Five tests let you evaluate content quality without becoming an expert yourself. Each test targets a different failure mode. Run all five before approving any content.
Test 1: The Paragraph Shuffle Test
What it catches: Content that looks substantial but says nothing. Thin content reads smoothly because it makes no specific claims that require logical ordering.
How to run it: Read paragraphs 3, 5, and 7 in isolation. Then mentally rearrange them. If the content makes equal sense in any order, it lacks logical dependency between sections.
Pass criteria: Rearranging paragraphs would break comprehension. Paragraph B requires information from paragraph A. Paragraph C builds on B.
Fail example: “SEO is important for visibility. Content should be valuable. Keywords help search engines understand your topic.” These three sentences work in any order because none builds on another.
If you can shuffle paragraphs without anyone noticing, the writer likely organized around keywords rather than logical progression.
Test 2: The Source Specificity Test
What it catches: Unsupported authority claims. Anyone can gesture at expertise. The question is whether those gestures point to anything verifiable.
How to run it: Find every claim that invokes external authority. Check whether the content names specific researchers, institutions, publication dates, or primary sources.
Pass criteria: Claims include names, dates, and links to primary sources. “Dr. Sarah Chen’s 2023 study in the Journal of Marketing Research found…” passes.
Fail example: “Studies show that most businesses benefit from SEO.” No researcher named. No publication cited. No methodology described. The claim cannot be verified or evaluated.
Unverifiable attribution functions as decoration, not evidence.
Test 3: The Falsifiability Test
What it catches: Claims too vague to be wrong. Low-quality content often avoids specific statements because specific statements can be contradicted. Broad qualifiers absorb all counterexamples.
How to run it: Identify the three main claims. For each, ask: could this be wrong? Is there a scenario where this claim fails? If no scenario exists, the claim carries no information.
Pass criteria: Claims are specific enough to be testable. “Email marketing outperforms social media for B2B SaaS with sales cycles over 90 days” can be verified or falsified.
Fail example: “Email marketing can be effective for many businesses.” This cannot be wrong because it claims nothing. “Can be” and “many” prevent any counterexample.
Claims that cannot be wrong usually carry no useful information.
Test 4: The “So What” Test
What it catches: Disconnected facts. Some content provides accurate information that fails to help the reader do anything. Facts without application are trivia.
How to run it: After each major claim, ask “so what?” If the content does not answer that question within two sentences, the writer has not connected information to reader impact.
Pass criteria: Every fact connects to reader action or decision. “Mobile traffic exceeds desktop for most sites” is followed by “which means your testing priority should shift to mobile-first.”
Fail example: “Google processes 8.5 billion searches per day. Search engines use complex algorithms. SEO has evolved over time.” Three true statements. Zero guidance on what to do with them.
Information without implication serves the writer’s word count, not the reader’s needs.
Test 5: The “Compared to What” Test
What it catches: Meaningless comparatives. “Effective,” “fast,” “affordable,” and “better” carry no information without a comparison baseline.
How to run it: Find every comparative adjective. For each, identify what baseline is being compared. If no baseline is stated or implied clearly, the comparison cannot be evaluated.
Pass criteria: Comparisons state their baseline. “This approach reduces load time by 40% compared to unoptimized images” specifies what “faster” means.
Fail example: “This tool provides fast results and excellent accuracy.” Fast compared to what? Excellent compared to which alternatives? The reader cannot evaluate these claims.
A comparison without a baseline is an adjective doing the work of evidence.
Pattern Recognition: When Tests Fail Together
Tests 3 and 5 often fail together. Content that avoids specific claims (Test 3) also avoids specific comparisons (Test 5). Both patterns indicate the same underlying issue: insufficient precision to be useful. When everything is qualified and nothing is compared, the content provides no basis for decision-making.
Tests 1 and 4 reveal a different problem. Content that fails the shuffle test usually fails “so what” as well. Both failures stem from the same structural gap: information has been collected without establishing relationships between pieces. The content knows facts but not how facts connect to each other or to reader action.
When you see these paired failures, revision is unlikely to help. The content needs rewriting by someone who can establish logical dependencies and reader implications from the start.
Running the Full Battery: A Concrete Example
Consider an article titled “Best CRM Software for Small Business.” You run all five tests:
Test 1 (Shuffle): The section on Salesforce could swap places with the section on HubSpot without breaking anything. Neither section references the other or builds on shared criteria. Fail.
Test 2 (Source): The article claims “studies show CRM increases revenue by 29%” but names no study, no researcher, no publication date. The number cannot be verified. Fail.
Test 3 (Falsifiability): Every CRM is described as “great for growing businesses.” No specific scenarios where one CRM works better than another. Fail.
Test 4 (So What): The article lists features but never explains which features matter for which business types or situations. Fail.
Test 5 (Compared to What): HubSpot is called “affordable” with no price comparison to alternatives or definition of the comparison baseline. Fail.
Five failures. This content should be rejected entirely rather than revised. The structural problems are too fundamental for editing to fix.
What These Tests Cannot Do
These tests filter low-quality content. They do not certify accuracy. Content can pass all five tests while still containing factual errors that only domain experts would catch. Two genuinely expert articles that disagree with each other could both pass these tests.
Use these tests as triage, not certification. They identify content to reject. They do not guarantee that passing content is trustworthy. For accuracy verification, you need domain expertise or access to someone who has it.
These tests raise your floor. They do not raise your ceiling.