Foodvisor - AI Calorie Counter reviews

For a chunk of users the recognition is accurate enough to become a genuine delight — they photograph a mixed table and get a sensible breakdown. That 'it actually guessed it!' moment is the single biggest driver of five-star reviews and emotional buy-in, and it's what gets people to pay for a year.

The very mechanic that annoys the efficiency crowd works, for the emotionally-engaged segment, as a real habit engine: users buy outfits for the plant with earned gems and log meals just to make it grow. It turns dull calorie accounting into a game and keeps people on hundred-day streaks — a rare case where a Tamagotchi genuinely moves retention in a utility app.

Barcode, search, AI photo and voice describe let the user pick the fastest path for the moment: barcode at home, photo at a restaurant, dictate on the run. It's this multimodality — not any single feature — that makes people abandon MyFitnessPal: the act of entering food feels fast and pleasant, which is what a daily habit lives or dies on.

Unlike most counters, the app puts fiber on the same footing as protein-fat-carbs and tells you how much you need. For users who care about more than calories, it's a rare reason to choose this one and stay — a concrete, recognisable plus that people call out as 'the only app that does this'. A small but loyal cohort sticks around specifically for it.

The food photo is what the ads sell and what makes people install — but the AI almost always undercounts the portion by 100-300 kcal. For an app whose only job is to keep you in a deficit, this sabotages the core: users trust the number, think they're on plan, and don't lose weight. The diligent ones start double-checking every value, at which point the whole 'fast photo' value proposition collapses.

The virtual plant that grows as you log is, for some, a cute motivator they open the app for. But for the 'I just want to record my food' segment it turns a 20-second task into a chore wrapped in animations. The core pain is the missing toggle: these users explicitly say they're leaving for MyFitnessPal precisely because the tool got slower with no added benefit.

The app paints meals red-amber-green and sends sad faces for 'bad' food. For the segment with an ED history this is a direct trigger: it demonises healthy foods (nuts get a warning, grapes get a green smile) and reactivates old shame. The app asks about disordered eating yet does nothing to safeguard — not a cosmetic issue but an ethical failure in a niche where a large slice of the audience is vulnerable.

The quality score is based on the average meal score rather than your macro targets, so the same food flips between green and orange and adding a glass of water 'upgrades' a meal. Users spot the absurdity (a cheeseburger rates 'good', a protein shake 'ok') and stop trusting the signature layer. The coaching that was meant to be the differentiator reads as random.

Before the first log the app runs a 10-20 minute questionnaire for a 'personalised plan'. It's a behavioural filter: people who came for a simple tracker abandon mid-survey, never reaching the product. Worse, the answers are barely used ('you're a busy parent!'), so the entry cost is high while the payoff for the user is nil.

The app pulls weight in from Health Connect but won't push the thing that matters out — calories, macros, nutrition. For the segment building an integrated health stack (Fitbit, Garmin, Google Health, a Gemini coach), it's a deal-breaker: their food data stays siloed. These users are loyal and pay annually, yet they churn specifically over the export gap — losing the most valuable cohort for a technical, not product, reason.

The app insists you hit a daily target and micromanages each meal, but offers no weekly summary — even though the weekly trend is what actually matters in a deficit. For people experienced with macro counting it's both inaccurate and psychologically punishing: you can't 'make it up' later in the week. The plea for weekly goals is the single most common mature feature-request from retained users.

Search returns a thin list, many everyday foods (spring onion, Asian dishes) simply aren't there, and adding your own is a slog of manual macro entry. This is a segment loss: users from non-Western cultures find their daily food isn't recognised and leave. For an app with AI ambitions, a narrow Western-centric database is a structural ceiling on growth in international markets.

When a user enters their own recipe and sets the serving count, the app multiplies the ingredient weights instead of dividing — a 5-serving meal at 1000 kcal becomes 25000 kcal per serving. It hits the most valuable, engaged segment: home cooks who track precisely. The bug recurs across dozens of reviews over months and quietly erodes trust in the whole tool's accuracy.

The daily target the app sets after the questionnaire runs noticeably higher than third-party calculators for some users. It's a silent failure of the core job: a person dutifully stays under 'goal' yet there's no deficit — and they blame themselves, not the app. The mirror failure is a goal set too low for those trying to gain weight: the 'weight loss only' assumption is wired in so hard it repels whole segments.

The advice section recommends exotica an ordinary user can't get (sea urchins, unidentified fish) and ignores stated preferences and allergies. Because the AI doesn't remember that someone is vegan or lives in a specific country, the suggestions read as generative noise. The feature meant to separate the 'coach' from a plain counter breeds memes instead of trust — and zeroes out the premium value.

The calorie counter ticks up when you add a meal, but the entry isn't saved unless you hit a non-obvious 'Done' button in the corner — so food 'mysteriously' vanishes. For a tracker whose value is a complete diary, this is fatal: people start screenshotting every meal as insurance. Once a user stops trusting that their data will persist, the entire point of the tool collapses.