Editorial · Methodology

How we decide "how long does it last?"

Every shelf-life number on this site comes from a documented chain: an official source, a manufacturer override rule, an editorial review pass, and a scheduled re-check. Here is exactly how that chain works.

The short version We rank sources from A (official US food-safety agencies) down to C (editorial synthesis). When a manufacturer's package instructions are stricter than the category-level guidance, the manufacturer wins. Every item is re-reviewed at least once every 12 months and the review date is visible on every page.

The four-tier source hierarchy

Every claim on every item page traces back to a source. We rank the sources by independence and rigor, and we cite them inline on every item page — not buried in a footer, not pointed at a generic hub page.

Tier A · Official direct source

A US government and statutory food-safety agencies

USDA FoodKeeper, USDA FSIS food product dating guidance, FoodSafety.gov, and FDA storage and labeling guidance. These are the anchor sources for storage durations, date-label interpretation, and the safety-vs-quality framing on every item page. UK and Ireland coverage (FSA UK, safefood, FSAI) will be added as those geographies receive proper category support.

Tier B · Manufacturer and technical guidance

B Package-specific instructions and industry technical bulletins

When a specific product's packaging or a manufacturer's FAQ specifies storage instructions that are stricter than the category default (for example, a refrigerated condiment whose label says "refrigerate after opening, use within 30 days"), the manufacturer instructions override the generic guidance. We never relax safety guidance based on manufacturer claims — only tighten it.

Tier C · Editorial synthesis

C Reasonable editorial defaults

For items where official sources don't address a specific state (e.g., "opened and refrigerated coconut milk") and no manufacturer guidance exists, an editor synthesizes a conservative duration from the nearest A-tier comparable item plus the underlying microbiology. C-tier durations are intentionally biased toward the shorter end.

The manufacturer-override rule

This single rule is the simplest version of the trust contract:

If the package says something stricter, the package wins. WillItExpire's category-level durations are baselines, not ceilings. If your bottle of olive oil says "use within 60 days of opening" and our page lists 2 months, we agree. If your jar of mayo says "discard after 30 days" and our page lists 60, follow the jar.

We do not flip the rule the other direction. A manufacturer that says a product is good for longer than the category baseline does not override our shorter duration — manufacturer marketing claims face commercial pressure that food-safety agencies don't.

Confidence levels

Every item page shows a confidence pill directly under the title. The level is derived mechanically from the strength of the sources cited:

High confidence Medium confidence Low confidence

High means the duration is supported by at least one A-tier source, typically USDA FoodKeeper plus FSIS dating guidance. Most items on this site are high-confidence.
Medium means the duration is supported by A-tier guidance for the category but not for the specific item state — for example, "opened plant-based milk" rests on dairy-category guidance plus B-tier manufacturer ranges.
Low is reserved for items where no A-tier or B-tier source addresses the specific question and the duration is editorial synthesis.

Risk classification

Next to confidence, every item carries a risk-posture pill that answers a different question: "If I eat this past the listed duration, what's actually at stake?"

Safety concern Quality concern Package-specific

Safety concern means the duration is about pathogen growth or toxin production. Eating significantly past it carries real foodborne-illness risk — raw meat, fresh dairy, leftover proteins, cooked rice (Bacillus cereus), seafood.
Quality concern means the duration is about flavor, texture, nutrition, or peak appearance. Past the duration the food is usually still safe if stored correctly — pantry items, most condiments, baked goods, oils.
Package-specific applies mainly to household items (medications, sunscreens, supplements) where active-ingredient potency is the metric, and the package's printed date is the source of truth.

Spoilage signs and the "when in doubt" rule

Every item page lists 3 spoilage signs that flip the calculation. These come from the same A-tier sources and from the underlying microbiology — slimy texture on protein, sour smell on dairy, mold on baked goods. The rule we apply to all of this:

When in doubt, throw it out. The cost of discarding food you weren't sure about is at most a few dollars. The cost of a foodborne illness is days of sickness, possible hospitalization, and (for some pathogens like Listeria in pregnant people or vulnerable adults) much worse.

Do not taste-test food you suspect might be spoiled. Many foodborne pathogens don't change flavor noticeably — a tiny taste is not a safety check.

Review cadence

Every item carries a "last reviewed" date in its footer. The default review cadence is every 12 months. When a source we cite publishes a meaningful update (e.g., USDA FoodKeeper revises its category guidance), affected items get a forced re-review regardless of how recently they were last touched.

When the data model itself changes — for example, adding opened-vs-unopened state variants — every affected item gets a re-review pass as part of the migration. The review date is bumped automatically when a meaningful change ships to the page.

Corrections and additions

If you spot an error, have a manufacturer source we should incorporate, or want to suggest an item we don't yet cover, file an issue on the project's GitHub repo. Verified user submissions through a structured workflow are on the roadmap (Tier 2 of the project plan); for now, GitHub issues are the supported correction path.

What we don't do

We don't reproduce manufacturer marketing claims. We don't accept paid placements or sponsored content. We don't generate item pages with AI without an editorial review pass. We don't make a duration shorter just to look more cautious; we don't make one longer just to look more permissive. The guidance you see is the guidance we'd give a family member.