What do free online file tools actually load?
Before we built ToolWow, we wanted to know what a typical "free online converter" actually ships to your browser when you open it. On 2026-10-24 we measured the first-load HTML of 15 popular free converter sites โ the exact bytes the server sends before any of your files move and before any JavaScript runs. This page is the full data, the method, and what we can honestly conclude from it.
The data
| Site (page tested) | HTML size | External scripts | Inline scripts | 3rd-party domains | Analytics | Ad-tech | Consent UI |
|---|---|---|---|---|---|---|---|
| ilovepdf | 53 KB | 2 | 9 | 1 | Yes | Yes | Yes |
| smallpdf | 422 KB | 1 | 40 | 2 | Yes | โ | Yes |
| convertio | 129 KB | 2 | 1 | 6 | Yes | โ | โ |
| cloudconvert | 48 KB | 1 | 1 | 0 | โ | โ | โ |
| ezgif | 4 KB | 0 | 0 | 0 | โ | โ | โ |
| online-convert | 245 KB | 14 | 13 | 1 | Yes | โ | Yes |
| pdf2go | 420 KB | 39 | 18 | 4 | Yes | Yes | Yes |
| jpg2pdf | 18 KB | 5 | 6 | 1 | โ | Yes | โ |
| sejda | 94 KB | 2 | 0 | 0 | โ | โ | โ |
| tinywow | 1,012 KB | 9 | 12 | 7 | Yes | Yes | โ |
| freeconvert | 374 KB | 64 | 9 | 3 | Yes | โ | โ |
| zamzar | 313 KB | 6 | 2 | 3 | Yes | โ | โ |
| aconvert | 16 KB | 4 | 2 | 2 | โ | Yes | โ |
| 123apps | 143 KB | 5 | 3 | 7 | Yes | Yes | Yes |
| pdf24 | 70 KB | 6 | 14 | 1 | Yes | โ | Yes |
| ourpdf | unreachable at measurement time (network error) | ||||||
| imgonline | unreachable at measurement time (404) | ||||||
โฌ Download the raw data (CSV) โ it is free to use and cite
"Analytics" = Google Analytics (gtag.js) present in first-load HTML. "Ad-tech" = at least one well-known advertising network domain (Google ad services, DoubleClick, PubMatic, Criteo, Taboola, Outbrain, Index Exchange, Amazon's ad system) referenced in the first-load HTML. "Consent UI" = a cookie/GDPR consent element present in first-load HTML. Two sites were unreachable from our test network at measurement time and are reported as such rather than estimated.
How we measured โ and what we did not
- One GET per page, desktop user agent, no cookies, no interaction, no JavaScript execution. What you see above is what the server sent first.
- First-load HTML only. Sites frequently inject additional scripts after JavaScript runs, so the real runtime load of these pages is almost certainly higher than this table. We deliberately did not run a headless browser: this is a reproducible baseline, not a runtime audit.
- One day, one network. Results can shift with A/B tests, geolocation and time. Re-measure before citing exact numbers; the method is simple enough to repeat.
- Presence, not behaviour. A domain in the HTML means it is referenced โ not what it does with your data. We make no claim about any site's data practices.
- One representative page per site โ a converter or merge page, or the homepage where that was the closest equivalent.
What the numbers actually say
- Analytics are the norm, consent is not. 10 of 15 pages ship Google Analytics in the first HTML, but only 6 include a consent UI in that same first HTML. Where analytics run before the user has seen โ or accepted โ anything, that is a design decision visible in the table.
- First-load HTML understates the final page. One site sent a single external script yet 422 KB of HTML and 40 inline script blocks; the heaviest first load was 1.0 MB of HTML before a byte of application logic ran. If you care about what runs when you open a file tool, the first HTML is the floor, not the ceiling.
- Lightweight is achievable. Three sites in the table shipped under 20 KB of HTML before JavaScript ran, and at least one of them (ezgif) did so with no analytics and no ad-tech markers in that first load. The category is not forced to be heavy.
- This is a snapshot, not a ranking. We chose sites because people actually use them. No site was chosen to look good or bad, and none has been contacted about this post.
What this means if you use online file tools
Your file is one thing; the scripts the page loads are another. Before uploading a document you would not want to hand to a stranger, open DevTools โ Network, click Convert, and watch what goes where. If a tool processes your file on a server, the only way to know what happens to it is that site's privacy policy โ a document, not a guarantee. The strongest version of "I don't want this file to go anywhere" is a tool with nowhere to send it: ToolWow's converters and tools run entirely in your browser, and you can verify it yourself on any tool page by watching the Network tab โ there are no upload requests because there is no server to upload to.
Measured 2026-10-24 ยท method above ยท data released as CSV for anyone to verify or re-run. Questions or corrections: contact us.