Most chat installations put a floating button in a corner and stop there. It works, and for a lot of sites it is all that is needed. But the corner is also the part of the page people have trained themselves to ignore, and it is nowhere near the sentence where the visitor actually got stuck.
A text launcher solves that. It turns an ordinary piece of text into something that opens chat, so the invitation can sit inside a paragraph, in the site footer, or in a navigation menu, right where the question forms.
The first thing to be clear about is that this is an additional way to open the chat you already have, not a second chat system. Visitor monitoring keeps running, proactive invitations keep working, and the conversation itself is the same conversation with the same history.
You can therefore add a text link without giving anything up, and without deciding between it and the floating button. Most sites that use text links keep the button as well, and the two serve different moments: the button for a visitor who has decided they need help, the link for one who has just hit the sentence that confused them.
Two snippets, because there are two states
The setup screen gives you two separate blocks of HTML rather than one: an online code and an offline code. Whichever applies is what a visitor sees, depending on whether anyone is available.
This is more useful than it first appears. A single piece of link text has to work in both worlds, and inevitably ends up vague. Two pieces do not. The online link can say something direct that promises an immediate answer. The offline link can say something honest that sets a different expectation, such as leaving a message, so nobody clicks expecting a person and finds a form.
Both are ordinary HTML, so the link can be styled to match the surrounding text, and there is no requirement for the two versions to look alike.
Deciding what offline should look like
There is a separate setting that controls whether the launcher appears at all when nobody is available. Hiding it and showing an honest offline version are both defensible, and the right answer depends on where the link sits.
In the middle of an article, hiding it is usually better. A visitor reading a paragraph does not need to be told about a channel they cannot use right now, and an offline link inside body copy reads as a dead end.
In a footer or a contact page, showing it is usually better. Someone who has navigated to the footer is looking for a way to reach you, and telling them they can leave a message is a genuine answer to the question they are asking. An empty space is not.
Each state also has its own tooltip text. These are easy to skip and worth a minute, because a tooltip is where you can be specific without cluttering the link itself.
A link that reads ask us about delivery can carry a tooltip that says a person usually replies within a couple of minutes. The offline tooltip can say when you are next open. Keep both short, and make sure the offline one never promises a speed you cannot meet while unstaffed. Setting the expectation in the tooltip is much cheaper than apologising for it in the conversation.
The launcher can either reuse the embedded chat already on the page or open chat in a separate popup window, and the screen states which behaviour is currently selected.
Reusing the embedded chat is the better default for nearly every site. The visitor stays on the page they were reading, which matters when the whole point of the link is that it sits next to the thing they were confused about. Sending them to a separate window takes the context away at the exact moment it is most useful.
The popup earns its place in narrower cases: a page where an embedded panel would fight with the layout, or a flow where you would rather the conversation survive the visitor navigating around. Choose it deliberately rather than by accident, and check the result on a phone, where a second window is a heavier interruption than it is on a desktop.
Changing the wording per page
The link text and tooltips can also be supplied through the script query string rather than fixed in the settings, which lets the same installation say different things in different places.
That is what makes a text launcher genuinely better than a generic button. A pricing page can offer to answer a question about plans, a delivery page can offer to check a shipping time, and a documentation page can offer to find the right article. The closer the wording sits to the visitor's actual question, the more the link gets used.
Keep the language consistent in tone across every entry point you publish, even as the specifics change. The screen itself makes this point, and it is a good one: a visitor who sees three different voices in three places is being told, quietly, that the company is not joined up.
How it differs from the no-JavaScript link
There is a separate feature for sites and pages where JavaScript cannot run, and the two are easy to confuse because both are described as links.
The text launcher is JavaScript. It knows whether you are online, it can open chat in place, and it participates in monitoring and proactive invitations. The no-JavaScript link is a plain hyperlink to a hosted chat page that works in environments where none of that is possible, and it gives up the state awareness in exchange.
Use the text launcher on your normal pages, and reach for the no-JavaScript link only where scripts genuinely cannot run, such as an email or a third-party page you do not control.
Where a text link earns its place
Four placements repay the effort. At the end of a help or documentation article, where the reader has just failed to find their answer. On a pricing page, next to the row people always ask about. In the site footer, where visitors go when they are hunting for a way to make contact. And inside a checkout or form flow, at the step your analytics show people abandon.
Resist the temptation to put one everywhere. A link that appears in every paragraph stops reading as help and starts reading as an advert, and the floating button already covers the general case.
What to measure
Compare the chats started from text links against those started from the button, and look at where they came from. A link that produces almost no chats is in the wrong place, or its wording is too generic to answer a question anybody was asking.
Then look at quality rather than quantity. Chats opened from a well-placed link usually start with a sharper question, because the visitor clicked at the moment they got stuck. If those conversations are shorter and resolve more often than the ones from the corner button, the placement is doing real work, and it is worth finding two or three more pages like it.
Also watch the offline path. If a meaningful number of clicks land on the offline version, that is a staffing signal about your hours, not a reason to hide the link.