Only accepted work is recordedAboutGuidesContactSign in
CitePepCitePep
Guides/Contributing/How to respond to a review request
Contributing4 min read · Beginner · Updated Sep 2026

How to respond to a review request

A strong response gets accepted, credited, and recorded. Here’s how to write one that lands.

What a response has to contain to be accepted, and how to pick the requests worth answering.

One thing to know about the bar before you start: of 1,758 contributions recorded on CitePep, 1,753 were accepted. That is not a low bar, it is a bar that sits at submission. What gets filtered is vague, unsourced work, and it gets filtered before a publisher ever sees it, because the response format asks for the specific text and the source.

Choose the requests where you’ll shine

Pick by topic, because your record is built per topic rather than in total. On CitePep the per-topic scores run from 1,140 in one subject down to single digits in others, and a contribution lands in exactly one of those buckets.

The practical effect: five accepted items in one subject outrank twenty scattered across twenty. One contributor here holds 162 topics and another holds 1, and the second is the stronger reviewer in their subject.

Take the requests where

  • It sits in a subject where you already have accepted work, or want to
  • You can verify the claims rather than assess whether they sound right
  • You have time for the full check, including the secondary locations
  • The per-topic score it feeds is one you are deliberately building

Do the work thoroughly

Skimming is visible in the output. A response that names paragraph-level problems without naming a sentence reads as a page that was scanned, and it gives a publisher nothing to apply.

What depth looks like in the record: of 1,249 categorised corrections here, 918 were a single clean error and 331 carried more than one in the same sentence. The second kind only surfaces when someone reads the sentence rather than the section.

Be specific, and bring sources

Quote the exact text, give the replacement wording, attach the source. The difference matters more than it sounds: a flagged problem goes back into someone else's research queue, while a proposed fix can be accepted as it stands.

A real example from the log, in the form that got accepted: an article placed Australia's CFC rules in Division 9 of Part III of the Income Tax Assessment Act 1936, and the correction named Part X as the right location. Wrong text identified, right text supplied, statute cited.

  1. Read the whole page, including its furniture

    The bibliography, FAQ and callouts carry their own copies of claims. 595 of the 1,753 accepted corrections here found a stale copy outside the body.

  2. Quote the text, supply the replacement

    Name the sentence and give the wording that should stand in its place. See how to write a useful correction.

  3. Attach the source, and check it carries the claim

    529 of 1,249 corrections were a citation resolving to something that said less than its sentence. Do not add a sixth.

i

An accepted response becomes a public record carrying the error category, your reason, and the text before and after. Decisions here have run fast, averaging under an hour from submission across 1,757 decided contributions, with the slowest at 79 hours.

Use each response to build your record

Aim responses where the record compounds. The score is per topic, so repeated work in one subject concentrates into something a publisher can look up, while the same effort spread across twenty subjects reads as twenty thin entries.

Two of the 8 contributors with accepted work here hold 1 and 2 accepted items. That is where every record starts, and the next response is what moves it.

Quick checklist

  • The request is in a subject where your score already sits, or should
  • You read the body and the bibliography, FAQ and callouts
  • Every note quotes the sentence it concerns
  • Each problem comes with the replacement wording
  • Each source was opened and confirmed to carry the claim
  • The topic is one you are deliberately building

Common questions

Where should I focus my responses?

In the subject where your score already sits. The record is per topic, so depth in one concentrates into something checkable while breadth reads as thin entries.

How do I improve a response that wasn’t accepted?

Check the three things first: did you quote the sentence, supply the replacement wording, and open the source to confirm it carries the claim. Most non-acceptances are missing one of those rather than wrong on the facts.

How quickly should I respond to a request?

Decisions here average under an hour once submitted, so the delay is yours rather than the publisher's. A careful response a day later beats a shallow one in an hour, because only one of them becomes a record.

Should I take requests slightly outside my expertise to build volume?

No, because volume does not aggregate the way people expect. The score is per topic: 20 items across 20 subjects leaves 20 scores of one, which is not the same as one score of 20.

What makes a response stand out to a publisher?

A response they can apply without doing research themselves: the sentence quoted, the replacement supplied, the source attached and verified. Everything else is preference.

Was this guide useful?