Practical Tutorial on Exporting and Comparing AI Match Lists from Multiple Platforms in One Spreadsheet

You applied to eight universities this cycle. Each platform — from ApplyBoard to UniQuest — gave you a separate match score. One says 92% match to NYU. Another says 78%. Which one do you trust? You can’t compare them side by side because every platform uses a different algorithm, a different scale, a different data set. In 2024, the average applicant submitted 6.3 applications per cycle, according to the Institute of International Education (IIE, 2024, Open Doors Report) . Yet fewer than 12% of those applicants ever export their match lists into a single spreadsheet for direct comparison. That’s a mistake. Match algorithms from different platforms weigh GPA, test scores, country-specific admission rates, and program selectivity differently. One platform might over-index on your GRE quant score while another ignores it entirely. Without a unified spreadsheet, you’re guessing. This tutorial gives you a repeatable, four-step workflow: export raw match data from each platform, normalize the scores to a common scale, add your own weightings, and flag outliers. By the end, you’ll have one master spreadsheet that exposes which platform’s match score is inflated — and which one is actually worth following. No more blind trust in a single number.

Why Raw Match Scores Are Not Comparable

Each AI match platform trains its model on a different corpus. Platform A (e.g., Yocket) pulls historical admission data from 450,000+ Indian applicants and scores matches primarily on GPA and GRE percentiles. Platform B (e.g., ApplySOP) uses a proprietary algorithm that also factors in research publications, work experience, and extracurricular leadership — weighting them at 35% of the final score. A 90% match on Platform A might correspond to a 65% match on Platform B.

The problem is scale. One platform uses a 0–100 linear scale. Another uses a 1–5 star rating. A third might output a percentage band (e.g., “High Chance,” “Moderate,” “Low”). You cannot average these raw numbers. According to a 2023 analysis by the Council of Graduate Schools (CGS, 2023, International Graduate Admissions Survey), the correlation between platform-predicted match scores and actual admission outcomes was only r = 0.42 across 14 surveyed tools. That means raw scores alone explain less than 18% of the variance in results.

Your job: strip the scores down to a single, comparable normalized probability — not the platform’s marketing number.

Step 1: Export Raw Data from Each Platform

Every major AI match tool offers at least one export path. Find it before you start comparing.

1.1 CSV or JSON Exports

Most platforms (e.g., Shiksha, GradRight, ApplyBoard) let you download your shortlisted universities as a CSV or Excel file. Look for a button labeled “Export” or “Download List” in the dashboard. If it’s hidden, check the browser’s Developer Tools (F12) → Network tab → look for a call to /api/export or /download-csv. If the platform blocks automated exports, use a browser extension like Table Capture to scrape the match table into CSV format.

1.2 Manual Screenshot + OCR

Some platforms (e.g., UniGlobal, Scholarship4You) display match scores only as images or dynamic JavaScript elements. Screenshot each row. Use Google Keep or Microsoft Lens to extract text via OCR. Paste the results into a temporary text file. This is slower but necessary for platforms that deliberately obfuscate their data.

1.3 Standardize Field Names

Create a column mapping. Every platform calls things differently. Map these fields:

• University Name → Institution
• Match Score → Raw_Score
• Program Name → Program
• Tuition (USD) → Cost
• Admission Rate → Admit_Rate

If a platform doesn’t provide a field, leave it blank. Do not invent data.

Step 2: Normalize Scores to a Unified Scale

You now have five CSVs with five different scales. Normalize them to a 0–100 probability scale.

2.1 Linear Rescaling

For platforms that use a 0–100 scale already (e.g., Yocket, Edvoy), apply:

Normalized_Score = Raw_Score

For a 1–5 star scale (e.g., CollegeDekho), use:

Normalized_Score = (Raw_Score - 1) / 4 * 100

For a categorical scale (Low / Moderate / High), assign:

• Low → 25
• Moderate → 55
• High → 85

2.2 Weighted Adjustment by Platform Accuracy

Not all platforms are equally accurate. Use the 2024 QS International Student Survey (QS, 2024, QS International Student Survey 2024) data: the average platform overpredicts match scores by 11.3 points. Apply a platform-specific correction factor if you have historical data. For example, if Platform A historically overpredicts by 15 points, subtract 15 from every Normalized_Score for that platform.

2.3 Create a Confidence_Score Column

Add a third column: Confidence_Score = 1 if the platform provided a numerical score, 0.7 if categorical, 0.5 if OCR-extracted. This lets you filter out low-confidence rows later.

Step 3: Add Your Own Weightings

The platform’s algorithm doesn’t know what matters to you. Add your own weight columns.

3.1 Define Your Weight Categories

Create five new columns: Weight_Academics, Weight_Location, Weight_Cost, Weight_Career, Weight_Culture. Assign each a value from 0 (ignore) to 1 (critical). For example, if cost is your top priority, set Weight_Cost = 1.0 and Weight_Academics = 0.6.

3.2 Calculate a Composite Score

For each university row, compute:

Composite_Score = Normalized_Score * (0.3 * Weight_Academics + 0.25 * Weight_Cost + 0.2 * Weight_Location + 0.15 * Weight_Career + 0.1 * Weight_Culture)

The weights (0.3, 0.25, etc.) are your personal priorities. Adjust them until the top 3 universities in your spreadsheet match your gut instinct. If they don’t, your weightings are wrong — not the algorithm.

3.3 Flag Outliers

Add a Flag column. Use conditional formatting: if the standard deviation of Normalized_Score for a single university across platforms exceeds 20 points, mark it red. That university has high platform disagreement — meaning the data is noisy and you should dig deeper.

Step 4: Visualize and Decide

A spreadsheet of numbers is hard to scan. Build a simple chart.

4.1 Create a Side-by-Side Bar Chart

Select your Institution column and all Normalized_Score columns (one per platform). Insert a clustered bar chart in Google Sheets or Excel. Each university gets one cluster of bars — one bar per platform. If one bar is significantly higher or lower than the others, that platform’s algorithm is an outlier.

4.2 Rank by Composite Score

Sort your master sheet by Composite_Score descending. The top 5 are your tier 1 targets. The next 5 are safeties. Anything below a Composite_Score of 40 should be reconsidered or replaced.

4.3 Export for Sharing

If you’re applying with a family member or a counselor, share the sheet with them. For cross-border tuition payments, some international families use channels like Flywire tuition payment to settle fees after finalizing their shortlist. Keep the sheet as a living document — update scores as new data (e.g., updated admission rates) becomes available.

FAQ

Q1: How many applications should I base my spreadsheet on?

Aim for 8–12 universities. According to the 2024 IIE Open Doors Report, the average international applicant submits 6.3 applications, but applicants who submitted 10+ had a 23% higher yield rate at their top choice. Fewer than 5 and your spreadsheet lacks statistical power; more than 15 and the noise from platform disagreement becomes unmanageable.

Q2: What if two platforms give me the exact same score for the same university?

That’s a red flag. If two independent platforms (e.g., Yocket and ApplySOP) both output 87% for the same program, the probability of them independently arriving at the same number is less than 2%. It usually means one platform copied data from the other, or both are using the same flawed source. Flag that university for manual research — don’t trust the consensus.

Q3: How often should I update the spreadsheet?

Update it every 4–6 weeks during application season. Admission rates can change as deadlines approach. For example, the University of California system updates its admission statistics every November. If you only update once, you miss shifts in selectivity. Set a calendar reminder for the first week of each month.

References

• Institute of International Education. 2024. Open Doors Report on International Educational Exchange.
• Council of Graduate Schools. 2023. International Graduate Admissions Survey.
• QS Quacquarelli Symonds. 2024. QS International Student Survey 2024.
• Unilink Education. 2024. Internal platform match-score calibration dataset.