Part 2: Checklist

Tick every box before you submit. An unticked box is something a judge will notice.

• I know if my project is Research or Engineering design.
• My category and subcategory match my main contribution.
• I have at least 3 real sources that I have opened and read.

• My abstract scores 12 or more out of 16 in the Abstract tab.
• My problem is backed by a number or a cited source.
• My objectives have numbers and units.
• I name the closest existing study and what mine does differently.
• My method lists variables, materials, sample size and number of trials.
• My results include at least one table or chart of real data, or I state honestly what is not done yet.
• I state at least two real limitations.
• My plan separates what is done from what is planned, with dates.

• Every section follows the rules in the Format tab.
• The numbers in my abstract match my results exactly.
• Every [number] in the text has a matching reference in Section 9.
• AI use is disclosed in Section 9, or I state that none was used.
• I can explain every sentence, method and reference without notes.
• Submitted before 28 October, not on the last evening.

Part 3: The abstract

The abstract is Section 1 and the first thing every judge reads. It must be 100 to 300 words in 1 or 2 paragraphs, and a judge should understand your whole project from it alone. Write it last.

These are the four parts the official form asks for. Winning abstracts usually split them into smaller moves, shown in the next column.

We analysed 5,427 ISEF finalist abstracts from 2022, 2023, 2024 and 2026. Every one reached ISEF, the global final, so these are shared habits of some of the strongest student work in the world. The bars show how common each habit is across them.

How common each habit is

Each bar is the share of these ISEF-level abstracts that include the habit. Aim to include them all; the rarer ones, like stating sample size and trials, are where you most easily stand out.

These are patterns from past ISEF abstracts, not guarantees. They describe what strong work tends to share, so treat them as habits to include, not boxes that win on their own.

The habit that most sets the very top apart: stating a specific new contribution. Only 20% of abstracts include it, but 43% of those that do won a Grand Award, against a 26% average. In one sentence, name what your project does that the closest existing work does not.

Length

Aim for about 230 to 250 words. Strong abstracts cluster there (the middle value is 241 words), and padding past 250 does not help.

We scored a blind sample of 181 abstracts on the 8-point rubric below, without knowing which had won. Abstracts that scored high won far more often, which means the rubric measures something real.

Winners averaged 10.6 out of 16; other finalists 9.2. Against a 26% average win rate, abstracts in the top band won 71% of the time, nearly three times as often. Score yours below and aim for 12 or more.

1. Put a number on the problem in sentence one. 38% of winners quantify the problem early, against 29% of other finalists. A project on lithium treatment opens by stating it causes chronic kidney disease in at least 20% of patients.
2. Use several numbers, in the results above all. Winners use a median of 5 numbers, others 4; 79% of winners give results as numbers. A walking-aid project reports ten times lower collision risk for under $120.
3. Compare against a control or what already exists. 60% of winners name a comparison. A biodegradable battery project reports performance comparable to a lithium cell at over five times lower cost.
4. State your specific new contribution. Winners are twice as likely to name clearly what is new in their work. Say what yours does that the closest existing work does not, in one sentence.
5. Name the gap in one sentence. A physics project says earlier studies observed an effect but its mechanism was unclear. The aim then follows directly from that gap.

• Sample size and trials. Only 22% of winners and 17% of others state them, yet the Ibdaa form asks for both in Sections 6 and 7. One clause ("5 samples, 3 trials each") is an easy, rare win.
• Limitations. Just 29% of all finalists state one, winners and others alike. The Ibdaa form rewards it directly, so a single honest limit sets you apart.
• Hype without data. Avoid "revolutionize" or "cure" unless a number backs it. An honest "not done yet" scores better than a claim with no evidence.
• No stated problem. Some abstracts open straight with the method. Always say what problem you solve and for whom.

Use these to plan, then write every sentence in your own words. Judges will interview you on it.

1. [Specific problem] affects [who]; [number or fact] [1].
2. However, current [solutions or studies] [specific shortcoming].
3. This project aimed to [measurable aim]. or We hypothesized that [prediction].
4. I [built, tested, measured] [what] using [materials or tools], with [n] samples and [n] trials, compared with [control or existing solution].
5. [Main result with a number] compared with [control]; [second result].
6. Although [one limitation], ...
7. This could help [specific people] by [specific change].

Both examples are invented to show the difference.

No specific problem, no method, no numbers, and a promise instead of a result.

Every part is there: a sourced problem, the gap, a measurable aim, a method with a control and trials, numbers, one limit, and a specific benefit.

Copy this prompt into an AI tool and paste your abstract at the end. It is set to give feedback only, not to rewrite your work.

You are a judge for Ibdaa 2027, the national science competition run by Mawhiba in Saudi Arabia. Review the student abstract at the end. Do NOT rewrite it and do NOT suggest replacement sentences. Give feedback only, so the student fixes it in their own words.

Rules: 100 to 300 words, 1 or 2 paragraphs. A judge must understand the whole project from the abstract alone. It must cover four parts:
1. Problem: a specific real-world problem, who is affected, and a number or cited source. Also the gap: what current solutions or studies miss.
2. What I did: one clear aim or hypothesis, then the method with materials or tools, sample size, number of trials, and what it was compared against.
3. What I found or built: concrete results with numbers, compared with a control or existing solution, plus one honest limitation. If data is not collected yet, an honest statement of what is complete.
4. Why it matters: specific beneficiaries or uses, not a general promise.

Reply in this format:
- Word count: [number]. Within 100 to 300: yes or no.
- For each of the 4 parts: Met, Partly met, or Missing. Quote the sentence that covers it, or say none does. Ask ONE question that would help the student improve it.
- Vague or hype phrases: list any claim without evidence (for example "very good results", "revolutionize", "cure").
- Top priority: the single most important fix, in one sentence.

Project type (Research or Engineering design): [WRITE HERE]
Category: [WRITE HERE]

Abstract:
[PASTE YOUR ABSTRACT HERE]

If you use AI for feedback, disclose it in Section 9, for example: "ChatGPT was used to give feedback on the abstract draft. All text was written by the student."

Part 4: Project sections

The form tells one story: the problem, what is known, the gap, your goal, what is new, how you did it, what you found, and what comes next. Open each section and tick it when your draft meets it.

Part 5: Format

1. Write in one language only, Arabic or English.
2. If a question has a Research and an Engineering design version, answer only the one that fits.
3. Put all text, tables, images and charts inside the answer boxes.
4. Keep each title to 20 words or fewer.
5. Stay within the word limits below.

Number sources in the order you first use them and put the number right after the claim:

Nitrate above safe levels reduces crop yield [1].

Use the same number every time you cite that source again.

Use one style for every entry in Section 9, numbered to match the text:

[1] Author, A., & Author, B. (Year). Title of the article. Journal Name, Volume(Issue), pages.
[2] Organization or Author. (Year). Title of the page or report. Website name. URL

• Give each a number and a title: Table 1: Nitrate concentration after filtration.
• Put units in every column header or axis label (mg/L, seconds, %).
• State the sample size and number of trials under it.

At the end of Section 9, name each AI tool and exactly what it did:

ChatGPT was used for language editing of Sections 2 and 3. All analysis and code are my own.

If you used no AI tools, say so.