Live selling script: the framework for one show of 60 to 120 minutes

Plenty of sales teams prepare for a livestream by writing a script that runs to a dozen pages with every sentence spelled out. Twenty minutes into the show the host starts reading it like a public notice, comments scroll past unanswered, and the session drifts into improvisation. The problem is not how much preparation was done. It is that the wrong kind of script was written.

How is a framework script different from a word-for-word script?

A framework script fixes three things: the time markers (what happens at which minute), the mandatory messages (the facts that must never be forgotten) and the standing call to action (the ordering line, always phrased the same way). Everything outside those three is left to the host. A word-for-word script does the opposite: every line is written out and the host simply reads. On a live broadcast that second approach almost always breaks, because a livestream is an environment full of interruptions: comments cutting in, someone asking the price mid sentence, technical faults. A host tied to the text either ignores the comments to keep reading, or abandons the script to answer and never finds the way back. A framework script allows the host to leave and return, because it is anchored to minutes rather than to sentences.

The framework for one show usually fits on a single page: a column for the minute marker, a column for what must be said, a column for the closing line, and notes on props. Anything longer belongs in supporting documents such as the product spec sheet and the objection list, kept off camera for quick reference.

How should a 60 to 120 minute show be split across the clock?

The framework below is a practice recommendation, not an industry standard. For a 60 minute show, keep the structure and cut the number of loops; for a 120 minute show, add loops rather than stretching each segment. The general rule: build each show around 3 to 5 products only. Beyond that, viewers stop remembering which item cost what.

Reference framework used by the Chạm AI content production team when planning a show. The numbers are operational markers, not market statistics.

Why reintroduce the product every 10 to 15 minutes?

A livestream has no fixed audience. Viewers arrive at minute 12, minute 40, minute 68, and most of them leave before the show ends. If the product introduction happens only once at the start, everyone who joins later hears a conversation without context: which item is being discussed, at what price, and where to buy it are all unclear. A 10 to 15 minute repeat cycle is a practice recommendation: frequent enough that newcomers do not have to wait long, spaced enough that people who stay do not find it grating.

A complete product loop should contain five parts. Miss any one of them and the loop is wasted:

1. Product name and price, spoken out loud rather than only shown on screen.
2. A single reason to buy, choosing the strongest benefit instead of reciting the whole spec sheet.
3. Visible proof: pick it up, open it, try it on, hold it next to a familiar object.
4. One objection handled up front, for example price, material or delivery time.
5. The standing call to action stating how to order in exactly one sentence structure, repeated identically in every loop.

The standing call to action is the detail most often skipped. When every loop phrases the ordering process differently, viewers have to assemble the information themselves and the chat team has to guess what was meant.

What should the 30-second hook say?

The first thirty seconds decide whether someone who just scrolled in stays, and it repeats many times during a show because everyone who joins gets their own "first 30 seconds". A practical formula: sentence one says what is on sale today; sentence two says why this show is worth staying for, whether that is a price that applies during the broadcast, a bundled gift or a trial; sentence three says what a viewer should do if interested. Avoid opening with drawn-out greetings or a warm-up story, because the viewer has no reason yet to spend time listening. Write three hook variants in advance and rotate them after each loop.

How do you handle comments and objections live on air?

Comments are the one part of a livestream that cannot be written in advance, but they can still be prepared for. Before the show, list the 8 to 10 most common objections in your category with a short answer for each. On air, keep the routine tight: repeat the objection in your own words so other viewers hear it too, acknowledge the fair part of it, then answer with concrete facts such as material, return policy or expected delivery time. Do not get drawn into an argument, and do not delete comments unless they break the channel rules.

Three families of objections come up constantly, each with its own direction. Too expensive: convert the price into cost of use over time or compare it with the alternative, without discounting outside policy. Not convinced about quality: move to visible proof and to the return terms. I will buy later: state the real conditions of the show, such as units remaining or the deadline, and state them only if they are true, because fake scarcity gets caught in the comments very quickly.

Aligning the voice of the host and the AI Agent in the chat

A live selling show usually has two voices speaking at once: the host on camera, and the team answering messages behind the scenes, a role increasingly handled by an AI Agent for live chat and automated advice. The risk is that the two voices do not match: warm and informal on air, "Dear valued customer" in the chat window; two-day delivery on air, three to five days in the reply. Buyers notice the gap immediately, and it costs trust at exactly the moment they were about to order.

The fix is a shared voice document and a shared fact sheet. The voice document sets forms of address, the vocabulary to use and to avoid, and how to respond when a customer complains, following the approach in our guide to building a brand voice for AI Agents. The fact sheet sets prices, stock, delivery times and return policy, and it has to be the single source for humans and machines alike. Before every show, update that sheet and reload it into the agent.

After the show: turning notes into a better framework

Once you sign off, record three things: the question that repeated most in the comments, the loop that produced the most orders, and the stretch where viewers dropped off. Those three facts are enough to adjust the framework for the next show without rewriting it from scratch. Repeat questions move into the objections you handle up front; the loop that worked moves earlier; the stretch that lost viewers gets shortened or replaced.

If your business wants to build a proper live selling operation, from the script framework and the production crew through to the post-show reply system, take a look at the content and deployment service packages from Chạm AI, or read the professional profile of Lâm, Content Strategist and Livestream Producer to see how we run a show.

FAQ - Frequently asked questions