Editorial and Quality Support Service
Pubrica’s editorial and quality support service provides independent editing and quality control of documents written by you, your team or another supplier, for researchers, clinicians, universities, publishers and life-science companies. Our subject-specialist editors work at an editing level we name before we quote (proofreading, copyediting, line editing, substantive or developmental editing) and check each document against the reporting guideline its design requires and the ethical standards of the target journal. You receive the edited document with tracked changes, a style sheet, a query log and a reference log. Because editing is sold under inconsistent names, we confirm which level you need, including when a less expensive level is sufficient.
Proofreading$0.025
Copyediting$0.035
Line editing$0.050
Substantive editing$0.075
Developmental editing$0.100
Read it → name the level → quote → edit → second reader
- The level named before the quote
- A style sheet built and handed over
- References checked against the source
- Read by a second person
- Our role acknowledged in your paper
Service Scope
Eight Editorial and Quality Support Services, Available Individually
Most documents need two or three of these services. Each can be ordered individually, and the editing level chooser on this page prices a level rather than a fixed package.
Who We Serve
Researchers and Organisations We Support with Editorial and Quality Control
We support six groups of clients, each of whom needs a document written elsewhere to be checked and corrected.
Service Limitations
We do not guarantee journal acceptance. We do not offer similarity reduction: we review the report, explain what it means and recommend a citation where one is needed rather than rewording text to lower the score. We do not offer AI-detection clearance, because no detector is reliable enough, and published false-positive rates are highest for authors writing in an additional language. We do not issue editing certificates; instead, we confirm in writing exactly what we did, which your institution may require. We are not peer reviewers: we check that a document reports what it should and claims only what it can support, not whether the science is correct. We do not edit assessed work where an institution’s regulations prohibit assistance.
Compliance and Standards
Six Editorial Frameworks and Their Current Versions
The frameworks below govern editorial work, with their current versions. Three of them have been revised in the last two years.
ICMJE Recommendations (updated January 2026)
The four authorship criteria are unchanged and all four must be met by every author. The January 2026 update added a new Section V on the use of artificial intelligence in publishing, split between authors, reviewers and editors, and a new Section III.L.2 requiring that all authors be able to review the data supporting the results. It is also the document that governs our own position: “writing assistance, technical editing, language editing, and proofreading” do not confer authorship. We are acknowledged, never listed. One practical note — the PDF carries the January 2026 text while icmje.org’s HTML pages still showed January 2024 when we checked, so cite the PDF.
COPE (and the two positions that matter here)
Two COPE documents govern most of what this service does. On similarity: COPE “does not recommend using a specific percentage of duplication or overlap as a benchmark for conducting an investigation” — so there is no threshold to hit, whatever an editing service tells you. On reusing your own earlier writing: the COPE position on handling text recycling, version 1 published 20 August 2024, defines it as reusing your own previous writing without being transparent or citing the original, and weighs the extent, the location in the paper and whether it was disclosed. Note that COPE’s older text-recycling document is authored by BioMed Central and carried as endorsed legacy guidance from 2016, not as COPE’s own position.
GPP 2022 (company-sponsored research)
Good Publication Practice for company-sponsored biomedical research, published in the Annals of Internal Medicine on 30 August 2022 and stewarded by ISMPP. It is the third update of the 2003 original, after GPP2 in 2009 and GPP3 in 2015, and it applies alongside the ICMJE criteria where a sponsor is involved. GPP3 is superseded and there is no GPP4 — that was the working title during development, dropped when the guideline moved to year-based naming. You will see “GPP3/GPP4” listed as a standard. Do not confuse GPP 2022 with the GPCAP 2026 update published this year: that covers conference abstracts and presentations and is a different guideline.
Reporting guidelines (the version, not just the acronym)
CONSORT 2025 for trials and SPIRIT 2025 for protocols, both published April 2025 and both stating that their predecessors should no longer be used. PRISMA 2020 for systematic reviews, which despite its name appeared in 2021. STROBE (2007) for observational studies, still current. STARD 2015, TRIPOD+AI (2024), ARRIVE 2.0, CHEERS 2022, COREQ and SRQR, and CARE for case reports. Our quality check locates every checklist item in the text by page and line, or records it as missing.
The editing standards (three, and they disagree)
There is no single industry taxonomy of editing levels, so clients are easily sold the wrong level. Editors Canada’s Professional Editorial Standards 2024 — note, not the 2016 edition — defines structural, stylistic, copy and proofreading standards over a foundational part. The Chicago Manual of Style splits manuscript editing into mechanical and substantive at 2.49 and 2.50, and treats line editing as a separate earlier pass at 2.53. The Editorial Freelancers Association names developmental, line, copyediting and proofreading, and defines proofreading comparatively — checking one stage against the previous — which is narrower than how the word is usually sold. We name which we mean, in all three vocabularies, before quoting.
Similarity and AI screening (and their documented limits)
Turnitin’s own guidance states that “plagiarism cannot be determined by the similarity score alone” and that there is no single acceptable or perfect score. On AI detection, COPE’s September 2025 guidance says detection tools “should not be relied upon in isolation” and that none is fully accurate; Turnitin now publishes no headline accuracy figure, suppresses results below 20%, and states that a score “should not be used as the sole basis for adverse actions”. The evidence on bias is the part that matters to this service’s clients: seven detectors tested on human-written TOEFL essays produced an average 61.3% false-positive rate, while essays by native-English school students were classified almost perfectly. We run these checks and interpret the output, but we do not sell a score.
Responsible Use of AI
Responsible Use of AI, with Human Editors and Traceable Changes
Journal editors, reviewers and regulators trace decisions back to evidence and to named individuals. For this reason, source traceability takes priority over speed whenever we use AI tools.
Why We Do Not Publish AI Accuracy Figures
Accuracy figures for AI tools are usually reported by the companies that sell them.
Pubrica therefore does not quote its own accuracy figure, which would carry the same conflict of interest. Instead, we tell you which tool was used on your project, at which stage, and what a specialist changed afterwards.
Methodology
Editorial and Quality Support Process: Six Stages
The appropriate editing level can only be identified by reading part of the document, so each project begins with a free sample edit.
Free Sample Edit Before Agreement
Send the document and we edit 1,000 words of it for free. The sample shows the level of intervention, how we treat your voice, and which points we query rather than change. It also shows us whether the level you requested is the level the document needs, and if it needs less, we quote less.
Editing Level and Exclusions Confirmed in Writing
A named editing level. Proofreading, copyediting, line editing, substantive or developmental — named in the three published vocabularies so you can compare our quote against anybody else’s, with what that level does not cover stated explicitly, so that our quote can be compared with others.
Style Sheet Preparation
Every terminology decision, spelling variant, abbreviation and formatting convention recorded as it is made. The style sheet keeps a long document internally consistent, allows a second editor to work on a chapter without contradicting the first, and is handed over to you at the end for use on later papers from the same project.
Editing with Tracked Changes and Author Queries
Tracked changes, and comments where a change would alter meaning. We do not silently rewrite a claim. If a sentence states more than the results support, we query it, show you the estimate it rests on and leave the decision to you.
Separate Quality Control Checks
We open and read references and check them for retraction and indexing, map the reporting checklist, review declarations and read the similarity report rather than relying on the score. These are run as separate passes because each finds different problems.
Second Reader Review and Editorial Report
A second editor who did not do the editing reads the document. You receive the edited document, a clean version, the style sheet, the query log with our reasoning, the reference log and the checklist, with a short note on what was outside the scope of the work.
Request a Free 1,000-Word Sample Edit
Within two business days and at no cost, you receive a sample edit, the editing level the document needs, the price, and a note on any issues that editing will not fix. If the document needs less than you requested, the note will say so.
A PDF affects the price, because editing it requires either working in comments, which is slower and harder to review, or converting it and restoring the layout by hand. Sending the editable original is the simplest way to reduce the cost.
Pricing
Choosing the Right Level of Editing
Enter the document type, its problems and its length, and the chooser names the editing level, prices it and shows what that level does not include. It also gives the level’s name in the three published taxonomies of the Editorial Freelancers Association, the Chicago Manual of Style and Editors Canada, which use different terms. If you select the similarity or AI options, it explains what those checks can and cannot establish.
01
The level you need, what it excludes, what it costs, and what a similarity score actually means.
Runs in your browser · nothing sent anywhereThe same level, under the three published taxonomies
Why the taxonomies differ. There is no industry-standard taxonomy of editing levels, which is why two honest quotes for “editing” can differ fourfold. The EFA names four services; Chicago splits manuscript editing into mechanical and substantive and treats line editing as a separate earlier pass; Editors Canada’s Professional Editorial Standards 2024 defines four levels under different names, over a foundational part that is not a level. Before comparing prices, ask each supplier which of the five levels above they will provide.
Sources, and a correction. Level definitions from the Editorial Freelancers Association’s editorial services definitions and 2025 rate survey, the Chicago Manual of Style 2.48–2.50 and 2.53, and Editors Canada’s Professional Editorial Standards 2024 — note that the 2016 edition is superseded. Similarity: Turnitin and iThenticate guidance pages; COPE, “Determining acceptable levels of plagiarism/duplication”; Elsevier PERK; Springer Nature’s Crossref Similarity Check guide for editors. Text recycling: COPE position, 20 August 2024. AI detection: Liang W, Yuksekgonul M, Mao Y, Wu E, Zou J, “GPT detectors are biased against non-native English writers”, Patterns 2023;4(7):100779; Turnitin’s own AI writing detection guidance; COPE, “Emerging AI dilemmas in scholarly publishing”, updated 17 September 2025. Language: Amano T et al., PLOS Biology 2023;21(7):e3002184. All checked 21 September 2026. The correction: “Good Publication Practice (GPP3/GPP4)” gets listed among the standards as though both editions were live. GPP 2022 is the current guideline — published in the Annals of Internal Medicine on 30 August 2022, stewarded by ISMPP, and the third update of the 2003 original after GPP2 in 2009 and GPP3 in 2015. There is no GPP4: that was the working title during development, and the guideline was renamed at publication. GPP3 is superseded.
Our Experts
Our Editorial Team
A named team of subject specialists stays with your project from scoping to delivery. Your project coordinator introduces each team member by name before work begins.
Writes the document against the governing template and regulatory guidance, keeps every statement traceable to its source data, and maintains the version history and comment log through each review round.
Reviews the clinical and scientific content for accuracy and confirms that each conclusion is supported by the data. Each expert is matched to your therapeutic or device area.
Checks that statistical methods, tables and conclusions are consistent with the analysis plan, and reviews how the data are presented before the document reaches regulatory review.
Performs an independent check of the final document against source data, the template and the applicable regulatory checklist. Findings are logged and resolved before delivery.
Sample Work
Sample Editorial Quality Control Report
This sample report concerns a manuscript that had already been through another provider’s highest editing tier, sent to us because the authors could not tell what had been done to it. It shows whether the level paid for matched the level received. Use the tabs, then open the full report.
Editorial QC report
Cohort study · 5,900 words · already edited elsewhere
Level delivered · language · checks · declarations
Level charged: premiumLevel delivered: copyediting
310 further changes at line level1,102 earlier corrections kept
- References opened and read
- 52. The previous provider had restyled all 52 and opened none — which is visible, because restyling propagates a wrong page range and reading does not.
- Do not support the statement
- 3. Two inherited from another paper’s reference list; one supports the opposite direction of effect in the population at issue.
- Retracted
- 1, retracted eleven months before this manuscript was drafted and cited in the introduction as current evidence. Nothing in a reference manager, a copyedit or a similarity report catches this. It has to be checked deliberately.
- Similarity report
- Delivered by the previous provider as a number and a colour. We read all 34 matches: 26 were correctly quoted and cited methods text, 5 were the authors’ own earlier methods section and needed citing as text recycling, 2 needed a citation added, 1 was a real attribution problem in the introduction. The number itself told the authors nothing, which is what Turnitin says it is for.
- Reporting checklist
- STROBE, not previously completed. Twenty-two items located by page and line; four incomplete, all four answerable from what the team already held.
- Numbers against tables
- Every figure in the text checked against its table. Two mismatches, both in the abstract, both introduced during an earlier revision rather than by the editor.
- Authorship
- Five listed. One met only the first ICMJE criterion — data collection, no drafting, no approval recorded, no accountability. Moved to the acknowledgements after a conversation with the corresponding author. That conversation is much worse after publication.
- Contributions
- No CRediT statement. Drafted with the authors across the fourteen roles, which also settled a disagreement about author order that had been unresolved for two months.
- The previous editor
- Not acknowledged anywhere. Elsevier requires that anyone participating in substantive aspects such as language editing be recognised in the acknowledgements; Springer Nature and BMC require that and the funding source of the assistance. Added, naming both providers, including us.
- AI use
- One author had used a language model to polish two paragraphs and had not mentioned it to anyone, including their co-authors. Declared, in the form the target publisher requires — which is not the same form at all four of the largest publishers.
- Data availability
- The statement offered more than the ethics approval permitted. Redrafted against the approval. An overpromised data statement is a problem you meet on the day somebody asks for the data.
- Ethics and consent
- Present, complete, correctly referenced. Recorded as passing, because a report that only lists faults is not a report.
Redacted extract · field, exposure and every figure changedEditorial QC report, as delivered
One manuscript, already edited, checked against what was actually bought. Use the tabs to see each part.
The level tab shows the most important finding. The previous provider did competent work, and we kept 94% of their changes. However, they did not deliver the level that was sold, and because the quote named no level, the authors could not have known.
The checks tab shows a retracted reference. A reference retracted eleven months earlier was still cited as current evidence. Copyediting, similarity reports and reference managers do not detect this; it is found only when every reference is opened in a separate pass.
If you request sample work, we prepare a QC report based on a document like yours, of the same type, discipline and reporting guideline.
Editorial and quality support · sample QC report
Editorial QC report — observational cohort study, previously edited elsewhere
An illustrative, anonymised report showing the checks, the categories of finding and the reporting format we deliver. Ask us and we will build one around a document like yours.
1Scope of the Check and Reason for the Request
A 5,900-word observational cohort manuscript, returned from another provider’s highest editing tier and brought to us because the authors could not tell what had been done. An edited document does not show which of the five editing levels it received.
| Element | How it was handled |
|---|---|
| Documents | The authors’ original, the returned edited version, the tracked-changes file and the previous provider’s invoice and scope description. |
| Level audit | Every change classified by the level of intervention it represents, so the delivered level could be stated rather than guessed. |
| Reviewer | An editor in the discipline, plus a second reader who did not do the editing. |
| Language | A full line edit over the previous copyedit, keeping every earlier correction that was right. |
| References | All 52 opened and read against the statements they support, and checked for retraction, expression of concern and indexing. |
| Similarity | Report run, and every match read and classified. Not scored. |
| Reporting | STROBE mapped item by item to page and line. |
| Declarations | Authorship against the four ICMJE criteria, CRediT, interests, funding, ethics, consent, data availability and AI use. |
What this is not: peer review, and not a verdict on the previous provider’s competence. It is a statement of what was done to a document, measured against what was bought.
2The level audit — 1,184 changes, classified
| Level of intervention | Changes | What this tells you |
|---|---|---|
| Mechanical: spelling, punctuation, articles, tense, number agreement | 1,102 | 93% of the work. This is copyediting, and it was done well. |
| Sentence-level: splitting, reordering within a sentence, word choice | 82 | 7%. Line editing happens at this level and this is not enough of it to constitute one. |
| Paragraph or section reordering | 0 | No structural work at all. |
| Queries raised to the authors | 3 | All three about a missing citation. None about a claim, a number or a method. |
| Changes that altered meaning | 1 | And it was made silently. See section 3. |
| Style sheet | None produced | Which is why the terminology drifts, and why the next paper from this project starts from nothing. |
Delivered level: copyediting. Charged as premium scientific editing. We are not alleging bad faith — the invoice named a tier, not a level, and tiers are not defined anywhere in this market. There is no industry-standard taxonomy: the Editorial Freelancers Association names four services, the Chicago Manual of Style splits manuscript editing into mechanical and substantive and puts line editing before copyediting, and Editors Canada’s Professional Editorial Standards 2024 defines four levels under different names again. They do not map onto each other. Until a supplier tells you which of those things they will do, a quote is not comparable to any other quote.
3The change that mattered
| Original | As returned by the previous editor | As we left it |
|---|---|---|
| “Exposure was associated with a higher rate of the outcome.” | “Exposure led to a significant increase in the outcome.” | “Exposure was associated with a higher rate of the outcome (adjusted HR 1.34, 95% CI 1.09 to 1.65).” |
The middle version reads better. It also asserts causation from an observational design, in the abstract, which is the objection this design always attracts and the fastest route to a reviewer deciding the authors do not understand their own study. It was made without a query, so the authors never saw the decision being taken.
The rule this illustrates, and the one we work to: an editor may change how a sentence reads without asking. An editor may not change what it claims without asking. Where the two conflict — and they conflict constantly in a discussion section — the claim wins and the query goes to the author. Our query log records every one of those decisions with its reasoning, which is also what makes answering a reviewer three months later a half-day rather than a week.
4Language Findings After the Previous Copyedit
| Finding | Count | Resolution |
|---|---|---|
| Earlier corrections kept | 1,102 of 1,184 | 94%. They were right. An audit that rejects everything the previous supplier did is an audit selling itself rather than reporting. |
| Sentences over 45 words in the methods | 11 | Split where they contained more than one procedure. Grammatical length is not readable length. |
| Over-hedging | 19 | “It may possibly be suggested that there could be some indication of” reduced to the claim. Over-hedging costs authors more than overstatement, because a reviewer reads it as a lack of confidence in the work. |
| Terminology drift | 3 variables, 2 abbreviations | Reconciled and recorded on a style sheet, which is handed over. |
| Unusual but clear constructions | Left alone | Two of four authors write in English as an additional language. The brief is clarity, not neutralisation, and a paper flattened into house prose has lost something it will not get back. |
| Further line-level changes | 310 | All tracked, all visible, 41 with a comment explaining why. |
5Checks Beyond Editing
| Check | Result |
|---|---|
| References opened and read | 52. 3 do not support the statement they are attached to — two inherited from another paper’s reference list, one supporting the opposite direction of effect in this population. |
| Retraction and expression of concern | 1 retracted, eleven months before the manuscript was drafted, cited in the introduction as current evidence. Removed and the claim re-evidenced. No copyedit, similarity report or reference manager catches this. |
| Indexing | 2 references from journals not indexed in any major database. Replaced. |
| Similarity report | 34 matches, all read: 26 correctly quoted and cited methods text, 5 the authors’ own earlier methods needing citation as text recycling under COPE’s position of 20 August 2024, 2 needing a citation added, 1 a genuine attribution problem in the introduction, resolved with the authors. The overall percentage is not reported in our findings because it is not a finding — Turnitin states that plagiarism cannot be determined from the score alone and that there is no single acceptable score, and COPE advises against using any percentage as a benchmark. |
| Reporting checklist | STROBE mapped. 22 items located by page and line; 4 incomplete, all answerable from documents the team held. |
| Numbers against tables | Every figure in the text checked against its source table. 2 mismatches, both in the abstract, both introduced in an earlier revision rather than by any editor. |
6Declarations
| Declaration | What was done |
|---|---|
| Authorship | Five authors listed. One met only the first ICMJE criterion — data collection, with no drafting, no recorded approval and no accountability. All four criteria must be met. Moved to the acknowledgements after a conversation with the corresponding author. |
| Contributions | No CRediT statement existed. Drafted with the authors across the fourteen roles of ANSI/NISO Z39.104-2022, which also settled an author-order disagreement that had been open for two months. |
| Editorial assistance | Neither provider was acknowledged. Elsevier requires that anyone participating in substantive aspects such as language editing or medical writing be recognised in the acknowledgements; Springer Nature and BMC require that plus the funding source, and warn that unacknowledged editorial services may constitute an undisclosed conflict of interest. Both providers named, including us. |
| AI use | One author had used a language model on two paragraphs and had told nobody, including the co-authors. Declared in the form the target publisher requires. The four largest publishers differ on whether basic grammar tools are exempt and on where the declaration goes, so “we declared it” is not a complete answer without naming the publisher. |
| Data availability | The statement offered more than the ethics approval permitted. Redrafted against the approval. |
| Ethics, consent, funding, interests | All present, complete and correctly referenced. Recorded as passing. |
7Deliverables Handed Over
| Artefact | Why it matters afterwards |
|---|---|
| The edited document, tracked, and a clean version | Every change visible and rejectable. You approve an edit; you do not receive a new document. |
| The query log | Forty-one queries with our reasoning, including every point where a change would have altered a claim. It is the document that answers a reviewer three months later. |
| The style sheet | Every terminology, spelling and formatting decision. Yours, and the reason the next paper from this project costs less. |
| The reference log | All 52, what was checked, what was replaced and why, including the retraction. |
| The similarity reading | All 34 matches classified, with the action taken on each. Not a percentage. |
| The completed STROBE checklist | Mapped to page and line, ready to upload. |
| A note on what we did not do | The shortest document in the pack and the one that prevents the next misunderstanding. We did not re-analyse the data, we did not assess whether the science is right, and we did not read the statistical code. |
One thing this report deliberately does not contain: a prediction of the outcome. We do not guarantee journal acceptance, and an editorial QC report cannot do so. It helps ensure that a manuscript is judged on its science rather than rejected for a retracted citation, an unsupported causal claim in the abstract or an authorship list that does not meet the criteria.
Benefits
Benefits and Limitations of Editorial Support
Editing does not improve the underlying study. It ensures that a sound study is not read as a careless one, and that you pay only for the level you receive.
Packages
Four Editing Levels Priced at 6,000 Words for Comparison
Editing is often sold as Standard, Advanced and Premium tiers that do not state which level of editing they include. The four packages below are the editing levels themselves, each costed at 6,000 words, a typical research manuscript. Longer or shorter documents are charged at the same rate per word, and the chooser above calculates the price.
Proofreading Before Submission or Publication
For final documents and typeset proofs returned by a publisher
From $1506,000 words · $0.025/word · minimum $110
Delivery 2–4 business days
- ✓Typographical errors, spelling, spacing, running heads
- ✓Figure, table and equation numbering and cross-references
- ✓Anything introduced by typesetting, checked against the previous stage
- ✓A final consistency read on a manuscript rather than a proof
Proofreading does not cover sentence structure, organisation or references. A hard-to-follow paragraph will remain hard to follow.
View a sample Select ProofreadCopyediting for Accuracy, Consistency and House Style
For drafts that read well but need corrections to mechanics, terminology and references
From $2106,000 words · $0.035/word · minimum $160
Delivery 3–6 business days
- ✓Grammar, usage, punctuation and spelling
- ✓Terminology and abbreviations made consistent, on a style sheet you keep
- ✓Reference styling and completeness, and every cross-reference checked
- ✓Journal house style, figure and table numbering
- ✓Tracked changes with a query where a change would alter meaning
Copyediting does not rewrite for clarity. It makes a manuscript correct rather than more persuasive.
View a sample Select CopyeditLine Editing That Preserves the Author Voice
For drafts that are correct but hard to read, the level most authors need
From $3006,000 words · $0.050/word · copyediting included
Delivery 5–8 business days
- ✓Everything in Copyedit, not billed twice
- ✓Clarity, flow and register, sentence by sentence
- ✓Paragraph order within a section; redundancy removed
- ✓Hedging that hides a finding, and overstatement that inflates one
- ✓Your voice preserved, with clear but unusual phrasing left unchanged
Line editing does not reorder sections or rebuild an argument. If the discussion claims more than the results support, substantive editing is needed.
View a sample Select Line editSubstantive Editing with Full Quality Control
For drafts with structural problems, and documents already edited elsewhere
From $4506,000 words · $0.075/word · all lower levels included
Delivery 7–12 business days
- ✓Everything in Line edit, at document scale
- ✓Section order and balance; the introduction narrowed to one question
- ✓Every number in the text checked against the table it came from
- ✓Every claim measured against the result it rests on
- ✓Reporting checklist mapped to page and line; references opened and read
- ✓Declarations reviewed; similarity report read rather than scored
Developmental editing — rebuilding rather than improving — is $0.100 a word and starts with a plan you approve before any rewriting.
View a sample Select SubstantiveDeliverables
The query log
Every querywith the reasoning
Including every change we did not make
Where a change would alter what the document claims, we raise a query with our reasoning rather than make an edit. The log also helps you respond to reviewers later.
Deliverables
The style sheet
One pageevery decision recorded
Terminology, spelling, abbreviations, format
Prepared before editing starts and handed over at the end. It keeps terminology consistent across a long document, allows a second editor to work without contradicting the first, and reduces the cost of later papers from the same project.
Deliverables
The reference log
Each oneopened and checked
With retractions and indexing checked
Every reference is read against the statement it supports, with a record of what was replaced and why. This pass identifies retracted papers still cited as current, which copyediting, similarity reports and reference managers do not detect.
Priced alongside any level: developmental editing at $0.100 a word; reference verification as a standalone pass; similarity report read and classified; reporting-guideline mapping on its own; declarations and authorship review; journal reformatting for a resubmission; figure and table preparation; translation review against a source text; and expedited delivery where capacity allows, quoted on capacity rather than on urgency. Institutional and publisher arrangements are quoted per title or per cohort rather than per document.
Quality Assurance
Our Service Commitments and Limitations
We do not guarantee journal acceptance. Our service commitments and their limits are set out below.
The Pubrica revision commitment
Revision Support Until Agreed Requirements Are Met
We revise the document until it meets the agreed level, with no limit on rounds within the agreed scope, and the same editor remains available when reviewers or supervisors respond.
01
Editing Level Confirmed in Writing
The level is named in the EFA, Chicago and Editors Canada vocabularies, with its exclusions. If the delivered work does not meet that level, we correct it at our cost within the agreed scope.
02
Queries for Any Change in Meaning
Every point where an edit would alter what the document claims is raised with you as a query, with the supporting estimate or evidence, and recorded in the log.
03
Second Reader on Every Document
An editor who did not do the editing reads the document before it is returned, because editors reviewing their own work tend to read what they intended.
04
Corrections at Our Cost
If we introduced an error, missed a reference or cited the wrong guideline version, we correct it free of charge within the agreed scope and inform you first.
Service limitations. We do not guarantee journal acceptance, we do not influence editorial decisions, and we have no relationship with any editor or reviewer that could affect one. We do not offer similarity reduction: we explain what each match is and recommend a citation where needed rather than rewording text to lower the score. We do not offer AI-detection clearance, because published false-positive rates make it impossible to provide reliably, and those rates are highest for authors writing in an additional language. We do not issue certificates; instead, we confirm in writing exactly what we did, which an institution may require. This service is not peer review. We do not edit assessed work where an institution’s regulations prohibit assistance. We do not alter a claim your evidence does not support without asking: we query it, show you what it rests on, and if you want it kept as written, we restore it and record our advice.
Free 1,000-Word Sample Edit
We edit 1,000 words of your own document so that you can see the level of intervention and how we treat your voice before you commit. The sample also confirms whether the level you requested is the level the document needs, and if it needs less, we quote less. It takes two business days, at no cost and with no obligation.
Confidentiality of Unpublished Work
A non-disclosure agreement is signed before any material moves, and editors and project staff are bound by internal confidentiality agreements as well. Unpublished data, drafts, decision letters and supervisor correspondence are held under the UK GDPR on access-controlled systems within a quality management system certified to ISO 9001:2015 — certificate H3CR9TCU20251212INDQM10N — kept only as long as the project needs them and never used to train anything. We do not reuse your text on anyone else’s project and we do not publish your work as a sample: every sample on this site is redacted or composite.
Acknowledgement of Editorial Support
Our involvement goes in your acknowledgements, naming the company and what we did. The ICMJE is explicit that writing assistance, technical editing, language editing and proofreading do not confer authorship; Elsevier requires that such contributors be recognised; Springer Nature and BMC require that plus the funding source, and warn that unacknowledged editorial services may constitute an undisclosed conflict of interest. Where you used another supplier before us, they belong in the acknowledgement too, and we will say so.
Preservation of the Author’s Voice
Our aim is clarity rather than uniformity. Where a construction is unusual but clear, it stays. This matters most for authors writing in an additional language, the largest group using editing services. Where we apply consistency, it is recorded on the style sheet, so you can see and reverse any decision.
Client Stories
Client Feedback on Editorial and Quality Support Projects
Testimonials are rephrased from client feedback and do not claim publication outcomes, because editing does not determine them.
“They told us the previous editor had done a copyedit, not the tier we paid for, and showed us the change counts.”
“Our similarity score had frightened us for a week. They read all thirty-four matches and twenty-six were our own correctly cited methods.”
“They queried a sentence instead of improving it. The improved version would have claimed causation from a cohort study.”
FAQs
Editorial and Quality Support FAQs
If you are unsure which level you need, send the document and we will edit 1,000 words free of charge and advise you. Ask an expert. No quote is required.
We support journal manuscripts of every design; systematic reviews and meta-analyses; case reports; theses and dissertations; grant applications and research proposals; book chapters and scholarly reports; regulatory and clinical documents; and institutional or publisher work at volume. We cover medicine and the health sciences, the life sciences, engineering and computing, psychology and the social sciences, economics and management, and the humanities. We also review documents already edited elsewhere, as shown in the sample report on this page.
No. We bring manuscripts to high editorial and formatting standards, but we do not guarantee journal acceptance. Acceptance is an editorial decision made by journal editors, with whom we have no relationship. In the best-documented analysis of reviewer comments, the leading reasons for rejection were statistical problems, overinterpretation, inappropriate instrumentation and sample size, with “text difficult to follow” ranking fifth. Editing addresses the fifth of these, not the first four.
There is no acceptable score, because the score is not a measure of plagiarism. Turnitin’s own guidance says so in almost those words: “plagiarism cannot be determined by the similarity score alone”, the number is “simply the percentage of text in a submission that matches other sources”, and there is “no single ‘acceptable’ or ‘perfect’ Similarity score”. A correctly quoted, correctly cited passage raises the number exactly as a copied one does. Nor is there a published threshold. COPE states that it “does not recommend using a specific percentage of duplication or overlap as a benchmark”. Elsevier tells its editors a high score does not necessarily indicate plagiarised text. Springer Nature publishes bands for editors — screen quickly above 10%, check carefully above 20% — and says explicitly that individual source scores matter more than the overall index. The “under 15%” figure that circulates widely traces to no publisher, no guideline and no primary source we could find. What we do is read every match and tell you which need a citation, which are your own earlier work and need handling as text recycling, and which are a real problem.
No. Seven GPT detectors tested on human-written TOEFL essays produced an average 61.3% false-positive rate, and 19.8% of those essays were called AI-generated by every detector, while essays by native-English school students were classified almost perfectly (Liang et al., Patterns, 2023). The tools key on text perplexity, so they penalise writing with less lexical variety — which means they systematically penalise the authors most likely to buy reassurance about them. Turnitin’s own claims have moved a long way from the “98% confidence” of its 2023 launch: it now publishes no headline accuracy figure, does not surface results below 20%, warns that submissions under 300 words are unreliable, and states that a score “should not be used as the sole basis for adverse actions”. COPE’s September 2025 guidance is the line to hold: detection tools “should not be relied upon in isolation”. What we will do is help you write an accurate declaration of whatever AI you actually used, in the form your publisher requires.
Use the chooser above, then send the document and we will confirm the level by editing 1,000 words free of charge. In summary: if the problem is typographical, proofreading. If it is grammar, terminology and references, copyediting. If the sentences are correct and hard to read, line editing, the level most authors need. If the structure or argument needs work, substantive editing. If the document needs rebuilding rather than improving, developmental editing, which starts with a plan you approve before any rewriting. The reason this question is hard is that there is no industry-standard taxonomy: the Editorial Freelancers Association names four services, the Chicago Manual of Style splits manuscript editing into mechanical and substantive at 2.49 and 2.50 and treats line editing as an earlier separate pass, and Editors Canada’s Professional Editorial Standards 2024 defines four levels under different names again. They do not map onto each other, which is why two honest quotes can differ fourfold.
An editor in your subject area, matched to the document type as well as the discipline, edits your document, and a second reader who did not do the editing checks it on every project. You are told who is doing the work, and the same editor stays with the document through revisions and responds when reviewers or your supervisor return comments. Where we cannot staff a level properly in your field, we decline the project.
Not without asking. Editors may change how a sentence reads, but not what it claims, without your agreement. Where the two conflict, which is common in discussion sections, we raise a query with the estimate or evidence the sentence rests on. The sample report on this page shows an example: another provider changed “was associated with” to “led to” in an abstract without a query, which asserts causation from an observational cohort. Every query we raise, and our reasoning, is recorded in a log you keep.
We provide written confirmation of exactly what was done, at what level and on what date, which your institution may require for a declaration. We do not present it as an editing certificate, because certificates carry no editorial weight. No publisher we checked requires one, treats one as bearing on a decision, or rejects a paper for lacking one. The only publisher-side mention we found is MDPI, which says that if you use a service that issues a confirmation certificate you may send a copy to the editorial office — permissive, with no stated effect.
Sometimes, and the answer is your institution’s rather than ours. Universities differ sharply: some permit language editing within stated limits, some require a declaration naming the editor and the work done, some prohibit third-party assistance in assessed work entirely. Please send your institution’s regulations before the thesis. Where help is permitted we work inside the limits — which usually means language and consistency but not structure or argument — and give you written confirmation of exactly what was done so your declaration is accurate. Where it is prohibited, we decline, and we will not help anyone work around it. For journal articles the position is different and simpler: we found no publisher that prohibits professional editing, and all the major ones require only that it be acknowledged, with Springer Nature and BMC also asking for the funding source of the assistance.
A non-disclosure agreement is signed before any material moves, and editors and project staff are bound by internal confidentiality agreements as well. Unpublished data, drafts, decision letters and supervisor correspondence are held under the UK GDPR on access-controlled systems within a quality management system certified to ISO 9001:2015, certificate H3CR9TCU20251212INDQM10N, kept only as long as the project needs them and never used to train anything. We do not reuse your text on anyone else’s project, and we do not publish your work as a sample — every sample on this site is redacted or composite. A document Pubrica holds is confidential whoever wrote it, so the sample on this page names neither the client nor the earlier editor, and never will.
Related Services
Related Writing and Publication Services
Services for writing a document, checking it before submission and preparing it for the journal.
1IF IT IS NOT WRITTEN YETwriting rather than editing
Scientific and academic writingWriting from your data when a document needs to be drafted rather than edited. Medical writingThe clinical and regulatory side of the same discipline.Insights
Further Reading on Publication and Funder Policies
Three articles on funder rules that affect where an edited paper can be published.
Request a Free Sample Edit of Your Document
The sample edit shows the level of intervention and how we treat your voice, and confirms the level your document needs. It takes two business days, at no cost and with no obligation, and if the document needs less than you requested, we quote for the lower level.