Document M-001 - v1 - ratified 2026-05

The Don't Eat Framework v1

A public scoring framework for Indian packaged foods. It explains what we measure, how a verdict is assigned, and where the system deliberately stops short.

Active versionv1
Assessment axes06
Verdict tiers04
Source groups07
What's new in v1

Changes from the previous version.

  • v1 realigns the framework to ICMR-NIN Dietary Guidelines
  • for Indians (2024) — India's national dietary authority —
  • and to the FSSAI thresholds being adopted for front-of-pack
  • nutrition labelling under Supreme Court direction.
Read full note (39 more)
  • Five substantive changes from v0:
  • — Sugar load axis switches from per-serve to per-100g
  • (solid) and per-100ml (liquid) thresholds. Adds a
  • children-targeting modifier: sugar in top-5 ingredients
  • plus child-directed marketing escalates the flag.
  • — Processing tier axis no longer treats NOVA Group 4 as
  • an automatic red flag. NOVA 4 escalates to red only when
  • combined with a fail on another axis. Fortified staples
  • and minimally-additive packaged foods that are NOVA 4
  • with otherwise clean composition receive a yellow flag,
  • not red. This aligns with peer-reviewed critique of the
  • NOVA framework's heterogeneity.
  • — Fat / oil type axis renamed to Fat profile. Palm oil is
  • no longer flagged independently, in line with ICMR-NIN's
  • explicit recommendation to consume a variety of vegetable
  • oils including palm. The axis now centres on total
  • saturated fat (per 100g/100ml) and trans fat. Palm oil
  • remains relevant in the Marketing claims axis when
  • products are positioned as premium, natural, or
  • cold-pressed.
  • — Sodium axis switches from per-serve to per-100g/100ml,
  • aligned with the WHO-SEARO threshold (≥625mg/100g for
  • solids, ≥300mg/100ml for liquids) being adopted by FSSAI.
  • — Marketing deception axis (renamed to Marketing claims)
  • gains explicit deception patterns with regulatory
  • precedent: "health drink" claims (NCPCR/FSSAI 2024),
  • "100" claims (FSSAI ban May 2025), unsubstantiated
  • "sugar-free," "natural," and "immunity" claims, and
  • child-targeted health marketing.
  • The decision rule is unchanged: any one red flag results
  • in Don't Eat. Two or more yellow flags result in Think
  • Twice. Otherwise, Eat. The verdict is computed
  • deterministically from axis flags.
  • All products previously assessed under v0 are flagged for
  • re-assessment under v1 and pass through editorial review
  • before publication. The methodology version applied to
  • each product is shown on the product page.
  • — v1 refinements (2026-05): The Pending Review verdict now requires four or more axes with insufficient evidence (previously three), so a single missing data point no longer overrides an otherwise complete assessment. The Marketing claims and Regulatory history axes default to GREEN when Tier-A sources (brand site, Indian e-commerce listings, FSSAI / ASCI / NCPCR registries) return no findings — absence of adverse evidence is now read as a clean record, with Insufficient reserved for genuinely unreachable sources. The Processing tier axis is auto-classified from the back-of-pack ingredient list when it is on file, using NOVA-4 markers (polyols, emulsifiers in the 4xx range, modified starches, hydrolysed proteins, artificial flavours, high-fructose corn syrup, maltodextrin). When a contributor submits a back-of-pack label, the assessment is automatically re-scored using the label data while the original web-sourced product description is preserved and marked with a contributor-update note.
  • — v1 refinements (2026-05): The Pending Review verdict now requires four or more axes with insufficient evidence (previously three), so a single missing data point no longer overrides an otherwise complete assessment. The Marketing claims and Regulatory history axes default to GREEN when Tier-A sources (brand site, Indian e-commerce listings, FSSAI / ASCI / NCPCR registries) return no findings — absence of adverse evidence is now read as a clean record, with Insufficient reserved for genuinely unreachable sources. The Processing tier axis is auto-classified from the back-of-pack ingredient list when it is on file, using NOVA-4 markers (polyols, emulsifiers in the 4xx range, modified starches, hydrolysed proteins, artificial flavours, high-fructose corn syrup, maltodextrin). When a contributor submits a back-of-pack label, the assessment is automatically re-scored using the label data while the original web-sourced product description is preserved and marked with a contributor-update note.
Section 01 - Verdicts

Four outcomes, one reading order.

The verdict is a shortcut, not a black box. Each product is read through the same six axes, then the decision rule below converts red, yellow, green, and insufficient flags into a consumer signal.

Eat

No disqualifying signal.

Zero red axes and fewer than two yellow axes. Claims and composition broadly align.

Think Twice

Context matters.

Two or more yellow axes. Usually not an automatic no, but worth reading before regular use.

Don't Eat

Red flags present.

One or more red axes, or materially deceptive marketing. The product page shows the failing axis.

Pending Review

Awaiting evidence.

Four or more axes returned insufficient evidence — usually because the back-of-pack label is not yet on file. Submit the label to upgrade the verdict.

Section 02 - Decision rule

How flags become a verdict.

Rules are evaluated top to bottom. The first matching rule decides the verdict.

01
If 1+ red axis
verdict: Don't Eat
02
If 4+ axes returned insufficient evidence
verdict: Pending Review
03
If 2+ yellow axes
verdict: Think Twice
04
If every axis is decided (red, yellow, or green)
verdict: Eat
05
Otherwise
verdict: Pending Review
Section 03 - The six axes

What we measure.

Each axis has red, yellow, and green criteria. A red flag is a disqualifier; yellow is a watch point; green means the product passed that check on available evidence.

01
Sugar load
ICMR-NIN 2024 · WHO Free Sugar Guidelines
Red

≥10g added sugar per 100g (solid) or ≥5g per 100ml (liquid), OR sugar in top-3 ingredients, OR product targets children with sugar in top-5

Yellow

5–10g per 100g, or sugar in top-5 ingredients

Green

<5g per 100g (solid) / <2.5g per 100ml (liquid), no added sugar in top-5

02
Processing tier
NOVA classification (Monteiro et al.) as referenced in ICMR-NIN 2024
Red

NOVA Group 4 AND fails any other axis (1, 3, 4, or 5)

Yellow

NOVA Group 4 with otherwise clean composition, or NOVA Group 3 with marketing claims

Green

NOVA Group 1–2

03
Fat profile
FSSAI trans-fat regulations · ICMR-NIN 2024 SFA guidance
Red

Detectable partially-hydrogenated oils, OR trans fat >1g per 100g, OR saturated fat >5g per 100g (solid) / >2.5g per 100ml (liquid)

Yellow

Saturated fat 3–5g per 100g, or single-source refined vegetable oil in a product marketed as premium / natural / healthy

Green

Sat fat <3g per 100g, no trans fat, oil source disclosed clearly

04
Sodium
ICMR-NIN 2024 · WHO-SEARO threshold
Red

≥625mg sodium per 100g (solid) or ≥300mg per 100ml (liquid)

Yellow

300–625mg per 100g (solid), 150–300mg per 100ml (liquid)

Green

Below yellow thresholds

05
Marketing deception
FSSAI Advertising Standards · ASCI Code · NCPCR rulings · ICMR-NIN 2024
Red

Front-of-pack claim contradicted by ingredients/nutrition: 'health drink' with sugar in top-3, '100% natural' with synthetic additives, 'sugar-free' with sugar substitutes >5% calories, unsubstantiated immunity claims, child-targeting with health claim, or active FSSAI/ASCI/NCPCR ruling

Yellow

Soft framing ('energy', 'active', 'fortified', 'premium') unsupported by composition; palm oil in product marketed as premium/natural/cold-pressed

Green

Claims align with composition AND no active FSSAI/ASCI/NCPCR rulings. Also GREEN if the product is sold without front-of-pack health/marketing claims, or if Tier-A sources (brand site + Indian e-commerce listings) were checked and no deceptive claims were observed. Insufficient is reserved for cases where the brand pages and quick-commerce listings could not be reached.

06
Regulatory history
FSSAI public record · ASCI complaints register · NCPCR actions · court filings
Red

Recall, ban, court order, or sustained adverse regulatory finding in the last 10 years.

Yellow

Pending complaint, unresolved advisory, or NCPCR-flagged practice not yet adjudicated or any social media issue, controversy regarding product or brand that surfaced.

Green

Clean public record. No FSSAI recalls, no ASCI upheld complaints, no NCPCR action, no pending court matters. GREEN is the correct flag when FSSAI / ASCI / NCPCR / news searches return no findings — absence of adverse findings IS a clean record. Insufficient is reserved for cases where the registries themselves were unreachable.

Section 04 - Scope

What's in scope.

  • The Six-Axis Standard applies to:
  • — Indian packaged food and beverage products sold at
  • retail in India, including via e-commerce.
  • — Products with a declared ingredient list and a valid
Read full note (24 more)
  • FSSAI license number on pack.
  • — Branded products. Generic and unbranded items are out
  • of scope.
  • — Variants of a product (different flavours, formulations,
  • or pack sizes) are assessed individually when composition
  • differs materially. Identical compositions across pack
  • sizes share an entry.
  • Categories currently covered: snacks, beverages, dairy and
  • dairy alternatives, breakfast cereals, malt-based and
  • nutrition drinks, biscuits and confectionery, condiments
  • and sauces, ready-to-eat meals, bakery, instant noodles
  • and pasta, packaged juices and concentrates, protein bars
  • and supplements.
  • Out of scope:
  • — Fresh produce, meat, fish, eggs, and unbranded staples
  • (rice, atta, dal, oil, salt, sugar) when sold without a
  • brand-level ingredient label.
  • — Restaurant and quick-commerce prepared meals.
  • — Imported food without an FSSAI import label.
  • — Infant formula and specialised medical nutrition
  • (separate regulatory framework).
  • — Alcoholic beverages.
  • — Nutraceuticals and supplements regulated outside the
  • Food Safety and Standards Act.
Section 05 - Limits

What we don't claim.

  • What this assessment does not claim:
  • — Not medical or dietary advice. The Six-Axis Standard
  • assesses products against a published framework. It does
  • not account for individual nutritional needs, health
Read full note (34 more)
  • conditions, or dietary requirements. For personalised
  • guidance, consult a registered dietitian or physician.
  • — Not a recall list. We do not initiate or coordinate
  • product recalls. Statutory recall authority belongs to
  • FSSAI. Where a recall is on public record, it appears in
  • the Regulatory history axis.
  • — Not a brand judgment. Each product variant is assessed
  • on its own composition and marketing. A brand may have
  • some products in the Eat tier and others in the Don't
  • Eat tier. The framework treats each entry independently.
  • — Not exhaustive. Absence of a product from the index does
  • not constitute endorsement. New entries are added
  • continuously; corrections are accepted at /corrections.
  • — Subject to composition change. Manufacturers reformulate
  • products. Ingredient lists, nutrition panels, and
  • marketing claims can change between batches. Every
  • assessment is dated. We re-assess when a reformulation
  • or label change is identified.
  • — Subject to data availability. Where primary-source data
  • (FSSAI label image, manufacturer-published nutrition
  • panel, transcribed physical pack) is unavailable, the
  • relevant axis is marked "insufficient data" and the
  • verdict reflects only the axes scored.
  • — Subject to methodology versioning. Products assessed
  • under earlier methodology versions retain their original
  • verdict until re-assessed. The methodology version
  • applied is shown on every product page.
  • — Not a substitute for label literacy. The most reliable
  • guide to what's in a product is the back-of-pack
  • ingredient list and nutrition panel. The Six-Axis
  • Standard exists to make that information findable and
  • comparable, not to replace it.
  • — Contributor labels are reflected. When a back-of-pack label has been submitted by a reader, the assessment is recomputed using the verbatim ingredients and nutrition panel, the original web-sourced description is kept as editorial context, and a small note on the product page records that a contributor-supplied label was used.
  • — Contributor labels are reflected. When a back-of-pack label has been submitted by a reader, the assessment is recomputed using the verbatim ingredients and nutrition panel, the original web-sourced description is kept as editorial context, and a small note on the product page records that a contributor-supplied label was used.
Section 06 - Sources

Authorities cited.

These are the reference bodies and source groups used by the framework. Individual product pages cite the exact documents used for that product.

  • ICMR-NIN Dietary Guidelines for Indians 2024
  • WHO Free Sugar Guidelines
  • WHO-SEARO sodium thresholds
  • FSSAI Advertising Standards & advisories
  • ASCI Code & complaints register
  • NCPCR rulings on child-targeted food marketing
  • NOVA classification (Monteiro et al.)
Section 07 - Versions

Changelog and archive.

The current methodology is v1. Earlier versions remain public so any historical assessment can be read against the framework that produced it.

  1. v1
    active
    The Don't Eat Framework v1 - 2026-05-04
  2. v0
    archived
    The Don't Eat Framework - 2026-04-01
Archive

Expand a previous version to read its axes and the changelog that retired it.