Knowledge Base Videos: How to Add Video to Help Articles (2026)

Most help articles are read by someone who is stuck. They have a task half done, a support ticket half written, and a few minutes of patience left. A short video can get them unstuck faster than three paragraphs of text. A long, outdated or badly placed one makes things worse.

This guide covers how to use video in a knowledge base well: which articles deserve one, how to structure the article around it, how long it should be, where to put it, how to keep it accessible and searchable, and how to keep a library of videos current as your product changes. Most of it works with any help center tool and any video host.

#Short answer

Add a video to help articles that describe a multi-step task in the interface, where seeing the clicks is faster than reading about them. Keep each video to one task, usually under two minutes. Put it near the top, under a one-line summary, and keep the written steps below it. Add captions and a transcript, record so the video is easy to replace, and check which articles get watched and which still lead to tickets.

#Which articles deserve a video

Not every article needs one. Video costs more to make and to maintain than text, so put it where it helps most.

#Good candidates

• Multi-step tasks in the interface. Setting up an integration, configuring permissions, building a report. Anything with five or more clicks across several screens.
• Tasks where people get lost. If support keeps answering "where is that setting?", a video showing the path is usually clearer than a description of it.
• Onboarding and first-run tasks. New users do not know the layout of the product yet. Watching someone move through it builds that picture.
• Your most viewed articles. Start with the articles that already get the most traffic or are linked from the most tickets.

#Use a screenshot instead when

• The answer is one setting on one screen.
• The article is a reference page: limits, plans, keyboard shortcuts, API fields.
• The interface changes very often and you cannot keep a video current.

#Use a GIF when

A GIF works for one short action of a few seconds: a drag and drop, a hover menu, a toggle. It loops, plays without a click and needs no sound. It is the wrong choice for anything longer, because viewers cannot pause or scrub it and file sizes grow fast.

#Structure: article and video together

A video does not replace the article. Some readers cannot play sound, some prefer to scan, some need to copy a value, and search engines read text far better than video. The pattern that works:

1. Title as a task. "How to invite a teammate", not "Team management".
2. One-line summary. What the reader will be able to do, and any requirement such as an admin role or a paid plan.
3. The video. One task, start to finish.
4. Written steps. Numbered, one action per step, matching the video. A screenshot for any step where the location is not obvious.
5. Troubleshooting. The two or three things that commonly go wrong.
6. Related articles.

Keep the written steps and the video in sync. If step 4 in the text is step 6 in the video, readers notice. The easiest way is to make both from the same recording. Our guide on turning a screen recording into a step-by-step guide shows how.

#How long should a help center video be

As short as the task allows. Someone in the middle of a task wants the answer, not an introduction. A rough guide:

• Single task: 30 seconds to 2 minutes.
• Setup or configuration with several parts: split into separate videos, one per part, and link them.
• Concept explainer: 1 to 3 minutes.

Cut the intro. Do not open with a logo animation or "Hi, in this video we will". Start on the screen where the task begins, with one sentence of context. End when the task is done. If a video runs long, it is usually covering more than one task.

#Recording for a knowledge base

Help center videos are watched in a small player, often on a laptop beside the product. Record with that in mind:

• Use a clean browser profile, demo data and Do Not Disturb.
• Zoom the interface so labels are readable in a small embed.
• Move the mouse slowly and pause before each click.
• Record one task per file, so one changed screen means one re-recorded video.

Our screen recording tips cover each of these in detail.

#Narration

Narration helps, but plenty of people watch support videos on mute. Write the narration so it names what to click and why, in plain words, and make sure the video also works with captions alone. An AI voice is a practical choice for a help center library because a changed line can be regenerated without a microphone session, and every video sounds the same. See how to add an AI voiceover to a screen recording.

#Where to place the video in the article

• Near the top, under the one-line summary. Readers who want the video should not have to scroll for it.
• Not as the only content. The written steps belong directly below it.
• One video per article in most cases. If an article needs three videos, it is probably three articles.
• Use a clear poster frame. The first frame should show the screen where the task begins, not a blank or loading page.
• Do not autoplay with sound. Let the reader start it.

#Captions and transcripts

Captions are essential, not optional. They help people who are deaf or hard of hearing, people watching on mute in an office, and readers whose first language is not the one spoken. The W3C has a clear primer on captions and subtitles.

• Use closed captions (a separate track, such as a WebVTT or SRT file) in your own player so viewers can turn them on and off.
• Burn in captions only where a player cannot show a caption track.
• Proofread product names. Automatic captions often get them wrong.
• Publish a transcript or the written steps on the page. Text on the page is what search engines and your help center's own search can index. A video alone is close to invisible to both.

If you narrated a video yourself and only need caption files, our free SRT caption generator makes SRT and VTT files from the audio.

#Hosting and embedding

There are four common ways to host knowledge base videos. Each has trade-offs.

• Your help center tool's own video support. Simple, but players and caption options vary between tools, and some only support embeds from outside hosts.
• A public video platform, with the video set to unlisted if you do not want it found there. Free and reliable, but the player may show recommendations or branding you do not control.
• A business video host. Usually gives a cleaner player, caption tracks, viewing stats and no unrelated recommendations, at a cost.
• Self-hosted MP4 with an HTML5 video element and a caption track. Full control, but you handle storage, bandwidth and stats yourself.

Whichever you choose, check three things before committing: that captions work in the embedded player, that the embed is responsive on a phone, and that you can replace a video without changing its embed code. That last one matters more than it seems, because updating a video should not mean editing every article it appears in.

#Keeping videos current

Outdated videos are the main cost of video in a knowledge base. A button moves, a menu is renamed, and the video now shows a screen that no longer exists. Readers trust the help center less every time this happens.

• Record one task per video, so a UI change touches as few videos as possible.
• Keep a list of every video, the article it lives in, the screens it shows and the date it was recorded.
• Tie updates to releases. When a release changes a screen, the product or support team flags the affected videos.
• Review on a schedule for anything the release process misses, starting with the most viewed articles.
• Remove a video rather than leave a wrong one. A text-only article is better than a misleading video.

#Translations

If your help center serves several languages, decide early how videos will follow. Options, from cheapest to most work:

1. Translated captions on the original video.
2. Translated narration, recorded by native speakers or generated with an AI voice per language.
3. Re-recording in a localized interface, so the screen matches the reader's language.

Captions are a reasonable start. For tasks where button labels matter, readers following a video in their language while looking at an English interface can get confused, so match the interface language when you can. Have a native speaker check each version.

#Measuring whether videos help

Do not assume a video helped because it exists. Useful signals:

• Play rate: what share of article visitors start the video.
• Completion: how many watch to the end, and where they drop off.
• Article feedback: most help centers have a "Was this helpful?" control. Compare before and after adding a video.
• Tickets on the topic: whether tickets about the task fall after the video goes live.
• Search terms: what people search for in your help center and do not find.

#How Nook helps with knowledge base videos

Nook is built for exactly this pattern: a video and a written article from one recording. You upload a screen recording you already made, and Nook splits it into steps, writes a script line per step, adds a voice (an AI voice or your own with filler words and silences removed), captions and zoom on clicks, and renders an MP4. From the same steps it makes a written guide with a screenshot per step, which you can export as Markdown for tools like Notion or Confluence, or as Word or PDF.

You publish the video with a share link or embed code, or download the MP4 on paid plans and host it wherever your help center expects. For tasks where readers learn best by doing, you can make an interactive demo from the recording instead: one screenshot per step, a pointer that marks where to click and a short card explaining each step, so the reader clicks through at their own pace. Interactive demos have no voice.

Nook also blurs private data it detects, such as emails, phone numbers, card numbers and API keys, in both the video and the guide, and translates into 14 languages besides English. Check the blur and the translations anyway.

The limits: Nook does not record your screen, and it does not notice when your interface changes. When a screen changes, you re-record that task and run it again. There is no timeline editor, by design. Recordings can be up to 15 minutes on paid plans, which is plenty for a single task. See help center videos for more on this use case.

#Common mistakes

1. A video on every article, including ones a screenshot answers.
2. Video without written steps, so readers who cannot play it are stuck.
3. Long intros before the task starts.
4. Several tasks in one video.
5. No captions, or automatic captions that mangle product names.
6. Videos that show an old interface.
7. No list of which videos exist, so nobody knows what a release broke.
8. Real customer data visible in the recording.
9. Never checking whether anyone watches.

#Checklist

#FAQ

#Should every help article have a video?

No. Add video where someone needs to follow several steps across screens. One-screen answers work better as an annotated screenshot, and reference pages work better as text. Videos cost more to keep current, so spend that effort where it helps.

#How long should a knowledge base video be?

Usually 30 seconds to 2 minutes for a single task. If a video runs longer, it often covers more than one task and should be split into separate videos and articles.

#Do help center videos help with SEO?

The video itself is hard for search engines to read. The text around it is what gets indexed, so publish written steps or a transcript on the same page. Those also help your help center's own search find the article.

#Where should I host help center videos?

Options include your help center tool's own video support, a public video platform set to unlisted, a business video host, or self-hosted MP4 files. Pick one where captions work in the embedded player, the player works on phones and you can replace a video without changing the embed code.

#How do I keep knowledge base videos up to date?

Record one task per video, keep a list of videos with the screens they show, and flag affected videos whenever a release changes a screen. Keep the source recording and script so an update is a quick re-record.

#Should I use a video or an interactive demo in a help article?

A video suits tasks people want to watch once, ideally with narration. An interactive demo suits tasks people want to try step by step at their own pace. Our comparison of interactive demos and videos goes into the trade-offs.

#Do I need captions on help center videos?

Yes. Captions help viewers who are deaf or hard of hearing, people watching on mute and non-native speakers. Use a caption track the viewer can toggle, and proofread product names.

#Start with your most viewed article

Pick the help article that gets the most traffic or the most related tickets, record that one task cleanly, and publish the video above the written steps. Then do the next one. If you want the video, the written guide and captions from one recording, Nook does that from a file you upload. See help center videos or make your first video free.