 27,957 venues as of this writing. We add new ones every week and remove the ones that close. The database is never "done." It’s a living document of Manhattan’s food and nightlife scene.
When someone takes the Moodap quiz and gets a perfect match in 25 seconds, they don’t see the months of data work behind it. They just think "wow, that was easy."
Good. That’s the whole point.
— The Moodap™ Team