How much does a live chat widget slow down your site?

Why this number matters more than it looks

A chat widget is the last thing on the page and the first thing people forget to measure. It loads after your content, so it rarely shows up in the number everyone watches, and it is usually installed by pasting a snippet somebody sent over Slack. Nobody owns it and nobody checks it.

The cost lands somewhere specific: on a mid range Android phone over mobile data, which is exactly the visitor least likely to wait. Half a megabyte of JavaScript is a second or two of nothing happening, and the visitor does not attribute that to your chat widget. They attribute it to your site.

Weight is not a ranking factor on its own. What it does is push Largest Contentful Paint and Total Blocking Time in the wrong direction, and those are. If you are buying traffic to a landing page, you are paying for the delay twice.

Download size is not the same as slowness

Bytes have to arrive, and then they have to be parsed and executed, and the second part happens on the visitor device rather than on the network. That is why the ordering changes depending on which number you look at.

Zendesk downloads less than Tawk.to and costs the visitor more, because Tawk.to's bulk is a picture and Zendesk's is code. A picture is cheap to receive and free to ignore. Code has to run.

How we measured our own widget

The TGLiveChat figure is the only one on this page we produced ourselves, and here is exactly how, so you can disagree with it precisely rather than in general.

Reproduce the core figure with one command:

curl -s https://cdn.tglivechat.com/index.mjs | gzip -9 -c | wc -c

To measure any widget on any site, including ours on yours, open DevTools, go to Network, filter by the vendor domain and hard reload. Read the transfer column rather than the resource size column, because transfer is what crossed the wire after compression. Then look at the Performance tab for scripting time. That is the whole method, and it takes about two minutes.

What to do if your widget is heavy

You have three options and they are not equally good. The first is to defer it: load the widget on first scroll or behind a placeholder button rather than on page load. That works with any vendor and it is the biggest single improvement available, at the cost of the chat not existing for the first second.

The second is to check whether your vendor already defers. Zoho SalesIQ and LiveAgent both hold most of their code back until somebody opens the chat, which is why they measure well despite being full products. If yours does this, you may have no problem at all.

The third is to change vendor, which is worth it only if the weight is buying you nothing. A 300KB widget attached to a helpdesk your team lives in is a reasonable trade. A 749KB widget where 500KB is an unused image is not.