Recording tips

TurtleScene builds everything from what it sees in your recording: it finds each click, turns it into a step, and times the narration around those steps. A few habits while recording make the steps more accurate and save you editing later.

#Plan one flow per recording

Record one task from start to finish, such as "create a project" or "invite a teammate", rather than a tour of everything. A single flow gives a clear script, a clickable demo with a clear end, and short social cuts.

• Decide the path before you press Start recording. Know which buttons you'll click, in order.
• Keep it short. Most product walkthroughs land between 1 and 5 minutes, and your plan caps each recording (3 minutes on Free up to 20 on Team).
• If you have several flows, record them separately. Each becomes its own demo, and each uses one AI recording. See Plans, credits & limits.

#Set up before you record

• Sign in first. Log in to your product before you start, so the login screen and your password aren't in the recording.
• Use the Open app window. Open app opens your product in its own 1280 × 800 window. Share that window or tab rather than your whole screen, so other apps and notifications stay out of the video.
• Start on the first screen of the flow. Whatever is on screen when recording starts is the first thing viewers see in the video.
• Close what you don't need. Extra tabs, chat pop-ups and banners in your product all end up in the frames.

#Hide personal and sensitive data

Everything visible in the recording appears in the narrated video, in the clickable demo's screenshots, and in formats such as the GIF and the help article. It is also analysed by AI to find your steps.

• Record in a demo or test account with made-up names, emails and figures.
• Don't open real customer records, inboxes, invoices or settings pages that show keys or tokens.
• Check the preview before you continue. If something slipped in, click Discard & re-record. A discarded take is never uploaded.
• Watch a video all the way through before you upload it with Upload a video. The same rules apply to it as to a recording.

If something private still ends up on screen, you can blur it after recording: in the demo editor, click Blur and draw a box over it. That area is destroyed in the narrated video, its formats, the clickable demo and stories made from the recording. See Blur private details. The first analysis that finds your steps runs before you can add a blur, so keep anything you'd rather no service sees off screen. To start again, delete the recording from Recordings and record the flow again. See Recordings library & reuse.

#Click clearly and pause on each screen

Click detection works from the video alone: AI watches the recording, works out where and when each click happens, and places a step there. The clearer each click is on screen, the better the steps.

• Pause for a second or two on each screen before you click. In the narrated video, each step's card shows before the click, so the button needs to be visible and still. In the clickable demo, the frame just before the click becomes that step's screenshot.
• Make one deliberate click per step. Avoid clicking around to find things, or clicking the same thing twice.
• Let the page settle after scrolling. If you need to scroll to a button, scroll, stop, then click.
• Don't rush to match the narration. You don't need to leave time for the voiceover. Where a step's narration needs longer than the screen was on show, the finished video holds on the frame before the click until the line ends. See Steps, zooms & timing.

If a step is missed or lands in the wrong place, you can fix it in the editor: move a hotspot by clicking the right spot on the video, add one with Add hotspot here, or run detection again with Re-analyze.

#Don't worry about waiting

Loading screens, slow saves and pauses while you think are fine. After the demo is made, Find waits to cut looks for still stretches between steps and suggests cutting them, and you can mark any other stretch with Cut from here. Suggestions are only offered between detected steps and are never applied without your review. See Trim & cut waits.

Mistakes can be cut out the same way, so you don't always need to re-record. If the mistake changes what's on screen for the rest of the flow, recording again is usually quicker.

#Narrate while you record (optional)

If you'd like the script in your own words, tick Record my microphone before you start and talk through the steps as you go. Afterwards, on the editor's Script tab, click Use what I said while recording. TurtleScene transcribes what you said, matches it to steps by when you clicked, cleans out the ums and repeats, and the AI voice reads the result. Your own voice is never in the finished video.

• Say what you're about to do just before each click, so your words land on the right step.
• Speak in full sentences. If there's too little speech, TurtleScene says We couldn't hear any narration in this recording. and you can let AI write the script instead.
• Record somewhere quiet.

If you don't want to narrate, leave the microphone off and let AI write the script. See Script & voice.

#Give the AI some context

When you set up the demo, the Context field is optional, but it helps. One or two sentences on what the flow does and what you click, for example "Creating a new project: the user clicks New, names it, picks a theme and generates the first screen", improves both the steps and the narration.

Choosing the right Demo kind helps too. See Create a narrated video.

#Keep good takes

If a take is good but you're not ready to make anything, click Save to Recordings. It uses one AI recording now, and the first demo or story you make from it later is included. See Recordings library & reuse.