Sourceworthiness & Proof Audit
Find unsupported claims, weak generalities, and places where specifics would make your page worth citing
What This Prompt Does
AI systems preferentially cite pages with concrete, verifiable information. This prompt flags every unsupported claim, every generic phrase, and every spot where a number, example, or credential would make the page more trustworthy. You leave with a prioritized list of credibility upgrades.
When to Use It
- •A page sounds polished but AI systems never reference it
- •A competitor's page (with real numbers) gets cited while yours doesn't
- •Preparing cornerstone content or a press-friendly page
- •Training a writer on the difference between marketing and proof
The Prompt Template
Act as a credibility and content quality analyst. Your goal is to identify where this page lacks proof, evidence, or trust-building detail that would make it more usable in AI-generated answers. Context: - Page text: [PASTE PAGE CONTENT] - Business type: [BUSINESS TYPE] - Topic: [TOPIC] Task: Identify: 1. unsupported claims 2. weak or generic statements 3. places where examples, numbers, or credentials would help 4. the top 5 ways to strengthen credibility Process: 1. Review the page line by line. 2. Find claims that are broad or unsupported. 3. Identify where specifics would make the page more trustworthy. 4. Recommend the top improvements. Constraints: - Use simple language. - Do not suggest fake statistics or invented proof. - Be specific. - Keep the answer under 400 words. - Favor clearer facts, direct answers, and useful specifics over polished marketing language. Output format: Use these headings: - Unsupported Claims - Weak Generic Language - Better Proof to Add - Top 5 Credibility Fixes
How the Prompt Is Structured
"Do Not Suggest Fake Statistics"
Without this guardrail, AI will hallucinate convincing-sounding numbers. The constraint keeps every recommendation in "what you could add if it's true" territory.
Separate "Unsupported" from "Generic"
Unsupported claims ("fastest in the state") are factually unmoored. Generic language ("excellent service") is factually empty. Each needs a different fix.
Numbers, Examples, Credentials
The three categories of proof. Each one trades abstract promises for concrete evidence — and AI systems reliably prefer the concrete.
Line-by-Line Review
The Process phrase "line by line" forces granular analysis. Holistic reviews miss the individual filler sentences that add up to a weak page.
Example Output
Unsupported Claims
- • "Fastest response times in the area" — no proof offered.
- • "Industry-leading technicians" — no credentials or references cited.
Weak Generic Language
- • "Excellent customer service"
- • "We go the extra mile"
- • "Our commitment to quality"
Better Proof to Add
Average response time in minutes. Years in business. Number of customers served. License number and state. A named trade-association affiliation. 3–5 real customer reviews with first names and neighborhoods.
Top 5 Credibility Fixes
- Replace "fastest response" with "average 45-minute arrival in central Bozeman."
- Add license number and insurance carrier name.
- Add "Serving Gallatin County since 2014" with customer count.
- Add 3 real reviews with names, neighborhoods, and service dates.
- Replace "industry-leading" with specific certifications held by staff.
Tips for Better Results
Only Add What You Can Prove
Every replacement must be verifiably true. If you can't support "45-minute response," measure it first and then publish.
Keep the Proof Visible
Numbers buried in a PDF don't help. Put them in the page's visible copy so AI systems can read them.
Pair With Structured Data
Once you have real reviews, licensing, and business facts in the copy, add schema markup so machines can parse them unambiguously.
Update Numbers Quarterly
"Since 2014" is fine forever. But "500+ customers" should be updated as it grows — fresh numbers look alive.
Swap Marketing for Proof
We help teams replace generic marketing language with the specifics AI systems (and customers) find credible.