Text is the default for good reasons
Typed chat is searchable, asynchronous, easy to multitask, and leaves a transcript. It is the right default for the overwhelming majority of conversations. Voice, video, and co-browsing are escalations — powerful when the medium is the bottleneck, wasteful when it is not.
Reach for voice when typing is the slow part
Some things are simply faster spoken: an emotional complaint that needs a human tone, a complex back-and-forth where each typed round trip costs a minute, an accessibility need. Offering to switch to a quick call can turn a frustrating twenty-message thread into a two-minute resolution. Offer it; do not force it — plenty of visitors chose chat precisely to avoid a phone call.
Video for the moments a face matters
Video earns its place rarely but decisively: a high-value onboarding, a sales conversation where trust is the product, a demo that is easier shown than described. Because it demands the visitor's full attention and a presentable setting, treat it as opt-in and purposeful, never a default pop-up.
The single most maddening support pattern is guiding someone through a screen blind — "click the gear, no, the other gear." Co-browsing or a shared screen collapses that into "I see it; click there." It is the fastest path through navigation confusion, form errors, and setup walkthroughs, where describing the screen in words is the whole difficulty.
Escalate with consent and a clear exit
- Ask, don't ambush. "Want to hop on a quick call to sort this faster?" respects the visitor who chose chat on purpose.
- Keep the transcript. Whatever the medium, the conversation record and context should survive the switch.
- Make return easy. Some visitors will decline; the text chat should carry on without penalty.
The discipline is simple: escalate the medium only when the medium is what's slowing the resolution. MyLiveChat supports voice and video calling and co-browsing alongside text, so an agent can lift a stuck conversation into the right channel in the moment — without asking the visitor to install anything or start over.
Check these before you offer a richer channel
Escalating from text to voice, video or co-browsing is a bigger ask of the visitor than it
feels like to the agent, and offering it when it cannot work well is worse than not offering it.
A few checks make the difference between a rescue and an embarrassment.
Consider where the visitor is. Someone on a phone on a train cannot take a video call and may
not be able to take a voice call either; someone at a work desk may be unable to speak at all.
Asking “are you somewhere you can talk?” before initiating anything costs one line
and prevents the most common failure. Consider the device too — co-browsing a desktop
layout with a visitor on a small screen often makes the confusion worse rather than better.
Consider whether the agent is ready. A voice call started before the agent has read the
history means the visitor repeats themselves out loud, which is more irritating than repeating
themselves in text. Take the extra thirty seconds to read the thread first; the call will be
shorter for it.
And consider whether it is genuinely faster. If the answer is a link, a setting name or a short
instruction, text wins because the visitor can re-read it afterwards. The richer channels earn
their cost when the problem is spatial, visual or emotionally charged, and not otherwise.
Keep the written record
The hidden cost of moving to voice or video is that the transcript stops. Everything agreed on
a call is invisible to the next agent, absent from any later review, and unavailable to the
visitor who wants to check what was said. That gap causes real problems weeks later, when nobody
can establish what was promised.
The fix is a habit rather than a tool: before ending the richer session, return to the chat and
write a short summary of what was decided. Three lines is enough — what the problem turned
out to be, what was done, and what happens next. The visitor gets something to refer to, the
record stays continuous, and the next agent is not starting blind.
Handle consent explicitly and in plain language. Say what will happen, say whether anything is
being recorded, and make declining easy and unembarrassing. “Happy to stay in chat if you
prefer” should always be on the table. A visitor who feels pushed into a video call
remembers that far longer than they remember the fix, and the ones most likely to feel pushed are
the ones least comfortable saying no.
What to measure
Track how often a richer channel is offered against how often it is accepted. A low acceptance
rate is not a failure of the feature — it usually means it is being offered reflexively
rather than at the moments it helps, and the transcripts will show which.
Compare resolution on escalated conversations against similar text-only ones. The honest
finding is sometimes that the escalation did not help, which is worth knowing before you build
process around it. And watch how many escalated sessions end without a written summary in the
transcript; that number is a direct measure of how much institutional memory the channel is
quietly costing you.