popnetmedia/non-commodity-content

---

name: non-commodity-content description: Use when writing or substantially editing prose meant for publication: articles, blog posts, reviews, guides, long-form marketing copy, landing pages, email campaigns, social posts, product descriptions, case studies, newsletters. Establishes Google's April 2026 "non-commodity content" standard (unique + specific + authentic) with a pre-publish 5-question check. Skip for code comments, commit messages, internal-only docs, README files, slack messages, or any writing not intended for public/SEO consumption.

Non-Commodity Content Standard

Why this exists

At Google Search Central Live Toronto (April 21, 2026), Danny Sullivan codified what Google now actively rewards in the AI-search era. The keynote framed it as the operational version of E-E-A-T:

Tips for AI Search success:

1. Follow SEO fundamentals
2. Make use of structured data
3. Have a great page experience
4. More than anything else: unique, authentic, non-commodity content

Items 1–3 are table stakes. Item 4 is the dominant ranking signal. This skill ensures every published piece passes the non-commodity bar before it ships.

The Three Pillars

Every piece must demonstrate all three:

1. Unique: Brings a viewpoint, information, or content others lack or can't easily replicate. If a competitor could write the same article from public sources alone, it's commodity.
2. Specific: Talks about a specific instance, situation, person, or thing, not general rules, generic steps, or "X tips for..." advice.
3. Authentic: Demonstrates first-hand knowledge: you tested it, visited it, opened the box, photographed the wear pattern, talked to the person, ran the experiment, signed the contract.

Commodity vs. Non-Commodity (Examples Across Domains)

These are Sullivan's examples and a few extensions. Note the pattern: non-commodity titles name a specific instance and signal a non-replicable action ("I refused," "we waived," "I tested," "what happened when…").

Domain	Commodity (avoid as standalone)	Non-Commodity (publish)
Running store	"Top 10 Things to Consider When Buying Running Shoes"	"Why This Customer's Shoes Collapsed After 400 Miles: A Wear Pattern Analysis"
Real estate	"7 Tips for First-Time Homebuyers"	"Why We Waived the Inspection (And Saved $15k): A Look Inside the Sewer Line"
Interior design	"2024 Kitchen Trends You Need to See"	"Marble vs. Grape Juice: Why I Refused to Install Stone for a Family of Five"
Software / SaaS	"5 Tips for Better API Design"	"We Rewrote Our Pagination Three Times in Six Months: Here's the Final Schema and Why It Won"
Product review	"iPhone 17 Pro Review: Everything You Need to Know"	"I Used the iPhone 17 Pro as My Only Camera on a 12-Day Iceland Trip: Here Are the Three Failures That Surprised Me"
Restaurant	"Best Italian Restaurants in Brooklyn 2026"	"I Ate the Entire Carbonara Menu at Misi Last Saturday: Here's the One That's Worth the Wait"
Fitness	"How to Build Muscle After 40"	"I Followed 5/3/1 for 18 Months at Age 43: Here Are My Lifts, Photos, and the Two Mistakes That Cost Me a Year"
B2B / consulting	"Why CRM Migration Matters"	"We Migrated 14 Clients Off Salesforce to HubSpot in 2025: Here's the Cost Spreadsheet and the One That Failed"

The pattern across every domain: specific instance + first-hand evidence + non-replicable proof.

Operational Rules

Apply to every published piece:

1. Anchor every article in a specific event, person, or moment. Open with "Last Tuesday…" / "When I…" / "On our March trip…", not "Most marketers know that…" or "In today's fast-paced world…".

2. Lead with non-replicable detail in the first 2 paragraphs. A photo of the actual thing, a date, a price you paid, a person you spoke to, a side-by-side you ran. If the opening could appear on any site, rewrite it.

3. Include at least one piece of evidence a competitor can't fake: an original photo (no stock), a tasting/test date, a quote from someone you interviewed, an order confirmation reference, a measurement you took, a result you logged.

4. Title patterns to favor:

   • "Why I [refused / waived / kept / chose / abandoned]…"
   • "What Happened When I…"
   • "[Specific subject] After [Specific Event/Time]"
   • "I Tested X Against Y for Z [Days/Months/Miles]"
   • "[Number] Things I Got Wrong About [Topic] in [Time Period]"

5. Title patterns to avoid as standalone pieces (acceptable only when each section contains non-commodity evidence inside):

   • "Top 10…" / "X Best…"
   • "X Tips for…"
   • "Best of [Year]…"
   • "How to [Generic Verb]…"
   • "What Is [Category]?"
   • "The Ultimate Guide to…"

6. For roundups specifically: Every entry must contain at least one first-hand sentence: what arrived, what you tested, what surprised you, who sent it. Generic descriptions sourced from the brand site are commodity.

7. For reviews / how-tos / case studies: State the date, the setup, the conditions. The non-commodity signal is "this happened, here's the receipt."

8. For interviews: A real conversation occurred. Quote, attribute, and date it. If it didn't, it's not an interview.

9. For listicles and "best of" content: Apply the standard per-entry. A roundup of 10 products is commodity if all 10 entries are press-release paraphrases. It's non-commodity if each entry has a dated first-hand observation.

10. Educational / explainer content is the highest commodity risk. To pass: open with a specific incident or experiment that motivates the explanation, anchor each section with a first-person observation, close with your own preference or recommendation backed by experience.

11. Cover the query fan-out, not just the title query. AI search no longer returns one result for one query; it runs 5–20 adjacent queries in parallel (the fan-out), retrieves the best answers across the bundle, and synthesizes one cited response. Before publishing, list 5–8 adjacent questions a fan-out would generate from your title (the why, the how, the what-about, the vs., the after-how-long, the what-if). Answer the obvious ones inline. For the rest, link to a highly relevant internal page, ideally a pillar page in the same topic cluster. If no relevant internal target exists, that gap is a creation signal: trigger a plan for a new piece to fill it. This is the content flywheel, fan-out audits surface gaps → gaps drive new pieces → new pieces deepen topical coverage → coverage wins more citations.

12. Every creation plan includes imagery. Original or task-specific visuals are part of the non-commodity bar. Stock photography and brand-pulled product shots are commodity. When planning a new piece, the plan must specify what images will be produced and how:

    • If the project has Replicate API integration (check the project CLAUDE.md), generate images programmatically as part of the build.
    • If the project does not, the plan must include the image prompts themselves, one per intended image, with subject, style, framing, and mood, so the user can paste them into ChatGPT / DALL·E / Midjourney and produce them manually.

    No piece publishes without imagery decisions resolved at plan time.

Site-Level Prerequisite: Content Map

Per-piece rules only work if the surrounding library is healthy. Before publishing into an existing site:

• The project must have a content map: an inventory of every published piece with its title, primary topic, target query cluster, and pillar/cluster role. If one doesn't exist, build it first; this is non-negotiable for any site of meaningful size.
• Run a cannibalization check before planning every new piece. Search the inventory for pieces already targeting overlapping queries. If overlap exists, the default move is to expand and consolidate the existing piece, not publish a competing one that splits ranking signals.
• Run a coverage check. Where the inventory has gaps in a topic cluster, those gaps are creation candidates (per rule 11).

If the project lacks a content map, that's the first task, not the new piece.

Pre-Publish 5-Question Check

Before any piece publishes, every answer must be YES:

• Could a competitor write this same article from public sources alone? (Must be NO.)
• Does the opening contain a specific date, person, place, or event from my own direct experience?
• Is there at least one piece of evidence (photo, quote, measurement, receipt, test result, conversation) that proves I actually engaged with the subject?
• Does the headline name a specific instance rather than a generic category?
• Does my piece either answer the obvious fan-out questions inline, or link to internal pages that do? (No relevant internal target = create one before or alongside this piece.)

If any answer is NO, the piece is commodity content. Rework before publishing.

What this skill does NOT change

Sullivan's slide explicitly noted: "These align with traditional SEO success, too." Don't abandon the basics:

• Structured data (Article, Review, Product, ItemList, FAQPage JSON-LD): still required
• Internal linking, semantic headings, meta tags: still required
• Page experience (Core Web Vitals, mobile, no intrusive interstitials): still required
• Accuracy, fact-checking, citing sources: still required
• Agent-readable rendering: AI agents now access pages via screenshots, the DOM, and the accessibility tree. Semantic HTML, proper headings, alt text on every image, and critical content rendered server-side (not trapped in client-only JS): still required, now for AI consumption as well as humans.
• Published prices and a bookable action: on any site that sells a priced service, the price range and a machine-callable booking action are prerequisites for being quoted and transacted with at all. See the ai-search-readiness skill.

Non-commodity content is the ranking multiplier. The basics are the floor.

What Google Explicitly Says NOT to Do

Per Google's AI Optimization Guide (Search Central docs), the following tactics are explicitly debunked. Do not do these regardless of what other guides claim:

• No AI-specific files or markup. No llms.txt, no special "AI markdown," no machine-readable bundles built just for LLMs. Google does not consume these.
• Don't pre-chunk content for AI. Write normal sections for humans; the retrieval system handles segmentation.
• Don't rewrite for an "AI voice." There is no special phrasing or structure that earns AI favor over normal good writing.
• Don't chase inauthentic "mentions" across the web. Trying to manufacture brand mentions on third-party sites for AI visibility is wasted effort.
• Don't spin per-keyword variant pages. Generating near-duplicate pages for every query variation is scaled-content-abuse spam under Google's policies.
• Structured data isn't strictly required for AI search. Keep using it for rich results and classic SEO (see "What this skill does NOT change"), but don't treat it as the lever that earns AI citations.

How to apply this skill in a session

1. When invoked, run the 5-question pre-publish check on any draft before suggesting it's done.
2. When asked to write something new, anchor the structure in a specific real event the user can confirm or describe; ask for it if you don't have one.
3. When asked to edit existing copy, identify the commodity-vs-non-commodity score per pillar and propose specific, surgical edits rather than full rewrites unless the entire frame is commodity.
4. When in doubt about whether a piece is commodity, look at the opening sentence: if it could be the opening of any article on this topic, it's commodity.

Sources

1. Danny Sullivan keynote, Google Search Central Live Toronto, April 21, 2026: origin of the three pillars and the non-commodity framing. Reported by Search Engine Roundtable, Apr 23, 2026.
2. AI Optimization Guide, Google Search Central, https://developers.google.com/search/docs/fundamentals/ai-optimization-guide: origin of the debunked-tactics list, query fan-out mechanics, and agent-readability guidance.