We Tested 5 Chat Widgets with Google Lighthouse — Here’s What Happened
Real Lighthouse benchmarks of popular chat widgets. We measured performance scores, Total Blocking Time, unused JavaScript, and third-party cookies.
Every chat widget promises to be "lightweight." We decided to actually test that claim.
Your chat widget runs on every page of your site. If it's slow, everything is slow.
Google uses Core Web Vitals as a ranking signal — a bloated widget doesn't just hurt user experience, it hurts your SEO. We ran Google Lighthouse on five popular chat widgets to get real numbers.
How we tested
Simple and reproducible — anyone can verify:
Baseline: A minimal HTML page with no scripts. Scored 98 on Lighthouse.
One widget at a time: Each widget was added using its official embed snippet. No other scripts present.
Google Lighthouse via Chrome DevTools: Mobile throttling enabled — simulated Slow 4G, Moto G Power emulation.
Metrics tracked: Performance Score, Total Blocking Time (TBT), unused JavaScript, long main-thread tasks, third-party cookies, and Best Practices score.
Mobile throttling is key. Most audits run on fast desktops, which hides the real cost of heavy JavaScript. Simulated 4G shows what your actual mobile visitors experience.
The results
| Widget | Perf Score | TBT | Unused JS | Long Tasks | Cookies |
|---|---|---|---|---|---|
| GhostChat | 100 | 0 ms | — | 1 | 0 |
| Chatway | 100 | 40 ms | — | 2 | 0 |
| Crisp | 100 | 30 ms | — | 2 | 0 |
| Tidio | 99 | 100 ms | 182 KB | 8 | 0 |
| Tawk.to | 92 | 340 ms | 97 KB | 6 | 0 |
Good news on the privacy front: all five widgets scored 100 on Best Practices with zero third-party cookies. The real differentiator now is raw performance — Total Blocking Time and long main-thread tasks.
What the numbers mean
Tawk.to: the heaviest
Scored 92 — the lowest in our test. Blocked the main thread for 340ms across 6 long tasks. No cookies, but 340ms of blocking time on a blank page is significant. On a real site with existing scripts, images, and fonts, that impact compounds. A widget adding 340ms of blocking time can push a borderline site from "good" to "needs improvement" in Core Web Vitals.
Tidio: decent score, most long tasks
Scored 99 — respectable. But under the hood: 182 KB of unused JavaScript and 8 long main-thread tasks, the most of any widget tested. 100ms of TBT isn't terrible on its own, but 8 long tasks mean the main thread is being interrupted repeatedly. On a real site with existing JavaScript, that adds up.
Chatway: close, but GhostChat edges ahead
Scored 100 — matching GhostChat on the headline number. But the details matter: 40ms TBT and 2 long tasks vs GhostChat's 0ms TBT and 1 long task. On a blank test page that gap is small. On a real site it compounds. One of our users switched from Chatway specifically because it was "very heavy on our site and definitely slowed page load times."
Crisp: genuinely lightweight
Credit where it's due — Crisp performed well. 100 Performance, 30ms TBT, no cookies. It's among the lightest competitors tested. The difference vs GhostChat comes down to 30ms of blocking time (vs our 0ms) and what happens beyond the benchmark: Crisp's paid plans start at €45/mo, and enabling add-ons increases the JavaScript payload. See our full Crisp comparison.
The hidden cost: setup complexity
Lighthouse measures what happens after the widget loads. It doesn't measure how long it takes to get there. During our testing, we signed up for every widget to get their embed code. The difference was striking:
GhostChat: One script tag. Paste before </body>. Done. No account required to test.
Tawk.to: Multi-step onboarding wizard — organization name, website URL, widget customization screens before you get the embed code.
Tidio: 6-step onboarding checklist including connecting Instagram, Facebook, WhatsApp, installing desktop/mobile apps, and feeding an AI agent before you can start chatting.
Chatway: Quick signup, but requires creating a workspace and configuring the widget before you get the embed code.
Crisp: Relatively quick signup, but still requires account creation and workspace setup before you get the code.
If you just want to talk to your website visitors, most of these tools ask you to set up an entire support infrastructure first. GhostChat gives you a script tag and gets out of the way.
What this means for your site
Every 100ms of TBT costs conversions
Google's research shows that mobile pages losing 100ms of responsiveness see measurable drops in engagement. Tawk.to adds 340ms of blocking time — that's real visitors bouncing.
Core Web Vitals affect rankings
Since the Page Experience update, sites with poor CWV scores can lose positions to faster competitors. Your chat widget is part of that equation.
Long tasks add up
Tidio triggered 8 long main-thread tasks, Tawk.to triggered 6. Each one delays user interaction. GhostChat triggered just 1 — with 0ms blocking time, your visitors never notice it.
Mobile users feel it most
Our test simulated 4G — a best-case mobile scenario. On 3G or congested networks, heavy JavaScript payloads are devastating.
Why GhostChat scored a perfect 100
GhostChat scored 100 with 0ms Total Blocking Time. That's not an accident — it's the result of building light from the start:
~10 KB total, deferred loading. The widget loads with defer and doesn't block your page's critical rendering path.
No external dependencies. No React, no jQuery, no UI framework. Pure, hand-optimized JavaScript.
Single WebSocket connection. No HTTP polling, no beacon requests. One connection, minimal overhead.
Zero cookies, zero tracking. Nothing to consent to. No cookie banners required. Best Practices stays at 100.
Zero blocking time, zero cookies, perfect score. See exactly how our widget works — the source code is fully open. Want to see how your current chat widget affects your site? Run a free speed test.
Try GhostChat free
One script tag. Zero performance impact. Zero cookies. Set up in 30 seconds.
More from the Blog
Zero Blocking Time: Building a Widget That Won't Tank Your Lighthouse Score
Most chat widgets fail on slow networks. Learn how a lightweight chat widget delivers instant messaging even on 3G and congested connections.
The Micro SaaS Founder's Customer Support Problem (And How We Solved It)
You can spin up a SaaS in a weekend. But how do you support customers across all of them? Why traditional support tools fail indie makers.
Best Tools to Reduce Customer Support Response Time in 2026
Simple strategies to reply faster without hiring more agents. From Gmail threading to canned responses, here is how to speed up your workflow.