What the Professional report includes, and what it does not

You are deciding whether to pay for a report whose contents you cannot see until after you have paid, which is the problem with everything sold this way. And there is an asymmetry working against you: a product page always lists what a thing does and never what it does not, so you discover the limits on opening it, once you have already made the mental transfer of «this solves my submission». This article is the complete list in both directions: what comes out of the report, section by section and in the order you read it, and what you will not find inside no matter how hard you look. The sample is published in full and downloads as a Word file, so you can skip the text and go straight to it; if you would rather have the list before the file, read on.

The verdict, which is the first thing you read

The first screen carries three things at once. A 0-10 score for the whole manuscript, an editorial decision written with the labels journals actually use (desk-rejection, reject, major revisions, minor revisions, accept) and a separate score for introduction, method, analysis, results and discussion. That second row is the one most people look at twice, and rightly so: it is perfectly normal for a manuscript to carry a 7 overall and a 4 in analysis, and that tells you where the problem lives before you read a single critique.

Below it comes the editor's letter. Long prose, the kind you get when a journal replies to you properly, with the reasoning for why the decision is that one and not another: it is not a summary of the critiques, it is the editorial argument. Alongside, two to four real journals where the manuscript would fit as it stands today, each with its reason. And if you named a target journal when you uploaded, the report reads your manuscript against that journal's scope, gives you a desk-rejection risk of high, medium or low, and adds what OpenAlex and DOAJ say about it: how many works it indexes, whether it is open access, whether it appears in DOAJ and whether it charges to publish.

Two caveats printed inside the report itself, which I am repeating here because they are the kind of nuance a sales page loses. The journal fit is a reading of your text against that journal's published scope, not an answer from the journal. And the score is shown from 0 to 10 with no decimals on purpose: internally the scale runs 0-100, but showing you a 73.4 would suggest a precision the measure does not have. A 7 and an 8 are the same thing. A 4 and an 8 are not.

Both reviewers, in full

Professional brings two simulated reviewers with different roles: one reads from your substantive area —theoretical framing, the gap in the literature, contribution, discussion— and the other is a methodologist who goes straight at method and analysis. From each you get their background (where they are reading you from), their overall assessment, major comments developed as paragraphs, minor ones, and their recommendation to the editor. Every major objection carries the literal quote from your text that prompted it and how to resolve it, which is the difference between a critique and a complaint.

Here is the concrete boundary with the Basic report, and it is worth saying plainly: the Basic one develops one finding from the reviewer, the one that decides the review, with its quote and its fix, and adds the code checks (references against eight databases, p values recomputed, figures that contradict each other). Professional brings every critique the two reviewers find. I will not promise you a number, because the number depends on the manuscript and anyone promising one before reading it is making it up; what is promised is the scope. On top of that, each critique is labeled by consequence: whether it sinks the paper, whether you answer it in the letter without touching the data, whether it is cosmetic, or whether it is past fixing because the data are already collected.

One literal extract of what comes out, taken from a real report published in full on the sample page: «The two countries are compared on nine emotional and worry variables with no correction for multiple comparisons, and the seven significant ones are highlighted. At alpha = .05 with nine contrasts, the probability of at least one false positive is around 37%». That is the kind of objection you want to read before a reviewer does: in your version it can still be fixed, in theirs it is already a letter with a decision underneath.

What the code checks, and not a model

This is the part of the report that cannot be wrong in your favor or against you, and it is the reason the product exists. Every statistic in your text with a p next to it is recomputed from its degrees of freedom. No model reads it: it is arithmetic. The report flags the ones that do not add up and separates two different severities, because they are not the same thing: the figure being mistyped without the conclusion changing, or the p you report and the one the statistic yields falling on opposite sides of .05, which is when the outcome of the test depends on which of the two is correct.

Effect sizes get the same treatment: the partial eta squared you report follows from your F and your degrees of freedom by an identity, so it either adds up or it does not. The most frequent error that shows up here is not a calculation error, it is a column error —copying eta squared where partial eta squared belongs— and as soon as the model carries more than one factor those two values stop coinciding: the partial one is always larger, and the gap grows with whatever the other effects explain. A third figure is added that almost nobody reports and every reviewer asks about: the smallest effect your sample could have detected. If you upload your output file, the report also checks that the analyses were run correctly and that their figures match what the text says.

And the references, one by one, against eight databases: Crossref, OpenAlex, PubMed, Semantic Scholar, DOAJ, OpenLibrary, Google Books and arXiv, up to three hundred per report. A reference counts as confirmed only when at least two agree on title and authors; a DOI that resolves is not enough, because a valid DOI can be attached to the wrong paper, and that is exactly the failure a reviewer catches and you do not. The procedure, if it interests you on its own, is set out in how to check your references before submitting.

Here is the caveat that matters most to me in the whole report. The prose is written by a language model: the reviewers' comments and the editor's letter are drafted, not calculated. What that model cannot touch are the numbers around it, because they reach the text already settled from the code. And the report also tells you what it could NOT check and why, instead of staying quiet about it: tests that arrive without degrees of freedom, p values already corrected for multiple comparisons —which by definition do not follow from the statistic— or a bracket that might be the degrees of freedom or might be the N. Those are not your errors, and they are labeled as what they are.

From critiques to work: action plan and reanalysis

The same objections from above, turned into tasks. The action plan is prioritized by impact on acceptance —critical, high, medium— with the effort each fix demands and what fixing it unblocks. This is the part that lets you split work across a group, because not everything a reviewer flags is worth the same and without prioritizing you end up spending the week on the cosmetic. Where an objection is past fixing because the data are already collected, what appears is not an impossible task: it is the Limitations paragraph written out for you to paste, which is literally the only thing that can be done with that objection.

The reanalysis plan is the part most people do not expect to find. For every analysis the reviewer asks you to redo, the instructions in the software you actually use —R, SPSS, jamovi, JASP, Mplus, Stata, MATLAB, G*Power, Python— with the code where there is code to give; where the software is driven by clicking, you get the menu path with the real names instead. It is not a generic tutorial on «how to run a mixed ANOVA»: it is your model, with your variables and with the specific correction being asked of you.

Then come the pre-submission checks, in the order they happen to a real manuscript. The editorial screen: what would sink you before the manuscript reaches a reviewer, with the risk attached. What it still needs to sit at Q1 level, which is the distance between what you have and what a first-quartile journal publishes. And the cover letter to the editor, drafted from your own text, with brackets deliberately left unfilled for the word count, the conflicts of interest and the funding: those are things only you know, and I would rather leave them blank than invent them for you.

And one section nobody asks for and almost everybody needs: how to declare your use of AI in the submission. The ICMJE recommends that journals require authors, at submission, to disclose whether they used AI-assisted technologies in producing the work, and that chatbots are never listed as authors because they cannot be responsible for the accuracy, integrity or originality of the work (checked in September 2026). If you run your manuscript through this tool or through any other, that applies to you, and I would rather hand you the wording than leave you the problem.

What you keep when you close the browser

Two files. The full report in Word, and your own manuscript handed back with the review inside it: the editor's letter up front and every objection anchored in the margin to the exact sentence it criticizes. The second is the one that gets forwarded to a co-author, and that is not a formatting detail: reading a critique pinned to the paragraph it is about is nothing like reading a list of critiques detached from the text. Both also land in your inbox, so you are not depending on keeping a tab open.

You also take away a case file number that lives for thirty days, which is exactly how long the report itself is kept. It travels inside your Word file as well, so if you come back four months later the document still works even though the number has expired. When you have the corrected version you come back with it and you are told which of those objections you have closed and which are still standing, with the line of your new text that proves it; that cross-check costs nothing. Reading the whole manuscript again is another full round and is paid for like any other. I am putting the price here rather than in the small print, because this is what turns the tool into two rounds instead of a single verdict, and you should know where the line sits before you pay for the first one.

About what happens to your text, because that is usually the next question. In Basic reports it is processed and discarded when the report finishes. In Professional ones, while you pay, it waits encrypted and tied to the payment —it has to, so the report can still be generated if you close the browser halfway— for 24 hours at most, and it is deleted as soon as the report is generated; the resulting report is kept for thirty days so you can retrieve it. It is written that way in the published data processing agreement, with the sub-processors named one by one. You can read it before uploading anything, which is what I would recommend with any tool you hand an unpublished manuscript to.

What it does not do

It does not fix a bad design. This is the first limit and the most expensive one. If you collected forty cases by convenience with no comparison group, if the instrument is not validated in your population, if the variable that would explain the effect was never measured, then no report fixes it: not this one, not a human reviewer's, not your supervisor's. The data are already collected. All a pre-submission review can do is tell you now, while you can still reframe the question or narrow the reach of your conclusions, rather than in four months' time. That is a lot, and it is still not the same as fixing it. Which is why, in those cases, the report gives you the Limitations paragraph instead of a task.

It does not guarantee acceptance. Nobody can, and anyone guaranteeing it is describing a different business. The decision is made by a human editor with criteria neither you nor I control: fit to the issue, the queue of manuscripts already on their desk, what the journal has just published on your topic and sometimes simply how much room is left. What a pre-submission report does is cut down the number of avoidable reasons you can be rejected for; the other reasons remain untouched. If you want the real breakdown of why manuscripts are rejected in the first quartile, it is in why you get rejected from Q1 journals.

It does not replace your supervisor. It does not know your field from the inside, or why you chose that design two years ago, or the line your group has been building for fifteen years, or what your department expects from this particular paper, or who you would rather not pick a fight with in the discussion. It can point out that an analysis does not support the conclusion you hang on it; it cannot tell you whether the question was worth asking. That conversation is still with a person, and if the tool saves you from having it, you are using it wrong.

It does not recompute what lives only inside a table. Word tables do come through and the reviewers read them, but the recomputation works on running text: a result split across cells can arrive as «t 48 2.31 .025», and no test is recognized there. If you want that figure checked, it has to be written into the results as well, which is how APA asks you to report it anyway. And a scanned PDF, whose pages are images, yields no extractable text at all: upload the original Word file.

It does not check that a citation says what you claim it says. Reference verification confirms that the cited work exists and is the one you declare; it does not confirm that the work supports what your paragraph attributes to it. No database and no model verifies that, and a human has to look at it. That human is you.

And it is not peer review

This is the limit that matters, which is why it gets its own section. Peer review is a procedure with a specific structure: there is an editor handling your manuscript who chooses and invites the reviewers, those reviewers are people independent of you, and their judgment feeds an editorial decision you do not control. Elsevier describes it that way on its own page explaining what peer review is: the handling editor is the one responsible for choosing and inviting reviewers, and the system exists to validate academic work (checked in September 2026).

None of that is happening here. There is no editor, no invitation, no independence —you commissioned it, on your own text, and you decide what to do with the result— and no decision that binds you to anything. What there is, is a self-assessment: you, on your manuscript, before you submit it. That the report imitates the format of a review package (two numbered reviewers, an editor's letter, a recommendation) is a presentation decision, because that format is the one this audience reads at a glance; it is not a claim about what the thing is.

The distinction is not a footnote technicality. It has three practical consequences worth being clear on before you pay: you cannot cite this as prior review in your letter to the editor, it does not count as external review in any assessment or accreditation exercise, and if your journal asks about the use of AI tools, the answer is yes and it goes declared. Everything this report can do for you happens before submission, never inside the process.

And there is a good consequence, which is the exact other side of the same coin: because it is not peer review, you are not spending a bullet. You can run the manuscript, correct it and come back with the case file to be told what you have closed, as many times as you need, and none of it goes on any record. A real reviewer reads you once, says what they see, and then there is a decision that stays in the journal's record. Here there is no decision and no record, which is why you can afford to discover the problem with your method while the manuscript is still yours.

Frequently asked questions

What exactly is the difference from the Basic report?

The Basic one develops one finding from the reviewer, the one that decides, with its quote and its fix; gives you the verdict and the score; cross-checks your citations against your reference list and checks every reference against the eight databases (up to 30); and recomputes your p values and the figures that contradict each other. Professional adds every critique from both reviewers, the full editor's letter, the desk-rejection screen, references up to 60, effect sizes checked against your F and the power, the prioritized action plan, the reanalysis with code, the journals it fits, the cover letter and the two downloads. If in doubt, run the Basic one first: same manuscript, same pipeline, and you buy knowing what will arrive.

Does it work for a thesis or only for papers?

Both: you upload the paper or the thesis as Word or PDF. In a thesis the reference check is usually what saves you the most work, because a bibliography of two hundred entries is an entire afternoon of checking by hand and it is where errors accumulated over years tend to hide. What changes is the reading frame: a paper is read with an editor in mind, a thesis with a committee.

If the report says desk-rejection, do I bin the manuscript?

No, and it is worth understanding what that label is. A desk-rejection is what an editor does before sending a manuscript out to reviewers, and what you are reading is not any journal's answer: it is this report's reading of what would be screened out, and you have it now, while everything can still be changed. In practice the screen almost always comes down to two or three concrete things, or to the target journal not being the right one. Look at the list of suggested journals before you rewrite anything.

Can I show it to my supervisor or a co-author?

Yes, and the file is built for that. The version that actually circulates is not the standalone report, it is your manuscript handed back with each objection anchored in the margin to the sentence that prompted it: it opens in Word, LibreOffice or Google Docs and reads like any commented document. One piece of advice you did not ask for: if you send it to your supervisor, tell them it is a self-assessment you ran yourself, not an external review. That is what it is, and presenting it as anything else costs you credibility with exactly the person you can least afford to lose it with.

What if a critique is wrong?

It happens, and two cases need separating. The code checks do not hold opinions: if they say a p does not follow from its degrees of freedom, that is arithmetic and the figure is what needs revisiting. The reviewers' comments are drafted by a model and can indeed flag something your subfield does differently. When that happens, my rule is this: if a specialist reader misread your paragraph, a real reviewer can misread it the same way, and what usually fails is not the analysis but the explanation. Dismiss the critique if you are sure, but read the sentence that triggered it first.

Do I have to declare to the journal that I used this?

Check your journal's specific policy, because they vary, and use the section of the report that drafts it for you. The background reference is the ICMJE, which recommends that journals require authors to disclose at submission whether they used AI-assisted technologies in producing the work, and that a chatbot is never listed as an author because it cannot be responsible for the accuracy or integrity of the work (checked in September 2026). Declaring it does not penalize you; being found out without declaring it does. You can see how that section looks in the sample report and on the pricing page.

Keep both lists together, because they are only worth anything together. What it does: an editorial verdict with the score broken down by section, two professional reviewers with a quote and a fix on every objection, a prioritized plan, reanalysis with code in your own software, references against eight databases and your figures recomputed by arithmetic, delivered in Word with your manuscript annotated. What it does not: fix a design whose data are already collected, guarantee an acceptance an editor decides, replace your supervisor, or be peer review. If reading the second list makes you think your problem sits precisely there —the design, the question, the analysis nobody in your group handles— no tool solves that and a conversation does: that is what I do in consulting, and only in psychology and health sciences.