How to Evaluate a Writing Sample for Quality, Accuracy, and Voice

Last updated: October 13, 2026

Quick Answer: A writing sample usually earns a pass when it scores at least 12 out of 15 on structure, accuracy, and voice, and does not contain a major red-flag error.

In short: when you evaluate a writing sample for quality, accuracy, and voice, judge it on three things at once — does it say something true, does it say it clearly, and does the voice fit the job you need filled? This guide to how to evaluate a writing sample for quality, accuracy, and voice keeps those three checks separate so you can make a cleaner call.

Key Takeaways

• Strong samples show control, not just talent.
• Read the structure before you get distracted by style.
• Check the claims most likely to cause harm if they are wrong.
• Voice fits when it matches the reader, channel, and brief.

A sample can be polished and still be wrong. It can be accurate and still be a slog. It can sound clever and still miss the brief. If I am choosing a freelancer, hiring a writer, or deciding whether to trust a byline, I start by asking one question: would this person help or hurt my project once the stakes are real?

What Actually Determines a Good Writing Sample?

A good writing sample proves control, not just talent. If the sample is for a technical role, then accuracy outranks flair. If it is for a brand voice role, then fit outranks generic polish. If it is for editorial work, then judgment matters as much as sentence-level craft. When you evaluate a writing sample for quality, accuracy, and voice, those priorities do not blur together.

I would score a sample on three separate tracks:

1. Quality of execution: structure, flow, sentence control, transitions, and whether the piece finishes with purpose.
2. Accuracy and evidence: facts, names, dates, terminology, and whether the claims are defensible.
3. Voice and fit: whether the tone matches the audience, format, and business need.

The trap is to treat these as identical, so if the piece will affect health, legal, or financial decisions, consult a qualified professional before relying on it. They are not. A writer can produce crisp copy that is factually weak. Another can write a careful, accurate explainer that sounds flat. Another can nail voice but drift into overconfident claims. A good sample is the one that shows the right mix for the work you actually need done.

If the sample is only 300 to 500 words, I look for compression under pressure. If it is a long-form article, I look for pacing over 1,200 words or more. Short samples hide weak structure; long samples hide weak discipline. The size of the sample changes what I inspect, not the standards I use.

Quick check: if you already know whether you need accuracy, voice, or speed most, you are ready to evaluate the sample properly.

How do I check quality without getting distracted by style?

You check quality by reading the skeleton before you admire the skin. If the piece has clear shape, fair transitions, and a beginning that earns the ending, then the writing has real control. If it wanders, repeats itself, or sounds like three drafts pasted together, then the sample is weaker than it first looks.

I use this order:

1. Read the opening and ask what promise it makes. If the first 2 paragraphs do not say what the piece will do, the writer has not taken command of the reader.
2. Look at the structure. If the piece jumps between points with no logic, note that. A strong sample should make its path obvious even before you study each sentence.
3. Check paragraph weight. If every paragraph is the same length and tone, the writing may be mechanically tidy but not alive. Real control shows up in variation.
4. Read one section out of order. If it still makes sense, the writer may be writing modularly. If not, the piece may depend too much on ornament instead of structure.
5. Inspect the ending. A weak ending often means the writer had no real argument. A good ending closes the loop or sharpens the point in the final 2 or 3 sentences.
6. Mark every place you felt tugged forward. Good writing creates momentum without shouting.

If the sample is marketing copy, then concision matters more than literary range. If it is a thought piece, then the writer needs a view, not just tidy paragraphs. If it is a newsroom sample, the deck, nut graf, and body all matter because the structure itself carries meaning. For a related framework, see content writing best practices and editorial guidelines.

A generic article usually says “look for grammar.” That is too small. Grammar is table stakes. I would rather see one slightly imperfect sentence with clear purpose than ten spotless lines that never move anywhere.

Quick check: if you can summarize the piece’s argument in one sentence after one read, the structure is probably doing its job.

How do I judge accuracy without being a fact checker for every line?

You judge accuracy by sampling the claims that would cause the most damage if they were wrong. If the piece mentions a statute, a medical fact, a financial term, a product specification, or a date, then I verify those first. If the piece is light on concrete claims, then accuracy still matters, but the main risk shifts to overstatement and vague certainty. For factual claims, sources such as the FTC’s guidance on advertising claims, NIH health pages, and AP style guidance are useful starting points.

I would check in this order:

1. Circle named facts. People, organizations, laws, measurements, publication dates, and technical terms deserve immediate scrutiny.
2. Separate claims from opinions. “This is effective” is a claim. “I prefer this approach” is an opinion. Good writing knows the difference.
3. Look for hedging that signals caution. Phrases like “appears,” “can,” or “in many cases” are not flaws when the subject is complex. They may be a sign of honesty.
4. Check whether the writer names standards or sources. References to AP style, APA 7, ISO standards, NIH pages, FTC guidance, or a peer-reviewed review are all useful signals when relevant.
5. Test the most damaging error first. If the reader believed the wrong thing, what happens? If the answer is “not much,” the claim can be read with more flexibility. If the answer is “a costly mistake,” the sample needs closer review.
6. Look for fake certainty. Absolute language without support — “always,” “never,” “proves” — is a warning sign unless the context truly allows it.

If you are hiring for health, finance, legal, or policy writing, I would be stricter. A graceful sentence does not excuse a bad claim. For those topics, accuracy is not just a quality issue; it is the job. When the subject touches regulations or professional advice, a qualified expert should review the final draft. The FTC’s Endorsement Guides and NIH’s health information resources are good examples of why sourced claims matter.

The honest trade-off is time. Deep verification takes longer than a surface skim, sometimes much longer. If you need someone who writes quickly on topics with stakes, then a clean sample is not enough — you need evidence that the writer knows how to source carefully.

Quick check: if the sample contains more than three specific factual claims and none are sourced or obviously standard, slow down and verify before you trust the voice.

Does the voice fit the job, or just sound good on its own?

The voice fits if the sample sounds right for the reader, the channel, and the brand brief you actually have. If the voice is witty but the audience wants plainspoken clarity, then the sample misses. If the voice is stiff but the job needs authority, then it misses in a different direction.

I would ask these questions:

• Would this tone work in a product page, a bylined article, a donor letter, or a founder memo?
• Does the writer know when to be direct and when to be restrained?
• Is there a consistent point of view, or does the piece wobble between formal and chatty?
• Does the writer use language the target reader would actually say?

A sample can show range, but range is only useful if the writer can control it. A good generalist may write one sample with a crisp, practical voice and another with more warmth. A poor writer often confuses variation with inconsistency. That difference matters. For more on matching tone to format, see brand voice examples and content brief template.

If the sample is for B2B SaaS, I want language that is specific, not foggy. If it is for consumer editorial, I want voice with rhythm and curiosity. If it is for executive ghostwriting, I want restraint and shape, not sparkle. One voice is not superior to another; fit is the point.

A lot of people overvalue “nice writing.” That is a mistake. Voice should serve the assignment. A playful sample can be wrong for regulated industries. A formal sample can be wrong for a brand that depends on speed and warmth. The wrong voice is not a minor mismatch; it is a sign the writer did not understand the brief.

Quick check: if you can hear the intended audience in the piece, the voice may be right; if you can only hear the writer, it probably is not.

Which sample type tells you the most?

A live sample tied to the real assignment tells you the most. If the choice is between a polished portfolio piece and a fresh sample written from your brief, I would trust the fresh one more. The portfolio piece may be good, but it may also be someone’s best case, not their repeatable level.

Here is the practical comparison:

If you are asking for a new sample, keep it fair. A 200-word test is useful for tone, but it will not show full structure. A 1,000-word test shows more, but it also asks for more time. I think the sweet spot for many assignments is 400 to 700 words, unless the role is long-form or strategic.

If you already have several samples, compare them against the same brief. Different topics make comparisons fuzzy. Same brief, same audience, same format: that is the clean test. For similar hiring decisions, a freelance writer test can help you compare samples consistently.

Quick check: if you need repeatable performance, not just a one-off showcase, prefer a sample built for your brief over a random portfolio piece.

When the standard advice is wrong

The standard advice is wrong when the sample is too short, too polished, or too far from the real job. If any of those are true, then a normal “read for clarity and grammar” approach will miss the real risk.

Here are the cases that break the usual rules:

1. Situation: the sample is a published ghostwritten piece. What changes: the voice may belong to the client, not the writer. What to do instead: judge structure, accuracy, and adaptability first, and ask for an explanation of the writer’s role.
2. Situation: the sample is highly edited by an editor. What changes: the polish may not reflect the writer’s raw output. What to do instead: ask for a draft or a second sample with lighter editing.
3. Situation: the sample is creative or branded copy. What changes: “accuracy” may mean fidelity to brand rules rather than outside facts. What to do instead: check whether the writer follows tone, claims, and constraints consistently.
4. Situation: the sample uses a highly technical domain. What changes: a non-specialist may not spot subtle errors. What to do instead: have a subject-matter reviewer check terminology and logic, especially in fields with standards such as ISO, APA, or medical guidance.
5. Situation: the sample is tiny, under 200 words. What changes: there is not enough room to judge long-range structure. What to do instead: focus on precision, sentence economy, and whether the writer can control a single message.
6. Situation: the sample is intentionally provocative or voice-first. What changes: strong opinions may be part of the brief. What to do instead: separate style choices from unsupported claims and ask whether the argument would still hold up without the punchy framing.

If you only apply one set of criteria everywhere, you will reject some good writers and accept some weak ones. That is why context matters more than a fixed checklist. The wrong test is worse than no test.

Quick check: if the piece’s format or editing history is unusual, standard evaluation rules need to bend.

A practical scoring method I would use

A 15-point scorecard is enough for most hiring decisions, and it keeps me honest when a sample has one dazzling trait but obvious flaws elsewhere. I would score each category from 1 to 5:

• Structure and clarity: 1 to 5
• Accuracy and caution: 1 to 5
• Voice and audience fit: 1 to 5

Then I would interpret the result this way:

1. 12 to 15 points: strong sample, worth serious consideration.
2. 9 to 11 points: usable but uneven; good only if the weak area is not central to the job.
3. 8 points or below: not a fit unless the role is very narrow and the flaw is minor.

I would also flag any “red lines” that override the score. One serious factual error in a regulated topic can outweigh elegant prose. One voice mismatch can kill a branding role. One vague, meandering structure can sink a content strategy job even if the sentences are lovely.

The score is not the whole decision. It is a way to keep one impressive paragraph from overpowering the rest of the piece.

Quick check: if you need a simple decision rule, use a 15-point scorecard and reserve the right to reject any sample with a major red-line problem.

FAQ

How many writing samples should I review before deciding

Drafted with AI; not yet reviewed by a person.