Shelf Check
The food industry's accountability report card — every recall, every repeat offender, every cover-up, exposed with data.
Channel: Shelf Check
Tagline: The food industry’s accountability report card — every recall, every repeat offender, every cover-up, exposed with data.
Niche: Consumer-facing food recall intelligence — brand accountability scores, contamination trend analysis, geographic risk maps, and weekly data journalism powered by free government APIs (FDA, USDA, CPSC, EU RASFF)
Target audience: Health-conscious consumers, parents, allergy sufferers, food safety advocates, investigative journalism fans. 80M+ US households buy groceries — food recalls affect everyone but nobody tracks accountability.
Why now: FDA food recalls hit 1,576 in 2025 — the worst year on record. 613 recalls per Trace One, 187 already in Q1 2026 with 53 Class I (serious health risk). Consumer anxiety at all-time highs. Yet NO consumer-friendly site does accountability analytics. It’s all raw government lists or $$$$ B2B tools.
Content Example
”The Dirty Dozen: 12 Brands with the Worst Recall Records in 2025-2026”
Every recall tells a story. But when the same company keeps showing up? That’s a pattern.
Of the 1,576 FDA food recalls classified in 2025, most consumers saw only the ones that made headlines — the Boar’s Head listeria outbreak that killed 10 people, the Diamond Shruumz poisonings, the seemingly endless parade of “undeclared milk” in products that shouldn’t have milk. But behind those viral moments lies a deeper, more disturbing dataset.
We pulled every food enforcement action from the openFDA database for the past 18 months and cross-referenced with USDA FSIS data. Then we ranked companies by a composite “Accountability Score” weighing recall frequency, severity classification, response time, and geographic scope.
The results are not pretty.
At the top of our list: a frozen food manufacturer with 14 Class II recalls in 18 months — all for the same category of allergen violation. Fourteen times they shipped product with undeclared soy or wheat. Fourteen times the FDA classified it as a recall. And fourteen times, they kept doing it.
Our Accountability Score methodology weights four factors:
| Factor | Weight | What It Measures |
|---|---|---|
| Frequency | 30% | Number of recalls per 12-month period |
| Severity | 30% | Class I (deadly risk) vs II vs III |
| Response Speed | 20% | Days between contamination report and recall initiation |
| Scope | 20% | How many states/countries affected per event |
A score of 90+ means the company is effectively self-policing. Below 40 means you might want to check the label twice before buying.
[Interactive chart would show: Animated bar chart ranking 12 brands by Accountability Score, color-coded red/yellow/green, with expandable detail cards showing each company’s recall timeline]
[Geographic heatmap would show: US map with density overlay of states affected by each brand’s recalls]
Why this matters: The FDA doesn’t rank companies. The USDA doesn’t publicly shame repeat offenders. Nobody connects the dots across agencies. That’s what Shelf Check does — we take the government’s own data and hold brands accountable with math, not outrage.
Next week: “Allergen Roulette” — which product categories have the highest rates of undeclared allergen recalls, and which supermarket chains stock the most frequently-recalled brands.
Data Sources
- openFDA Food Enforcement API — https://open.fda.gov/apis/food/enforcement/ — All classified FDA food recalls since 2004. JSON REST API, free key (240 req/min). Updated weekly. Fields: reason, classification, recalling_firm, distribution_pattern, status
- USDA FSIS Recalls — https://www.fsis.usda.gov/recalls — Meat/poultry/egg recalls, RSS feed + XLSX downloads, annual summaries with pound volumes
- CPSC Recalls API — https://www.cpsc.gov/Recalls/CPSC-Recalls-Application-Program-Interface-API-Information — Kitchen/food equipment recalls, JSON API
- EU RASFF Portal — https://data.europa.eu/data/datasets/restored_rasff — European food safety alerts, open data download
- FDA Adverse Events API — https://open.fda.gov/apis/food/event/ — Consumer adverse reaction reports
- CDC FoodNet — Foodborne illness surveillance (background data)
Automation Pipeline
- Schedule: GitHub Actions runs 2x daily (06:00 UTC + 18:00 UTC) for data collection; weekly deep analysis run (Sunday 00:00 UTC)
- Collect: Fetch latest recalls from openFDA API (delta since last run), USDA FSIS RSS, CPSC API, RASFF open data. Store in local JSON/SQLite. Deduplicate by recall number/ID.
- Process: AI analyzes new recalls — classifies hazard type, extracts brand, maps geographic distribution, calculates running Accountability Scores, identifies repeat offenders, detects trend changes. Cross-references across agencies.
- Generate:
- Daily: “Today’s Recalls” digest page with risk cards for each new recall
- Weekly: Deep analysis article (repeat offender rankings, trend charts, category breakdowns)
- Monthly: “State of the Plate” comprehensive report with interactive visualizations
- Per-recall: Individual detail pages (good for SEO long-tail)
- Charts/graphics: D3.js or Chart.js for trend lines, bar charts, heatmaps; AI-generated header images
- Publish: Build static TypeScript site → deploy to Cloudflare Pages (instant global CDN)
Tech Stack
- Static site: TypeScript + Astro (fast builds, great SEO, partial hydration for interactive charts)
- Image generation: Chart.js/D3.js for data viz (rendered to SVG/PNG at build time); DALL-E or Stable Diffusion for editorial headers and mascot illustrations
- Data collection: Node.js scripts — fetch from openFDA, FSIS RSS, CPSC API, RASFF data portal
- Data storage: SQLite database committed to repo (or separate data repo) — enables historical queries
- Search: Pagefind (static search, zero-cost)
- CI/CD: GitHub Actions (2x daily data fetch + weekly analysis + on-push deploys)
- Hosting: Cloudflare Pages (free tier: unlimited bandwidth, instant deploys)
Monetization Model
- Ko-fi / Buy Me a Coffee — “Help us keep watching the watchers” — emotional hook is strong (food safety = protecting families)
- Newsletter premium tier — Free: weekly digest. Paid ($5/mo): daily alerts, brand lookup, personalized risk profile based on shopping habits
- Affiliate: Food safety products — BPA-free containers, food thermometers, allergen test kits (Amazon Associates)
- Sponsorship potential: Food safety certification companies, testing labs, organic/clean-label brands that WANT transparency
- Telegram channel with Stars — Quick daily recall alerts
- API access — Developers/apps pay for pre-processed recall intelligence ($29/mo)
- Projected month-1 revenue: $50-150 (donations + early newsletter subs from viral launch posts)
- Projected month-6 revenue: $800-2,000/mo (newsletter at 2K paid subs @ $5, affiliate, sponsorship inquiries)
- Projected month-12 revenue: $3,000-6,000/mo (API tier, established newsletter, regular sponsorships from food safety industry)
Channel Soul & Personality
Name: Shelf Check
Mascot: Inspector Morsel — a sharp-eyed cartoon raccoon in a lab coat with a magnifying glass, always sniffing out the truth. Raccoons go through your garbage; Inspector Morsel goes through the FDA’s.
Voice: Investigative journalist meets sardonic food critic. Think a data journalist who moonlights as Anthony Bourdain. Passionate about food safety but never preachy — more “let me show you the data and you’ll be furious” than “you should be scared.”
Signature style:
- Every article opens with a surprising data point
- “The Recall Receipts” — weekly segment showing exact numbers, no editorializing needed because the data speaks for itself
- “Brand Report Card” — monthly grading of major food companies (A+ through F)
- “Shelf Check Says” — the raccoon mascot delivers the verdict in a callout box
- Running tally at the top of the site: “FDA recalls this year: [X] | Class I: [Y] | Repeat offenders flagged: [Z]”
Visual identity:
- Color palette: Deep navy (#1a1a2e), warning amber (#f0a500), clean white, danger red (#e63946) for Class I
- Typography: Bold condensed headlines (Impact/Oswald energy), clean body text (Inter)
- Signature element: Every brand gets a “trust thermometer” graphic — mercury rises from green to red
- Data viz style: Clean, high-contrast, newspaper-quality charts — not corporate dashboards
Opinion & stance: Shelf Check believes the FDA is understaffed and underfunded, that food companies calculate recall fines as cost of doing business, and that sunlight (data) is the best disinfectant. It celebrates companies with clean records just as loudly as it flags bad ones.
Running jokes:
- “Another day, another undeclared allergen” — recurring category for the endless allergen mislabeling recalls
- The “Clean Plate Club” — companies with zero recalls get inducted with a fake ceremony graphic
- “Recall Bingo” — a monthly bingo card of predicted recall categories that readers can play along with
Launch Complexity: 3/5
Time estimate: 2-3 weeks for MVP (data pipeline + daily digest + basic Astro site). Full analytics suite with accountability scores in 4-6 weeks.
- openFDA API is well-documented and easy to query
- Most complexity is in the analysis layer (cross-referencing brands across agencies, calculating scores)
- Astro + Cloudflare Pages deployment is straightforward
- Main challenge: normalizing company names across FDA/USDA databases (same company, different spellings)
Content Quality Score: 5/5
- Every data point is sourced from official government databases
- Analysis methodology is transparent and reproducible
- Content is genuinely useful — parents checking if their kids’ food is safe, allergy sufferers tracking brands
- No equivalent consumer-friendly resource exists — this fills a real gap
- Auto-generated content has built-in variety (every recall event creates unique content)
Automation Score: 5/5
- openFDA API + FSIS RSS + CPSC API = fully automated data collection
- AI analysis of structured government data = reliable (not scraping messy HTML)
- GitHub Actions 2x daily = hands-off after setup
- New content generated automatically every day (new recalls happen constantly)
- 1,576 recalls in 2025 = ~30 new articles per week without trying
Revenue Potential: 5/5
- Food safety affects EVERYONE — total addressable audience is essentially all grocery shoppers
- High emotional engagement = high donation/subscription conversion (protecting family health)
- Endless SEO long-tail (every recall + brand + product = new keyword)
- B2B upside (API tier for apps, food companies monitoring their competitors)
- Sponsorship from “clean” brands and food safety industry
- Newsletter has clear premium value proposition
Total: 18/20
Why This Will Work
Psychology: Fear + empowerment = action. People don’t just want to know about recalls — they want to feel IN CONTROL of their family’s food safety. Shelf Check gives them a brand they trust more than the brands on their shelves. The accountability scoring creates an emotional hook: “Is MY grocery store’s brand safe?” That’s a question people will pay to answer.
Market logic: 1,576 recalls per year = 1,576 automatically generated SEO-friendly articles per year. Each one targets a unique long-tail keyword (“[brand] recall [product] [year]”). Over 12 months, that’s a massive content moat that’s impossible to replicate manually. And the data is free from the government — zero content acquisition cost.
Viral mechanics: Accountability rankings are inherently shareable. “Your favorite frozen pizza brand got an F on food safety” is a headline people screenshot and send to group chats. The monthly “Dirty Dozen” ranking will generate controversy and backlinks every time.
Defensibility: The analysis layer (Accountability Scores, cross-agency correlation, historical trends) creates intellectual property on top of public data. Anyone can pull FDA recalls — nobody else is scoring brands, mapping geographic risk, and writing smart analysis automatically.
Risk & Mitigation
- Legal risk from naming brands: MITIGATED — all data is from public government databases. Accountability Scores use transparent, documented methodology. This is journalism/analysis, protected speech. Include clear methodology pages.
- Data quality issues (company name variants): MITIGATED — build a company alias database early. “Nestlé USA” vs “Nestle USA Inc” vs “Nestle Prepared Foods” all map to one entity. This is engineering work, not a blocker.
- FDA API downtime or changes: MITIGATED — cache data locally in SQLite, fall back to web scraping of FDA dashboards if API breaks. Multi-source (FDA + USDA + CPSC) means no single point of failure.
- Content fatigue (too many recalls): MITIGATED — daily digest format keeps volume manageable. Weekly deep dives provide the analysis layer. Not every recall gets a full article — the AI filters for significance.
- Competition from government improving their own dashboards: UNLIKELY — government sites have remained basic for decades. Even if they improve, they won’t do editorial voice, brand scoring, or data journalism.