A free-plan user is not a trial user
Almost everything written about converting non-paying users assumes a countdown. The trial ends on the fourteenth day, so there is a natural moment to check in, a natural reason to reach out, and a natural deadline that makes the conversation feel timely rather than pushy.
A free plan has none of that. The user can stay exactly where they are indefinitely, and many will — happily, permanently, and without ever costing you a thought. That removes the deadline that most playbooks are built around, and it changes what a chat conversation with them is for.
It also changes the arithmetic. A trial user is a short-term cost with a decision point attached. A free user is an ongoing cost with no decision point at all, which is why the question of what support they get deserves an actual answer rather than drift.
Decide what free support includes, and write it down
Most teams never decide this. They offer chat to everyone because that is how it was set up, then quietly resent the volume, and agents start making the policy themselves one conversation at a time — which produces the worst outcome, where the answer depends on who happened to pick up.
There are three defensible positions. Chat for everyone, on the view that free users are your funnel and your reputation. Chat for paying customers with self-service for everyone else, which is honest if the self-service is genuinely good. Or chat for everyone with a scope limit: anyone can ask about using the product, and account-specific work is for paid plans.
Any of the three works. What does not work is having no position, because then the cost is unpredictable and the experience is inconsistent, which is the combination that eventually makes someone propose removing chat entirely.
The conversations genuinely worth having
Without a clock, intent has to come from behaviour instead. The signals are not subtle, and they beat any calendar-based rule.
A free user who hits a limit is telling you they have outgrown the plan — that is the moment, not a month later. A user inviting teammates is describing a bigger use case out loud. Someone asking whether a paid feature does a specific thing has already imagined using it. And a user asking about security review or invoicing is not exploring; somebody has asked them to justify a purchase.
Those four are worth a real conversation. A free user asking how to format a date is worth a good answer and nothing more, and treating every free conversation as a sales opportunity is how a support channel acquires a reputation nobody wants.
Do not punish the free user in chat
It is tempting to make free support deliberately worse so that the paid tier looks better. It works poorly, for two reasons that are easy to underestimate.
The first is that free users are frequently not the buyer. They are the developer evaluating you, the assistant testing three tools, the person who will recommend one of them in a meeting you are not invited to. Being brisk with them is a decision made about someone whose influence you cannot see.
The second is that public writing exists. A curt chat is a screenshot, and it will be captioned in a way you would not enjoy. Being straightforwardly helpful within a stated boundary costs very little and reads completely differently from being grudging.
If a request is genuinely outside what free includes, say so plainly and point at what is available. That is not rudeness; it is clarity, and people take it well when the boundary is stated as policy rather than implied as reluctance.
Asking for the upgrade without a deadline
With no expiry to lean on, the only honest hook is the limit the user has actually met. So talk about the limit, not the plan. You have hit the ceiling on X, here is what happens on the tier above, would you like me to explain how the billing works is a conversation about their problem. Have you considered upgrading is a conversation about your revenue, and people hear the difference immediately.
Make the next step the smallest one that is genuinely useful. Sometimes that is upgrading; often it is a walkthrough of the feature they are curious about, or an answer to the security question their manager raised. The chat conversation rarely closes the sale on its own, and it very often removes the one obstacle that was quietly blocking it.
Then log what actually happened, because the free-plan population is the one where teams are most likely to guess. Our guide to measuring chat conversion rate covers how to attribute this without kidding yourself.
When the answer is that they will never upgrade
Plenty of free users will never pay you anything, and that is a normal outcome rather than a failure. The goal for that population is not conversion; it is keeping the cost of serving them proportionate.
That is a self-service problem, and it is the one place where the effort really pays. Every routine question your help content and AI answers correctly is a conversation that never needed a person, which means the free tier stops competing with your paying customers for attention. The point is not to hide from free users but to make the common answers reachable without a queue.
It also produces a decent test of your content. If free users generate a lot of chats about the same three things, those are three articles you have not written yet, and writing them helps your search traffic at the same time.
How MyLiveChat fits
We run a free plan ourselves — one agent, unlimited chats, no expiry — so this is a question we have had to answer rather than theorise about. Unlimited chats matter here specifically because a free-user population is where volume is least predictable, and a per-conversation cost would make the sensible policy the stingy one.
For keeping the cost proportionate, AI answering grounded in your knowledge base handles the routine questions and hands off when it should, which is what makes chat-for-everyone affordable. The AI chatbot and knowledge base do that work, and our pricing is flat, so a busy month with free users does not arrive as a surprise invoice.
One boundary worth keeping: a public chat widget is not an authenticated channel, so anything that changes an account — a plan change, a refund, a data request — should be verified through the account itself rather than accepted on the strength of a chat message.
What to measure
Three numbers make the policy reviewable rather than a matter of opinion. The share of chats coming from free users tells you the actual load, and it is usually lower than the team believes. The share of those resolved without a person tells you whether self-service is doing its job. And the upgrade rate following a chat, measured over a sensible window, tells you whether these conversations pay for themselves.
Read them together and quarterly, not weekly. The free-plan question is a slow one, and the honest answer is often that support is cheap and the reputation is worth more than the queue time — but that should be a conclusion you reached from your own numbers, not an assumption either way.