How to check if your research idea is novel before you write the paper
To check whether a research idea is novel, write the claim in one sentence, search at least three sources for the closest existing work, read what those papers actually do, and list exactly what your idea adds. If you cannot name a specific difference in the problem, method, data or result, the idea is not yet novel enough to write up.
Most researchers find out their idea has already been published after months of work, not before. A short, structured check at the start costs a few days and protects the months that follow. This guide gives you a method you can run by hand, then shows where a tool can speed it up.
What does “novel” mean to a journal?
A paper is novel when it adds something the existing literature does not already contain. That does not mean a completely new field. Most published work is novel in a smaller way, and editors usually look for a clear difference on at least one of five things:
| What can be new | Plain meaning | Example |
|---|---|---|
| The problem | A question nobody has asked, or asked in this setting | A known disease studied in a new patient group |
| The method | A new technique, or a real change to an old one | A modified loss function that fixes a known failure |
| The data | A new dataset, region, period or population | Crop-yield records from a region never studied |
| The result | A finding that changes what we believe | An effect that earlier papers reported as absent |
| The combination | Two known ideas joined in a way that gives a new outcome | A known model plus a known sensor, with a new measured benefit |
Novelty is also a matter of degree and of venue. A small, careful step can suit one journal and be too incremental for another. Your job at this stage is not to guess the journal’s verdict. It is to find out honestly what your work adds.
Step 1: Write the idea as one claim sentence
If you cannot say the idea in one sentence, you cannot search for it. Use this template and fill every blank with specifics:
Template / exampleWe show that [method or approach] improves [measured outcome] for [problem] on [data or setting], compared with [closest existing approach].
Then write down five to ten search phrases that mean the same thing. Include the name you use, the name other fields use, and the older name for the same technique. Many “new” ideas turn out to be old ideas with a new label, and many old ideas are hidden behind a different label.
Step 2: Search in at least three places
One search engine is not enough. Different sources index different material, and the paper that overlaps with yours may sit in only one of them.
- Google Scholar: widest coverage, good for phrases and for “cited by” chasing.
- Scopus or Web of Science: curated journal indexes, good for filtering by subject and year (often through a university login).
- PubMed for biomedical topics, or a field database such as IEEE Xplore for engineering.
- arXiv and other preprint servers: recent work that journals have not published yet.
Two habits find overlap that keyword searches miss. First, read the reference list of the best paper you find (backward chasing). Second, open its “cited by” list and read the newest entries (forward chasing). Stop when new searches keep returning papers you have already seen.
Tip Search the last two years most carefully. A paper that appeared after your last literature review is the most common reason a “new” idea is already taken.
Step 3: Compare the closest papers in a table
Pick the five to ten papers closest to your claim. Read the abstract, the method and the results, not only the title. Then fill in a comparison table with one row per paper. The last column is the one that matters.
| Paper | Problem | Method | Data | Result | What my idea adds |
|---|---|---|---|---|---|
| Author A, 2024 | Same | Baseline model | Public set X | Accuracy 0.84 | Handles missing values; they drop them |
| Author B, 2025 | Similar | Hybrid model | Own data | Accuracy 0.88 | Different data; no test on set X |
| Author C, 2023 | Wider | Survey | n/a | n/a | They name my gap as future work |
The numbers above are an illustration of the layout, not real results. Fill yours with what each paper states. If a cell in the last column stays empty, that paper may already cover your idea, and you should read it again more slowly.
Step 4: Decide with a simple test
Look at the last column of your table and ask two questions. Does your idea differ on at least one of the five things above? And would that difference change a conclusion or allow something that was not possible before? A difference nobody would care about does not make a paper.
| What you find | What it means | What to do next |
|---|---|---|
| A clear, useful difference on one or more axes | Novel enough to continue | Write the novelty statement and plan the experiment |
| Differences exist but they are small | Incremental | Strengthen the idea, add a second contribution, or aim at a venue that accepts incremental work |
| A paper already does what you planned | Already done | Change the problem, the data or the method, or stop. Do not rewrite it as new |
Show your table to your supervisor or a colleague before you decide. A second reader often spots a paper you read as different but a reviewer will read as the same.
Step 5: Write the novelty statement
The novelty statement is the short passage in your introduction (and your cover letter) that says what is new. A clear version has three parts: what exists, what is missing, and what you add.
Template / exampleEarlier work predicts crop yield from weather data alone [refs]. It does not use soil-moisture readings, which several studies identify as a limitation [refs]. We add soil-moisture features and show a measurable change in error on [data].
- Name the gap with citations, so the reader can check it.
- State the contribution as a result you can measure, not as an adjective.
- Avoid “first to” unless you have searched thoroughly and can show it. If you are wrong, a reviewer will find the paper.
- Keep it to two or three sentences. Detail belongs in the body.
Common mistakes
- Searching only with your own wording and missing the older name for the same idea.
- Reading titles and abstracts only, then missing the paper whose method section matches yours.
- Treating “nobody has done this exact experiment” as novelty. The question is whether it adds knowledge.
- Stopping the search after the first week and never checking again before submission. Search once more just before you submit.
- Claiming novelty because a database search returned few results. Few results can mean a poor search term.
Where a tool can help
A manual check is slow because you read many papers just to rule most of them out. A tool can do the first pass, as long as you stay the judge. In EasySeer, the novelty check reads your handwritten or typed concept and your proposal, base papers and a sample experimental result, searches published papers, and scores each out of 100 with a reason and the file it came from. It then combines them into a final score shown against the threshold you set, and lists the open questions, such as a missing literature review or metric.
In the sample project, the final score is 71 out of 100 against a threshold of 75, so the report points to what to strengthen before any code is written. That is a prompt to improve the idea, not a verdict, and it is not a prediction that a journal will accept the work. Editors and reviewers judge novelty, and you and your supervisor should check every claim the report makes.
See the novelty check stageRead: how to read your novelty score
Questions
Does every research paper have to be novel?
Most journals expect a paper to add something new, but the kind of newness varies. Some venues publish replication studies, negative results or confirmatory work and say so in their aims. Read the journal’s scope and author guidelines before you decide how to frame the contribution.
How long should a novelty check take?
A careful manual check usually takes several days of reading, depending on how crowded the field is. A tool can shorten the first pass, but the final comparison of the closest papers should still be done by you.
What if someone has already published a similar idea?
Compare it carefully on the five points: problem, method, data, result and combination. If you still differ in a way that matters, cite the paper and state the difference. If you do not, change the problem, the data or the method, or consider a different project.
Can an AI tool tell me if my idea is novel?
It can help find and compare related work and show you where your concept looks thin. It cannot know every paper, and it cannot predict how editors will judge your work. Use it as a decision aid and verify the papers it points to.
Keep reading
How to run the novelty check
Set how many related papers to screen and the score you need, press Run novelty check, read the scor…
Guide · 2 min readHow to prepare your inputs
Only the handwritten concept is required. Add a proposal, base papers and datasets if you have them:…
Guide · 2 min readHow to read your novelty score
The novelty check scores your concept, proposal, base papers and a sample experimental result out of…
FeatureNovelty check: is the idea clear, feasible and new enough?
EasySeer reads your handwritten or typed concept, searches papers, scores the idea out of 100 agains…
Sources
Checked on 2026-10-24. Policies change, so read the current version before you submit.
Published 2026-10-24. Last updated 2026-10-24.