The complete internal analysis behind Market reality (which is the condensed, reader-facing synthesis). Rebuilt from scratch from independent and primary sources; every claim carries a source, a date and a confidence. Competitor detail appears here as analysis; the positioning recommendations stay self-standing (no “better than X”).
Every load-bearing statement is tagged so you can see where it comes from and how sure it is. Nothing here is interpreted or embellished; where something is inference it says so, where a figure is missing it says “open”.
Amt official · Ak academic/peer-reviewed · U independent (press, test portal, professional body) · MF commercial market research · A vendor’s own claim · Aff affiliate/SEO comparison page (often a competitor itself — excluded for opportunity claims)confirmed multiple/primary · plausible one solid source · unconfirmed single weak source or rumourClaims the prior material leaned on that did not hold up when checked. They are first because they are why the from-scratch rebuild was necessary.
U (confirmed).U (plausible, not Apple-confirmed).U.A (confirmed).U.The solution space. Fixpoints are hard limits (by instruction); degrees of freedom are deliberately open.
“Fits any everyday life” (broad capability) pulls against “sharp positioning that gets found” (focus). A position “for everyone” rarely pulls. The resolution to test: a broad product with a focused promise (relief / no judgement) that applies to all because it isn’t a segment feature. Decided on evidence in the positioning section — not asserted as a slogan.
MF)MF)MF, confirmed)The central business risk, and the reason the credits idea addresses a real problem.
U/MF (confirmed).Ak. US: most use free apps, would pay ~$1 max.Ak. → “no ads, no data sale, EU servers” is a paid-for value in DE/EU, far less so in the US.Real German search volumes (Keyword Planner, Aug 2025–Jul 2026 U/Amt-adjacent, confirmed). The demand sits in mainstream vocabulary; emealia’s differentiators are barely searched.
| ~5,000 / mo | ~500 | ~50 | ~0 / no data |
|---|---|---|---|
| kalorienzähler · kalorienzähler kostenlos · kalorien zählen app · kalorien tracken · abnehm app · ernährungstagebuch | abnehmen ohne kalorien zählen · calorie counter app · essenstagebuch | ernährungstracker · food diary app · makro/macro tracker · nährwerte app · kalorienzähler ohne anmeldung · ernährungsberatung app · ernährung schwangerschaft app · intuitive eating app · achtsam essen app (−100% YoY) · essen ohne schlechtes gewissen | food journal no calories (0) · kalorien zählen ohne wiegen · barcode scanner lebensmittel kalorien · diabetes ernährung app · ozempic ernährung app |
Total for the tested set: 32,150 searches/mo (DE); competition on the top terms uniformly “high”. App Store landscape (your own survey): under “Kalorien” a dozen-plus apps, almost all titled “KI Kalorienzähler”; “Essen” is owned by food delivery; “Abnehmen” shows Oviva (DiGA) and weight-loss-for-women clusters.
Evidence-backed contexts, not invented characters.
Ak). Round vs. precise goals: no significant difference (Frech et al. 2022 Ak). → supports ranges and “approximate is enough.”Ak). → labelled ranges are more honest than false precision.Ak); red “over budget” visuals produce intense negative emotion (BJPsych Open 2021 Ak). → the evidence base for the psyche promise.U; cohort n=461,382, 22.4% deficiency in 12 months, detection-bias caveat, 80.5% type-2 diabetes — do not round to “1 in 5 on Ozempic”).Why the category loses people, and the uncomfortable trade-off in the psyche stance.
Ratings ≠ retention. Many apps hold 4.5★ and ~3% day-30 retention — ratings come from the minority who fit the method, not the majority who leave. The four documented quit-reasons (perfection pressure, logging fatigue, numbers without meaning, messy real life) are the real story; each maps to an emealia design choice (see §What emealia can stand on).
Gamification/streaks lift engagement (+40%). emealia’s “no streaks” deliberately forgoes a proven retention lever — it is not a free advantage. Retention must be earned another way (usefulness of the read-back, speed, trust, or a non-punitive engagement format — see §Engagement).
Value demand, honestly. The values (no pressure, no judgement) are in demand — visible in abandonment of harsh tracking, a whole anti-diet field, rising “gentle” downloads — but not in search, and already served by several. So they retain and convert; they don’t get you found. Diet plans also fail: ~60% regain, little effect past 24 months (Frontiers 2018 Ak) → supports “small nudge, not a plan.”
The whole field, not just the five emealia is measured against. Competitor claims are vendor material unless marked.
| Player | Mechanism / input | Price / model (2026) | Note |
|---|---|---|---|
| MyFitnessPal | ~20M crowdsourced DB; voice log; AI Coach; ChatGPT integration | Free capped at 5 entries/day + ads; Premium $79.99/yr ($19.99/mo); Premium+ $99.99/yr ($24.99/mo) — 2 sources | Reach (280M); crowdsource DB error-prone (~27% of entries, some >50% off, Nutrition Journal 2019 Ak); privacy suit pending; acquired Cal AI |
| Cal AI (now MFP) | Photo-first estimation | Hard paywall, ~$29.99/yr | >15M downloads <2yr, “tens of millions” ARR; grew via ~250 paid influencers (TheNextWeb 03/2026 U) |
| Yazio (DE) | DB search + barcode; marketed as “AI calorie counter” | ~€83.90/yr list (heavy discounting/confusion; charged upfront) | Market-leader perception DE; highest-rated in FET directory (4.7); AI oversold, not “just barcode” (check-app.de U) |
| Lifesum | DB + 2024/25 AI pivot | ~$44.99/yr | Habit/plan focus; complaints on AI accuracy/price after pivot (Aff, not re-verified) |
| Lose It! | DB search | ~$39.99/yr | Simplicity; ~10% DB error (Aff) |
| Player | Mechanism | Price (2026) | Note |
|---|---|---|---|
| Cronometer | Search of lab-verified official DB (680,000+ foods: USDA, NCCDB, Canadian NF); labels source per entry; added AI photo logging Sept 2025 (Gold) | Free tier (84 nutrients, verified DB, no ads); Gold $49.99/yr (2 sources), ~$9.99–10.99/mo | Nutrient depth + official provenance with source label — exactly what is often attributed to emealia; slow/“cold”, point numbers not ranges |
| MacroFactor | DB search + adaptive targets from weight trend | No free tier (7–14d trial); $11.99/mo, $71.99/yr | Adaptive TDEE is the USP; no desktop |
The “45 seconds” is a subjective reviewer impression, and sources disagree: one says ~45 s per meal, another 45–90 s per food item (2–5 min/meal). Only defensible: the search-select-verify flow is noticeably heavier than free text; the exact seconds are not citable as fact.
| Player | Claim | Evidence status |
|---|---|---|
| Welling | “2.6 s avg log, 95.6% accuracy over 15,000 meals”; built for non-Western/mixed meals | All self-reported; no independent verification found A (unconfirmed) |
| Kalory AI | One-time ~$5, no subscription | App Store listing A |
| SnapCalorie | Photo + depth sensor, ~16% mean error | Review figure Aff |
Segment fact: photo-AI error is structurally 0.10–38.3% by meal, 25–35% for mixed dishes (Nutrients 2024 Ak). “Easy AI logging” is table stakes, owned by the largest distributors. None of the checked players shows uncertainty as a range or holds the no-red/no-verdict stance — a design difference, to be re-checked against specific candidates rather than asserted as “nobody.”
| Software | Vendor | Platform | Price (one-time) |
|---|---|---|---|
| PRODI | Nutri-Science | Windows desktop (client-protocol web widget only; DB: BLS 4.0+3.02, Expert + Souci·Fachmann·Kraut, USDA, AFCD) | 295–1,995 € + 99–459 €/yr updates |
| Nutri-Guide | Nutri-Science | PC | from ~215 € |
| EBISpro | ebispro.de | PC | ~400 € |
| Diät2000 / OptiDiet / DGExpert / … | various | PC | 150–1,500 € |
Built for professionals; no consumer product, no free-text/conversational input. — FET e.V. U (confirmed).
| Platform | What it does | Price |
|---|---|---|
| Foodzilla | Meal planning, white-label client app, client mgmt, booking, telehealth, payment processing, per-client targets | from $37/mo, free trial + free version; 4.3/5 (47) — GetApp U |
| Nutrium | Meal planning, client mgmt, appointments, billing/invoicing, client app, secure messaging, HIPAA | from $35/feature/mo; 4.2/5 (38) — Capterra U |
| Cronometer Pro | Dietary analysis for professionals / clinics / research | not public (“contact vendor”); 4.5/5 (11) |
Consequence for a professional view: the bar is high (full planning + app + billing + telehealth from ~$35–37/mo). emealia’s thin, non-clinical layer wins only by staying deliberately simpler — consistent with the “no invoicing/CRM” fixpoint.
Amt, esanum U.U)Who is behind it: largely “vibe-coded” apps built on Cursor, Lovable, Replit, Bolt by people with little coding background; Apple is throttling (review times 7–30+ days, update blocks, removals). Quality evidence: audits of 26 AI-built apps found 23 had zero tests, 17/21 no error tracking, 12/14 no rate-limits — problems mostly invisible (security, no tests), not on-screen (axonbuild U). The distinction that matters is engineering discipline, not “built with AI.” A technical founder who adds the invisible infrastructure (tests, auth, rate-limits, error tracking) is on the right side of that line; being AI-assisted is not itself the risk.
Visibility is the bottleneck, not the feature. Another “AI calorie counter” without sharp, defensible differentiation disappears into the noise regardless of its quality.
Prior analyses claimed an empty middle (“numbers without judgement / nutrient focus, that nobody does”). It does not hold.
Verdict: there is no empty niche. What emealia might hold is only the specific combination for a specific person; every single building block is occupied. “We do what nobody does” is false and must not be used.
Correction, important for positioning.
“Official data” is not a differentiator: Cronometer uses USDA/NCCDB/CNF and labels the source per entry (in its free tier); German professional tools rest on BLS. Crowdsourced DBs (MFP) are the exception at the low end, not the rule. The BLS is licensable (~€100 download / ~€2,000/yr server; Max Rubner Institut, ~15,000 foods × ~130 values). Whether a German consumer tracker uses BLS as its base (as emealia does) is unconfirmed — Yazio/FddB run their own DBs (to verify). emealia’s edge, if any, is the combination (BLS-German accuracy + labelled estimate for unknowns + ranges + speed + no judgement) — each part contested.
Where the field sorts. This is the mapping, not the recommendation.
The five verdicts (sensible / desirable / realistic / unrealistic / only-with-budget) are not mutually exclusive — a position can be realistic + desirable + only-with-budget at once. So positions are scored on independent axes; verdicts are derived combinations. Growth is not the top goal — sustainability for one person is.
| Axis | Question |
|---|---|
| Demand-fit | Real demand? Discovery- or retention-relevant? |
| Differentiation / defensibility | Dunford: real alternative, what emealia has that they don’t, “so what”, does it hold vs competition? |
| Category clarity | Understood instantly, without teaching a new category? |
| Marketing pull | Does the message solve a heard problem? |
| Findability | Serves searched vocabulary / clears the discovery bottleneck? |
| Buildable & operable (solo) | Buildable and sustainably operable by one person (support, content, review queue, regulation)? |
| Founder-market fit | Matches what the founder credibly embodies? |
| Monetisability | Pays despite the DE barrier? |
| Unit-economics plausibility | LTV:CAC ≥ ~3:1 under realistic organic acquisition (weighting sustainability over scale)? |
| Retention / PMF signal | Reason to believe people stay (against the 3% norm)? |
| Risk | Regulation (MDR), ED-safety, overclaim, platform dependence |
Sober probability: a broad consumer position against Yazio/MFP is unrealistic solo. A narrow, sustainable niche is possible but not likely without sharp focus and without several open bets landing.
+ advantage/met · o neutral/partly · – disadvantage/not met. Verdicts are combinable.
| Position | Demand | Diff. | Pull | Find. | Solo | Monet. | Reten. | Verdicts (combinable) |
|---|---|---|---|---|---|---|---|---|
| P1 · “real numbers / official sources” | + | o | o | + | + | o | o | sensible as a findability anchor, not as differentiation |
| P2 · “without pressure / no judgement” | –/+ | + | + (post-click) | – | + | o | + | sensible + desirable as a promise; unrealistic as discovery |
| P3 · everyday flexibility | o | o | o | – | + | o | o | product property, weak as standalone position |
| P4 · “companion, not scoreboard” | – | o | + (tone) | – | + | — | o | brand voice, not discovery |
| P5 · consumer×professional (thin) | – | o | – | – | o/– | o | + | add-on/retention; full tool unrealistic/only-with-budget |
| P6 · medical edge-groups | – | o | – | – | – | o | + | add-on/retention; medical build forbidden |
| P7 · family/household | ? | o | o | – | o | o | o | feature; market unproven — verify first |
| P8 · privacy / no data sale | – (search) | + | o | – | + | — | + | trust pillar (worth most in DE), not a category |
| P10 · “deficit view, warm not cold” | o | + | o | – | + | o | + | secondary differentiation feature |
| P-niche · everyday-fit + real reason + founder-fit | o (community) | + | + | – | + | o | + | sensible + desirable + realistic; slow/organic acquisition |
| Broad brand vs Yazio/MFP | + | – | o | + | – | o | – | desirable but unrealistic / only-with-budget |
emealia’s “not a chat” stance, tested against HCI evidence.
U). Pure chat calculation (ChatGPT) is 27–36% off on energy, 60–110% on protein, portion under-estimated in ~76%, and stateless (no log/history) Ak-based.No (see §premise). No-counting and nutrient-focus are occupied, growing fields. Only the specific combination + person is open; don’t claim an empty category.
Yes, but hidden (abandonment, anti-diet field), not in search, already served, and hard to monetise; plus the gamification trade-off. They retain/convert; they don’t acquire.
No evidence of a sized/searched segment. Batch/meal-prep is a convenience feature; “evening-parent-logger” is a design persona. A feature, not a positioning — verify with interviews.
Exists (Cronometer/FiberUp/Proto); small search but viable niches. A chooseable focus is a sensible cheap feature (fits the deficit view), not a differentiator — if one, protein.
Suggestions are broad (MFP AI Coach/meal plans, MacroFactor adaptive targets, Noom coaching, Foodzilla plan generation) — emealia is behind here, not ahead. emealia’s framing is deliberately a small nudge, not a plan (plans fail); that demand must be verified. Contributions ≈ Cronometer’s per-food nutrient breakdown; the band-contribution view is a UX novelty, not a new capability. The clearest unmet wish is “accurate and simple and not anxiety-inducing” — the contested no-counting/accuracy space.
Absolute claims demonstrable by a self-test — no comparison needed. This is the recommended way to position, given “no better-than-X.”
“Write down your whole day in a sentence.”
Measure and publish the seconds for an evening entry / a restaurant meal — an absolute number. Answers the top quit-reason (logging fatigue). Honest caveat: “fast and calm,” not “fastest possible” (voice exists).
“Rough numbers from real national food data — and it tells you when it’s a guess.”
Provable in the product; the label is visible. Quality is in the data — the part to protect from overclaim.
“It never turns red on you.”
Provable because it simply never does it; maps to the perfection-pressure quit-reason.
“No ads, nothing sold, data in the EU.”
Worth paying for in DE/EU specifically (privacy premium); much less in the US. A reason to weight markets.
U).Web app → App-Store findability is irrelevant; channels differ, and the keyword demand matters more (Google is the front door).
U).U).Tracker cluster ~€5–10/mo (Cronometer Gold $49.99/yr [2 sources], MacroFactor $71.99/yr, MFP Premium $79.99 / Premium+ $99.99, Yazio ~€83.90). Programme cluster (Noom ~$70/mo, Alma $199/yr) is irrelevant. emealia (subscription + credits, no free-forever) competes fairly on price with the paid tiers; the harder competition is the free tiers — a capable free tier gives official data + nutrients at no cost, and “free” is a top search term. The real question is what the paid experience gives that a free tier doesn’t (calm speed, no judgement, German-data accuracy, privacy), and whether credits + try-without-payment-data lowers the German barrier. Unproven; test early. At ~3% D30 retention and a small niche, unit economics are fragile without proven retention — and growth is not the goal; sustainability (retention × pay-rate) is.
A hypothesis to run in parallel to the “paid, no free-forever” assumption: because emealia will hold a curated official database, the plain tracker — manual search + barcode logging against that database, with emealia’s calm ranges, no judgement, German data and privacy — could be offered free, while the AI free-text flow, estimation and suggestions sit behind subscription or credits. Is that sensible, especially in Germany?
The failure mode above comes from putting all AI behind the wall. The benchmark evidence points the other way: freemium works when users can sample the core before the paywall. The credits model resolves it — free tier = unlimited manual database tracking + a bucket of free AI credits so the fast free-text flow is actually felt; then a subscription or more credits for continued/heavy AI. This caps AI cost (credits bound the spend), gives the differentiator a chance to convert, and is maximally on-brand for Germany (“free to use, AI when you want it” — no subscription trap).
It does not remove the core monetisation bet — it reshapes it: the question becomes “do enough free users, having tasted the AI, pay for more?” That is testable early with the credits mechanic, and should be tested before a free tier is committed to.
Added after a review-mining pass and a check of emealia’s own feedback files. Google Trends was attempted directly via the browser and returned HTTP 429 (automated access blocked); the Keyword Planner volumes above are the stronger source in any case (absolute, not relative).
The alpha/prototype/product feedback in the repo is the founder’s own build-and-design log and bug reports (draft→logged flow, band-vs-number decisions, the suggestion box, favourites for re-logging) — valuable for product intent, but not end-user demand data. The primary-market question (do target users have the pain, want the relief-not-precision frame, and pay?) remains open and needs interviews.
| Recurring signal | Where seen | Reading for emealia |
|---|---|---|
| Database inaccuracy is the #1 trust complaint | Reddit synthesis; MyFitnessPal forums | Confirms accuracy/provenance as a felt need — but it must be shown, not claimed |
| Feature-bloat resented — “jack of all trades, master of none”; search got slower; peripheral features (scores, podcasts, coaching) crowd the core | Cronometer forum | Strong support for a calm, focused, fast core; the direction to protect is restraint, not more features |
| Paywalls on basic tracking + clunky UI | Reddit synthesis | The free-tier/price tension again; a calm, clear UI is valued |
| Speed / effortless logging valued; voice explicitly wished for | Reddit synthesis | Reinforces fast entry as the felt need — and that voice is a demanded modality (a real check on any “text is fastest” claim) |
| Community/accountability valued by some | Reddit synthesis | A lever emealia’s no-social/no-pressure stance forgoes — a conscious trade-off, not an oversight |
| One-size-fits-all disliked; personalised goals valued | Reddit synthesis | Supports a chooseable focus / personalised targets (already in the product intent) |
Net: the felt needs are accurate, fast, simple, fairly priced, personalised — and bloat is actively punished. This lines up with the four quit-reasons and argues for depth over breadth: refine the core, resist feature accretion, prove accuracy and speed. The two demanded things emealia deliberately omits — voice and community — are trade-offs to make consciously, not gaps to fill by reflex.
Yazio runs a large mixed database with a per-country “Food Database Country” setting and ~6.1% average deviation from lab measurement (mid-table); it does not advertise official-source (BLS) provenance or per-entry source labels. So emealia’s BLS-first, source-labelled base remains a plausible German-accuracy differentiator against the market leader — still to be confirmed by a direct food-by-food comparison, not asserted. — Yazio Help Center / calorie-trackers.com 2026 U/A.
MF; Deloitte Mobile Health Consumer Trends 2026 (n=2,000) U/MF; Uncovska et al. 2023 Ak; Google Keyword Planner export (DE, Aug 2025–Jul 2026); media.market.us MF; Business of Apps U; Prince & Wallsten (privacy by country) Ak.Ak.Amt, esanum 2025, nutri-science.de, TechSpot, axonbuild, eated.io, fitbudd, calorie-trackers.com, Yu-kai Chou, Morgan Stanley, Indie Hackers, designative.info, Intercom, kcalm.app.