Evidence and Citations

This skill adds the evidence that makes answer engines cite you. Controlled research on generative engines is unusually clear here: adding sources, statistics, and attributed quotes lifts citation likelihood substantially, while padding and unsupported confidence do nothing or backfire (see ../../references/citation-ranking-factors.md). Engines act as consensus machines — they prefer claims they can verify. This skill makes your claims verifiable. It works for any business, B2B or B2C.

Structure is answer-engineering; this skill is the substance that goes inside the structure. Use them together on every page meant to be cited.

When to use this skill

• Strengthening content so AI will quote it.
• Replacing vague claims with specific, sourced facts.
• Adding expert quotes, original data, or inline citations.
• Auditing a page for "evidence density."

Before you start

Read aeo-foundation.md for the brand's real proof assets — original data, named experts, research, results — and any claim or compliance constraints. If it is missing, point to the aeo-foundation skill.

The four evidence levers

1. Statistics. Replace "fast" and "most" with specific, sourced numbers ("reduces setup from 2 hours to 4 minutes"; "in a 2026 survey of 1,200 teams, 63%…"). Specific, defensible figures are among the strongest citation drivers.
2. Original data. First-party research, benchmarks, and data tables are disproportionately cited because they are the only source for that fact — engines must point to you. Where a brand has data, surface it as the centerpiece.
3. Attributed quotes. Quote named experts with their credentials. Attribution signals authority and gives the engine something concrete to cite.
4. Inline citations. Link claims to authoritative external sources (research, primary data, standards bodies, reputable publications). Citing sources reliably lifts your own citation odds, and helps most for sites that are not already top authorities.

Workflow

1. Find the claims. Scan the content for every assertion an reader could doubt — every "fast," "best," "most," "leading," "easy."
2. Upgrade each claim. Attach a statistic, a data point, a quote, or a citation. If a claim cannot be supported, soften or cut it — unsupported confidence is penalized.
3. Lead with your own data. Where the brand has original numbers, build a passage (or a table) around them; this is the most ownable citation magnet.
4. Set evidence density. Aim for a concrete, sourced point roughly every few paragraphs, not one buried fact per article. Avoid the opposite failure — stat-stuffing that reads as noise.
5. Source quality. Prefer primary sources, original research, standards bodies, and reputable publications over thin secondary blogs. Keep links live and accurate.
6. Keep it honest. Accuracy is the moat: a misrepresented stat or a broken citation erodes trust with both readers and engines.

Output format

• The content with claims upgraded to evidence: statistics, data, quotes, and inline source links in place.
• A short list of the claims that need a source the brand must supply (e.g. "needs an internal figure for setup time").
• Suggested original-data opportunities the brand could publish to own citations.
• (assumption) tags on any numbers or quotes used as placeholders.

Common pitfalls

• Vague modifiers left unsupported ("blazing fast," "industry-leading").
• Confident claims with no evidence — actively deprioritized by engines.
• Stat-stuffing that overwhelms the point.
• Citing weak secondary sources, or letting citations rot into dead links.
• Misstating a figure for effect — the fastest way to lose trust.

See before/after examples for weak-vs-evidence-backed rewrites.

Related Skills

• answer-engineering — the structure these evidence-rich passages live in.
• author-authority — the named, credentialed voices behind the quotes and claims.
• structured-data — marks up authors and articles to reinforce credibility.
• digital-pr — turns original data into earned coverage and corroboration.