agentleFS
keuperakeupera/seo-skills/skills/best-of-listicle-pages/SKILL.md

best-of-listicle-pages

Skill0 starsChanged 2 months agoMITView on GitHub

Write roundups, listicles and product reviews that survive Google's reviews system and get cited by AI assistants, using real testing rather than repackaged spec sheets. Use whenever the user mentions listicles, roundups, "best of" posts, top 10 lists, product reviews, review content, affiliate content, buyer's guides, or gift guides. Also use when a review or roundup site has lost traffic after a core or reviews update.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 6 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every one of those is impossible to fake convincingly and trivial to demonstrate if you actually used the thing.

    line 23
  2. Distinctive:

    Disclose ownership: If the publisher owns one of the products listed, say so where it's listed.

    line 82
  3. Distinctive:

    Every product is a winner: No evaluative judgement, so no reason to trust it.

    line 107
  4. Distinctive:

    Don't pad to ten: when there are four good options.

    line 116
  5. Distinctive:

    Don't put "2026" in the URL: Put it in the title, update it annually, keep the URL stable.

    line 118
  6. Distinctive:

    Don't copy the manufacturer's description: Every competitor has it.

    line 119

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Roundups, Listicles and Reviews

Google's reviews system specifically targets the thing most roundups are: a summary of what's already on the internet, assembled by someone who never touched the products. Sites built on that model have been losing traffic in waves since 2021, and it isn't reversing.

What survives is evidence of first-hand experience. That's expensive, which is exactly why it works.

The bar Google set

Google's own guidance for review content asks for evaluative content that demonstrates:

  • Expert knowledge of the product, shown rather than claimed
  • What the product is like physically, or how it performs in use
  • Unique information beyond the manufacturer's, and quantitative measurement where possible
  • How it compares to competitors and to previous versions
  • Decision-relevant differentiators and who each option is actually for
  • Trade-offs, including reasons not to buy

Every one of those is impossible to fake convincingly and trivial to demonstrate if you actually used the thing. That's the design.

The structure

H1        Best [category] for [specific use case] in 2026

          [Verdict block, 50-60 words: the pick, the runner-up,
           the budget option. Named, with one reason each.
           Someone who reads only this leaves with an answer.]

          [Comparison table: 5-8 options, the specs that decide it,
           price, and the one-line "best for"]

H2        How we tested
          [Real methodology: what was bought, how long it was used,
           what was measured, what was excluded and why, who did it.
           This section is the credibility of the whole page.]

H2        1. [Product] — best overall
          [Photo you took. Not the press image.]
          [What it's like to use. Specifics, measurements, quirks.]
          [Who it's for. Who it isn't for.]
          [What went wrong or annoyed you.]
          Specs table | Price | Where to buy

H2        2. [Product] — best for [specific case]
          ...

H2        What to look for when buying
          [The buying guide: which specs matter, which are marketing]

H2        How we chose / what we excluded
H2        FAQ
          [The specific pre-purchase questions, answered directly]

          [Disclosure, plainly stated, near the top and again here]

The parts that decide whether it works

Test the products. There's no substitute and no shortcut. If testing everything isn't possible, test some and say clearly which you tested and which you assessed from specs and other reviews. That honesty is worth more than a fabricated uniform treatment.

Take your own photographs. The single strongest signal of first-hand experience, and the one aggregators can't produce. A phone photo of the actual product on a real desk beats a studio press image.

Measure something. Battery life you timed. Decibels you recorded. Load times you benchmarked. Weight on your own scale. A number nobody else has is why anyone would link to or cite the page.

Say what's wrong with each option. A roundup where every product is great is a roundup nobody trusts. The negatives are what make the positives readable.

Rank, and commit. "Best overall" must be one product. A list of ten options with no ranking is a page that refused to answer the question it asked.

Name who each is for. "Best for people with small hands", "best if you already own the ecosystem", "best under 100 euro". This is what gets quoted by assistants, because it maps onto a user's stated constraint.

Update honestly. New versions, discontinued products, changed prices. And when you update, change the content, not just the date.

Disclosure

  • Affiliate relationships must be disclosed, clearly, before the first affiliate link. In the US this is an FTC requirement, and equivalents exist elsewhere. A line in the footer isn't disclosure.
  • Disclose gifted or loaned review units.
  • Disclose ownership. If the publisher owns one of the products listed, say so where it's listed.
  • Use rel="sponsored" on affiliate links and rel="nofollow" where appropriate. Undisclosed affiliate links passing PageRank is a link scheme.

Disclosure doesn't hurt conversion the way people fear. Getting caught undisclosed does.

Why AI assistants cite these pages

A model answering "what's the best X for Y" needs something that has already done the comparing. What it can use:

  • An explicit ranked recommendation rather than a neutral list
  • "Best for [specific case]" labels that match a constrained prompt
  • Measured numbers attributable to a source
  • Tables it can parse cleanly
  • Stated methodology, which is a trust signal for a model the same way it is for a reader
  • Balanced assessment, weighted above pages that only praise

And the constraint: it has to be in the HTML. Product data loaded client-side, or a table that renders after hydration, doesn't exist for most retrieval agents. See featured-snippet-passages.

Diagnosing a roundup site that lost traffic

Almost always one of these, in order of likelihood:

  1. No first-hand experience. Content assembled from other reviews. This is the reviews system working as designed, and the fix is expensive: test things, or narrow to a category you can genuinely cover.
  2. Scaled thin content. Hundreds of near-identical roundups across unrelated categories. Prune hard.
  3. Stale. Products discontinued, prices wrong, "2024" in the title. Update or remove.
  4. Every product is a winner. No evaluative judgement, so no reason to trust it.
  5. Undisclosed affiliate relationships across the site.
  6. Site-level quality. Core updates assess the site, so 400 weak roundups drag down the 20 good ones. Pruning is usually a bigger lever than improving.

Recovery from a reviews or core update is measured in months and comes from the site being genuinely better, not from a tweak. Say that plainly rather than promising a fix by next month.

Don't do these

  • Don't generate roundups from spec sheets. It's the exact pattern being demoted.
  • Don't pad to ten when there are four good options. A four-item list that's right beats a ten-item list with six fillers.
  • Don't rank by commission. It's transparent to readers over time and it's the reason the whole category lost trust.
  • Don't put "2026" in the URL. Put it in the title, update it annually, keep the URL stable.
  • Don't copy the manufacturer's description. Every competitor has it.
  • Don't publish a roundup for a category you can't get access to. Better to write nothing than to fake it.

Related skills

comparison-alternatives-pages, eeat-author-pages, featured-snippet-passages, schema-markup, content-refresh, image-seo

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.