Food scienceJul 2, 20267 min readBy N. Menon

Your dal entry is probably wrong.

We pulled the 14 most-copied "dal makhani" entries from popular food databases and sent the real dish to a certified lab. The best entry was close. The worst was off by nearly half. This is how a guess becomes a "fact."

If you've ever logged dal makhani, you trusted a number someone typed years ago. Maybe they weighed a restaurant portion in another country. Maybe they copied it from another app. Maybe they guessed. The number then spread from database to database, losing its history at every hop — until it landed in your tracker looking exactly as confident as a lab result.

That's the quiet problem with food tracking: the numbers all look equally true. A gram of lab-tested paneer and a gram of somebody's guess wear the same font.

What we did

We bought dal makhani from six Bengaluru kitchens — two homes, two small restaurants, two cloud kitchens — homogenized each batch, and sent samples to an NABL-accredited lab for a full macro panel. Then we compared the lab results against the 14 most-copied database entries for the dish.

Fig 1 — The same dish, 14 database "facts," one lab panel.

What we found

Per 100 g, the lab average came out at 142 kcal. Here's how the popular entries compared:

Sourcekcal / 100 gvs. lab
Lab panel (avg of 6 kitchens)142baseline
Best database entry151+6%
Median database entry118−17%
Worst database entry76−46%

Read that last row again. If you're cutting on 1,800 kcal and dinner is two katoris of dal makhani, that one entry hides about 190 kcal a day. Over a 12-week cut, that's the difference between a result and a plateau you blame on yourself.

A marked estimate beats a confident lie. The danger isn't the guess — it's the guess dressed up as a fact.

Why entries drift

  • Recipes vary wildly. Home dal and restaurant dal are different foods — butter and cream do a lot of quiet work.
  • Copies lose context. "Per serving" becomes "per 100 g" somewhere along the chain, and nobody notices.
  • Nobody signs their work. Once a number is separated from its source, there's no way to question it.

What we do differently

Every entry on the Mentzer record carries its source: lab certificate, brand label, USDA reference, or community — and estimates are tagged est right on the number. When you tap a food, you see where its numbers came from. When we don't know, the app says so.

Dal makhani now carries our lab panel, split into home-style and restaurant-style entries, because they genuinely are different dishes. Log the one you actually ate.


Foods you want us to send to the lab next? Tell us: hello@mentzer.app.

NM
N. MenonFood data · Runs the lab pipeline

Track on numbers you can check.

Private beta. We open in waves.