HAR vs JMeter — Gap Analyser

Use HAR to audit whether your JMeter script covers what the browser actually loads. Don't compare the numbers. Compare the coverage

---

The question this tool answers

JMeter says your API responds in 350ms. The browser says the same call took 500ms. Which one do you trust?

Both — they're measuring different things:

	JMeter	HAR
Measures	How long each scripted request took	How long the user actually waited
Sees	What you explicitly scripted	Everything the browser fired — including 3rd party scripts, lazy loads, render-triggered calls
When slower	Backend under concurrent load — DB contention, cold caches, GC pauses	Browser overhead — DNS, SSL, connection limits, real network

Passing your JMeter test doesn't mean your users are happy. This tool bridges the gap.

---

Quick start (3 steps)

Step 1 — Export a HAR file from Chrome/Firefox

• Open DevTools → Network tab
• Navigate through your app 2–3 times (warm cache = realistic returning user)
• Right-click any request → Save all as HAR with content

Step 2 — Export a JMeter Summary Report CSV

• Add a Summary Report listener to your test plan
• Run your test
• Right-click the Summary Report → Save Table Data → save as .csv

Step 3 — Run the comparison

python har_vs_jmeter.py your_file.har summary_report.csv report.html

Open report.html — done.

---

What the report shows

• 4 stat cards — HAR total load time, request count, JMeter samplers, matched count
• Request-level comparison table — HAR time vs JMeter avg, diff, dual bar chart, match score
• Unmatched HAR requests — 3rd party scripts, fonts, ad networks the browser loaded that JMeter never saw
• Unmatched JMeter samplers — samplers with no HAR equivalent
• "Why the diff?" reference — built into every report, explains whether browser overhead or backend load is the likely cause

---

Requirements

• Python 3.7+ (standard library only — no pip install needed)
• A .har file exported from Chrome or Firefox DevTools
• A JMeter Summary Report CSV (exported from the Summary Report listener)

---

URL Matching

HAR requests are matched to JMeter sampler labels by URL path similarity.

HAR URL	JMeter Label	Match
/api/session	/api/session	✅ Exact (score 100)
/api/session	/api/session-1	✅ Recorder suffix auto-stripped
/SearchHotel.php	/SearchHotel.php-85	✅ Recorder counter stripped
/api/session (×2 in HAR)	/api/session	✅ Both HAR hits matched (many-to-one)
ads.example.net/ad.js	—	→ Unmatched HAR (3rd party)

---

Sample files

The examples/ folder has a ready-to-use demo set:

File	Description
demo.har	10 browser requests — 3 APIs + static assets + a slow 900ms 3rd-party ad script
demo_summary.csv	JMeter Summary Report for the 3 API samplers
demo_test.jmx	Minimal JMeter test plan matching the HAR

Try it:

python har_vs_jmeter.py examples/demo.har examples/demo_summary.csv examples/demo_report.html

---

Why the diff? (built into every report)

HAR > JMeter — the browser pays a tax JMeter doesn't

• DNS + SSL/TLS handshake (JMeter reuses keep-alive connections)
• Render-blocking dependencies
• Per-host connection limits (~6 concurrent)
• Real network conditions — latency, packet loss on the user's actual connection
• 3rd party scripts you never scripted

JMeter > HAR — the backend pays a tax HAR doesn't

• Concurrent load — many virtual users hammering the backend simultaneously
• DB connection pool exhaustion
• Lock contention — concurrent writes cause locks a single session never triggers
• Cold vs warm caches under load
• GC pauses / CPU saturation

---

JMeter Listener plugin (optional)

A JMeter GUI Listener plugin is also available in Releases for those who prefer a UI wrapper inside JMeter.

Note: The standalone Python script above is the recommended approach. The plugin wraps the same Python engine — the standalone script is simpler and more reliable for most workflows.

---

Project structure

HAR_vs_Jmeter/
├── har_vs_jmeter.py        ← main Python script (run this)
├── README.md
├── examples/
│   ├── demo.har
│   ├── demo_summary.csv
│   └── demo_test.jmx
└── plugin/                 ← optional JMeter GUI plugin
    ├── pom.xml
    └── src/

---

Built by

Visvashwarr Venugopal — Performance Engineer, 11+ years across Oracle and TCS. Now freelancing.

• 🔗 LinkedIn: linkedin.com/in/visvashwarrvenugopal

---

License

MIT — free to use, modify, and distribute. Attribution appreciated.