How it works
How ScholarSignal evaluates your idea
The evaluation connects your idea to the literature retrieved for it, then checks the contribution, the design and possible venues. These are the steps, what each one produces, and where it stops.
From idea to report
Four stages run after you submit. You can close the tab after the first.
You describe the idea
A title and a few sentences to a full abstract: the question, setting, data if any, and what the paper would claim.
Optional: stage and target journal
Live literature search
Your idea becomes construct-aware OpenAlex queries, filtered for relevance and expanded along the citation graph.
Up to 12 closest existing works
Checks, gates and three reviewers
Automatic design checks, then an editor rubric applied from three reviewer perspectives to your idea and the retrieved work.
A verdict, its reasons and the objections
Report and radar
The report arrives by email and stays in your account; your topic joins a private radar of matched calls and papers.
Venues, next steps, weekly radar
How the literature search narrows down
Originality can only be judged against real papers, so the search runs before any judgement is made.
If the search finds fewer than four close works, the report says so and treats the neighbourhood as weak evidence rather than proof that the idea is new.
The four gates
Each gate has a fixed scale. Together they lead to Go, Refine, Pivot or Retire. See one example per verdict.
Gate 1
Contribution viability
Would readers of the target literature understand something differently, given the closest existing work?
Gate 2
Claim–design alignment
Can the planned data and design support the main claim, or does the claim outrun the design?
Gate 3
Journal fit
Where does this literature publish, and what have your target journal and the listed journals near your idea printed closest to it recently? The report lists those articles as models.
Gate 4
Desk-reject risk
Read as an editor with a few minutes: is there tension, surprise and resolution, or a data report?
Checks that do not depend on the model
Before any model reads your idea, fixed rules check the description for patterns the research literature documents. They are passed to the evaluator to verify and shown in your report with the guide on fixing them.
Causal wording without an identifying design
Causal verbs with a survey or observational data, the most common overreach in the social sciences.
Predictor and outcome from one survey
A single-source survey invites common method variance unless measures are separated in time, source or method.
Intentions standing in for behaviour
Stated intentions predict purchase only loosely; claims about loyalty or sales need a behavioural measure.
Samples, sites and instruments
Small samples, one site or a student sample, and instrumental variables that need a strong case.
After the evaluation, two more checks run on its output: a contribution statement that leans on the data, and a main effect already tested in three or more of the closest works.
Three reviewer perspectives, one verdict
The evaluator reads your idea three times, as a desk editor, a methods reviewer and a reviewer who knows the closest literature. Each run first writes the strongest reasons to reject, then judges the gates. The report follows the middle verdict and shows all three, so disagreement stays visible. The example below is illustrative.
What it stands on, and where it stops
The rubric and the evidence
The rubric is distilled from roughly 15 published editorials and editor essays on what survives desk review, the public record of what editors say they reject and why. No private review files or rejection letters are involved, and nothing is fine-tuned: the rubric lives in the prompt, not in model weights.
When your idea touches a topic covered in the evidence hub, the evaluator also sees what published research already establishes there, so restating a known finding is recognised as such.
The model
Evaluation and ranking use language models through OpenRouter; embeddings represent the meaning of project text for matching. Model configuration can change, and reports include model or provider information where the service supplies it. Requests use a routing restriction that asks OpenRouter to exclude providers that collect submitted text for training. This is a provider policy restriction, not a guarantee of zero retention.
Limits
- The rubric has been checked against a small set of papers with known outcomes. It has not yet been validated against submission outcomes at scale.
- The literature search is broad but not exhaustive: OpenAlex coverage and query construction both have gaps, so a missed neighbour is always possible.
- A verdict is a structured second opinion, not a prediction of what any editor will do with your paper. Journal-fit scores are not probabilities of acceptance.
- Incomplete abstracts and thin retrieved literature limit what can be concluded. Check key works and venue requirements yourself.
- Opportunity dates may be unknown or unverified. A verified date requires a recorded check of the official source; no opportunity listing guarantees that a call remains open.
Your data
Your projects are not published as public examples. The service stores your submitted versions and results, and uses processing providers to produce evaluations and radar matches. Anyone you share a private result link with can read that result, so keep these links private. Details about accounts, providers, saved history and data requests are in the full privacy note.
Free and Pro
Both plans use the same evaluation process.
Free
€0
One lifetime evaluation and one active project, with the report by email and in your workspace.
Pro
€15 / month before applicable taxes
Five active projects and ten completed evaluations per paid billing period.
A revision creates a new report with its original inputs preserved. Failed evaluations do not consume your allowance. Archiving a project frees an active slot without removing its history. After Pro ends, you can still read past results and select one project for your free radar. Checkout and automated radar delivery depend on availability shown in the product.
See it on your own idea
One free evaluation. It takes a few minutes to describe your idea.