"Cauliflower turned into broccoli": a small Tokyo team's photo calorie app squeezes into the charts in Taiwan

Nometer ("Snap Before Meal") launched in Oct 6 and about a month later was around #30 on Taiwan's Health & Fitness Top Free. It has no influencers, no media coverage, and no related TikTok videos. It relies on a traditional-Chinese title stuffed with search terms, near-daily updates, and a full set of invite and social features. The chart's bar is low, though, and it has only 59 ratings so far. The product is worth taking apart, but the category is already crowded.

A four-star review

In Oct 2026 23, a Taiwanese user gave Nometer four stars. They said that when work is busy they often only have time to snap the meal, not upload it, and the app can't backfill, "so one meal goes missing from the record". They also mentioned, "occasionally there's Chinese wording… (cauliflower turned into broccoli)". The last line: "otherwise it's been pretty good to use".

That review is almost a portrait of the app: it's carefully made and aimed at Taiwanese users, but the team behind it isn't in Taiwan. The developer is the godo kaisha Revu (REVU LLC), registered in Shinjuku, Tokyo in Aug 2026 3 with a capital of 1M yen, representative ZHANG XIAOTIAN. Their website calls themselves a "small team across cities and time zones". Nometer is their second product; the first is the SSH tool ConchTalk, launched in June, with 0 ratings in the US store so far.

What it is

Snap a photo of a meal and AI estimates four numbers (calories, fat, protein, carbs) and flags a likely margin of error. If it's wrong, you don't edit the numbers one by one. Just tell it in a sentence, like the line in the website comic "I only ate half the chicken", and the AI recalculates the whole meal. You confirm before it's saved.

On top of that is a heavy social layer: a friends feed (you only share the meals you pick), group chats, direct messages, a regional explore page and leaderboards, a cosmetics store, and an AI companion called "Nanami" who chats with you about what you ate and what to eat next. There's also Apple Watch support, home-screen widgets, a monthly calorie-balance calendar, and read/write access to Apple Health.

The app is free to download. In the Taiwan store, Nometer Pro is listed at NT$90 per month or NT$790 per year (App Store in-app purchase list, 11-06). New monthly subscribers get 7 off the first month. Starting from 1.1.0, Nanami is only available to Pro subscribers.

How big it is now

(App Store Taiwan daily chart and ratings. Apple 11-06 shows 59 ratings in total, 4.98 stars)

Outside Taiwan, only Hong Kong has occasionally charted, at Top Free 55 to 141, with 2 ratings in total. The other hundred-plus stores have never charted. Third-party AppInsider estimates 1k+ global installs.

Take this rank with a grain of salt. 11-06 It sits at #33 on the Taiwan Health Top Free chart with only 54 Taiwan ratings. Neighbors: SlimGo at #29 has 82 ratings, and InOut at #31 has 387. On a chart this thin, a few extra ratings a day moves you up a lot (estimated tens to a couple hundred downloads a day behind it). Top Grossing has held at #104 to #184, so some people are paying subscriptions, but the amounts are small.

Where users come from

Outside, there's almost no buzz. Search "饭前一拍" or "Nometer" and you'll find nothing but its own website and store page; no TikTok videos, nobody talking about it on X, Threads, or Dcard. Almost all of its traffic seems to come from inside the App Store:

• The title is the search term. The Taiwan title is "Nometer 飯前一拍｜熱量計算、飲食紀錄" (Nometer, "eat-before-you-snap": calorie calculation and food logging), and the subtitle is "拍照AI 估算卡路里與營養,輕鬆減重" (AI photo estimates of calories and nutrition, easy weight loss). Both of the terms Taiwanese users search for most are in there. Nearly all its volume is in Taiwan. The Hong Kong store, which uses Simplified Chinese or English, gets almost nothing, which fits (inference, since I can't see search ranking data).
• A new version almost every day. It shipped about 16 versions in its first month, going from 1.0.6 to 1.1.2. 10-14 AppInsider first recorded it, and the next day it was in the Taiwan Health Top Free top 100 (AppInsider).
• Asks for ratings in the app. Only 4 of its 59 ratings have written text. The average is 4.98 stars, and 1.0.8 also has a "write a review" entry in Settings. Lots of ratings, little text, near-perfect scores: that usually means a rating prompt (inference). High scores make the store page convert better.
• Invites and sharing. When a friend finishes their first meal analysis, both the inviter and the invitee get 7 days of Pro. Each user has a short invite code. After the third meal logged, it prompts you to share your "today's calories" card to Instagram Stories or TikTok. These mechanics exist, but I can't see whether they work.
• A comparison post. In Oct 11, its official blog published 〈2026 Taiwan calorie-counting app roundup: 饭前一拍, Cal AI, SlimGo, MyFitnessPal〉, putting itself in the same table as the big brands, with a TDEE calculator on the side. A Japanese version compares あすけん, カロミル and Cal AI. Only three posts so far, so the traffic it brings is probably still small.

Is it paying for ads? I couldn't find it in Meta's ad library, and Apple Search Ads can't be checked from outside. But the 1.0.2 update notes say "Improvements to ad attribution and tracking stability," so it has at least wired up ad attribution. It may be spending on a small scale already, or getting ready to (inference).

Who it's competing with

"AI photo calorie counting" in Taiwan and Hong Kong is already crowded. In the Taiwan Health category alone, there are over a hundred apps with calories, calorie counter, or diet in the name, dozens of them new this year, and new ones keep showing up every week.

There's another class of free competitors: iPhone's Visual Intelligence, ChatGPT, and Gemini can all take a photo and tell you the calories. Nometer's answer is that 1.0.8 plugged into Visual Intelligence, so any photo can be logged as "log this meal in Nometer." It turned the competitor into an entry point.

Does the differentiation work?

Its positioning is clearly stated: not the diet app with the most numbers, but the one that is "most honest with numbers and gentlest with people." It gives four numbers, shows error margins, recalculates a whole meal with one sentence, doesn't reward eating less, shares only meals you pick, and has a companion character, Nanami.

But going by the current data, users aren't coming in for these things. Most of them search for "calorie calculator" and arrive without knowing who Nanami is. The reviews praise it for being "convenient" and ask for features like backfilling entries and logging sleep manually. Nobody mentions companionship or "honesty."

There's also a mismatch between what the app says and what it does. The store description says "there is no public leaderboard here," yet the version history includes 1.0.11 "step leaderboard made public" and 1.0.17 leaderboard rankings with medals, plus country and region charts. The product is actually moving toward gamification and community, which pulls against its "gentle, no comparison" positioning.

Social and companion features only count once they actually drive growth. Right now it's adding 0 to 11 ratings a day, and that's not yet visible.

If it were you

As a C-tier opportunity, following it isn't worth it. Photo calorie counting in Taiwan is already a red ocean. The leaders are Cal AI, CaLoMei, and Gymwork, and new entrants show up every week. Nometer itself hasn't proven that its "honest + companion" approach brings volume. Copying it means following someone who hasn't broken out into a packed category. What's valuable is its tactics, which you can apply to other categories and other languages.

Who it's for: One or two people who already want to build an AI tool app for a small language or Chinese-speaking market. Ideally your native language is the target market's language, so you can write store copy that doesn't read like a translation and won't call 花椰菜 "西兰菜" (broccoli).

Who it's not for: Anyone who wants to build another generic AI calorie app and expects organic growth in Taiwan or Hong Kong. Anyone who can't run ads, has no content channels, and only plans to win by being "gentler." In this category, the apps that grew big either leaned on influencer ads (Cal AI) or claimed the local mindshare early (CaLoMei).

How to do it: Take its four moves and apply them to a category that isn't crowded:

1. Put the terms locals search for directly in the title and subtitle, not brand slogans;
2. Publish a "best local XX app" comparison, putting yourself in the same table as the big brands to catch long-tail searches for brand names;
3. "Correct with one sentence, recalculate the whole thing" instead of making users edit numbers one by one. Works for any AI estimation product;
4. Plug into system entry points like iOS Visual Intelligence. Send invite rewards only after the invitee completes their first core action, which prevents abuse and makes sure the people you bring in actually use the app.

If you insist on making a food logger, don't go head-to-head in Taiwan. Go to the markets it hasn't touched: its Japan and Korea users are so few they have to be merged into a single explore bucket, and the Android version isn't out yet (the website says "coming soon").

How to get users: Early on, rely on App Store search: keyword titles, frequent updates, in-app rating prompts, plus one or two local comparison posts. Once retention data holds up, test Apple Search Ads on category terms with a small budget. Build in share cards and invite codes from day one, since they cost almost nothing.

When to move: No rush, this isn't a window that will suddenly close. But the "photo calorie counting" keyword is being fought over by more new apps in every market, and the earlier you claim the search results for a language, the cheaper it gets.

What would show you got it wrong: ① Nometer's ratings climb into the hundreds before December, or it stays in the Taiwan Health Top Free top 10, or Top Grossing holds steadily in the top 50. That would mean "honest + companion" really does keep people paying, and the differentiation is worth copying seriously. ② Its Android or Japanese version charts locally soon after launch, which would show the playbook travels to other markets and there's more room to follow than it looks right now. ③ The reverse: if in a month or two it drops off the charts and updates stop, that means keyword titles and new-listing traffic alone can't sustain it for long.

Data and sources

• App Store Taiwan and Hong Kong daily charts and ratings (10-28 to 11-06); Apple lookup and store page (11-06, including in-app purchase prices, version history, and reviews).
• AppInsider's launch timeline and install estimates.
• Official site nometer.app (Traditional Chinese homepage, FAQ, and comparison posts on the blog); company registration info from revuteams.com.
• External mentions: web search, TikTok, and X searches for 「饭前一拍」 and "Nometer" found no third-party coverage as of 11-06. No Meta ad records found; Apple Search Ads cannot be queried externally.
• Wherever "inference" or "estimate" is written, it's an inference or estimate.