How to build a video knowledge base for customer support

To build a video knowledge base, use support data to choose the recurring questions, record each answer in a safe demo account, turn the recording into a narrated video, and publish it beside written steps. With ScreenDub, the same project can produce a standard MP4, captions, and a step-by-step PDF. The production tool makes the answer easier to create; the support team still decides what is accurate, what is safe to publish, and when it needs a review. This guide covers prioritization, production, publishing, measurement, and maintenance.

What a video knowledge base should solve

A video knowledge base is useful when customers repeatedly need to see a workflow, not merely read a definition. The strongest entries answer one question, show the real product path, and leave the customer with a clear next step. Pair the video with written instructions so the customer can watch, scan, search, or follow the steps with sound off.

Do not treat the library as a replacement for every support interaction. Account-specific problems, bugs, billing decisions, and sensitive cases still need a person. The library should remove repetitive explanations so agents have more time for those cases.

Choose which support questions deserve a video

Start with evidence from the help desk, not with the feature the team is most excited to show. Sort how-to tickets by volume, time to resolve, repeat follow-ups, and how difficult the flow is to explain in text.

• High-volume how-tos: repeated setup, navigation, export, and configuration questions.
• Visual workflows: tasks that cross several screens or hide a control in a menu.
• First-week blockers: the questions that prevent a new customer from reaching first value.
• Stable procedures: flows unlikely to change before the content can pay back its production cost.
• Do not start with: one-off account issues, unresolved bugs, or material containing private customer data.

Build the first three answers

A small first release makes the workflow testable. Pick the three highest-impact repeat questions, publish them, and observe whether agents and customers can use the links. Expand from real demand rather than trying to create a complete library before anyone sees it.

1. Write the question exactly as a customer asks it.
2. Define the successful end state and any prerequisite permission or account setup.
3. List the one or two mistakes that most often create a follow-up ticket.
4. Choose the safest sample data and the narrowest product area to record.
5. Assign a product owner who can approve the final wording.

Production workflow in ScreenDub

1. Prepare a demo account. Open the product at the correct starting state and replace customer-like details with sample data. Close unrelated tabs and notifications.
2. Record the answer. Capture a browser tab, window, or full screen at a readable pace. Keep the video to one question and pause after important state changes.
3. Review locally. Trim mistakes, crop the frame, and blur private information before the recording is uploaded. See Editing your recording.
4. Draft and edit narration. ScreenDub splits the recording into segments and drafts a narration line for each. Use a support-oriented style, then correct the feature name, explanation, warning, and expected result.
5. Write for the searcher. Say the exact feature or task name early. Explain the reason for the action instead of reading visible button labels.
6. Export the answer. Render the MP4 with captions and create the step-by-step PDF from the same project. Put the video and written version in the same help article.

Use one shape for every support answer

Opening: name the question

Start with the customer’s wording or a close variant: “Here’s how to connect your Stripe account.” This confirms that the viewer found the right answer and gives the help center another clear phrase to index.

Middle: show the shortest safe path

Demonstrate the normal flow from a realistic starting state. Mention prerequisites before the first click, explain decisions that are not visible, and show the confirmation that proves the task worked.

Recovery: cover the common failure

Add one concise recovery instruction when it prevents a predictable follow-up: check a role, refresh a connection, or contact the correct administrator. Do not turn one answer into a catalog of every edge case.

Close: give the next action

End with the next help article, checklist item, or support path. A customer who completed the task should know what to do next; a customer who could not should know how to escalate.

Make the library consistent

• Voice: use a consistent voice and language for related answers.
• Visuals: use the same title, chapter, caption, and branding treatment.
• Titles: start with an action or customer question, not an internal feature codename.
• Length: keep one question per video; split when the viewer needs to scrub for a separate answer.
• Written alternative: publish the PDF or written steps beside the MP4.

ScreenDub supports captions that can be burned into the video or exported as a separate subtitle file. It can translate a project into 80+ languages, but have a fluent reviewer check product names, warnings, and regional workflows before publishing a translated answer. See Captions and Translating narration.

Publish where the question appears

• Help article: keep the searchable written answer and the video together.
• Support macro: give agents a precise link for the repeat question.
• Product help or empty state: surface the answer beside the blocked workflow.
• Onboarding sequence: include setup answers before the customer reaches the first blocker.
• Internal agent guide: use the same source to train agents on the approved path.

Measure usefulness without confusing views with resolution

ScreenDub creates and exports the content; use your help-center analytics, product analytics, and support data to evaluate the outcome. A view is only a starting signal. The stronger test is whether the customer completes the task or stops asking the same question.

• Repeat-ticket volume: does the same question occur less often?
• Follow-up rate: do customers still need an agent after receiving the link?
• Resolution time: can the agent answer the question with one accurate resource?
• Task completion: can customers complete the workflow after watching or reading?
• Search behavior: do customers find the answer using the terms in the title?
• Content feedback: where do viewers report confusion or an outdated screen?

Keep the library true as the product changes

1. Record the product version, source recording, owner, and review date with each article.
2. Review affected answers when navigation, permissions, pricing, or policy changes.
3. Re-record the affected workflow and correct its narration when the interface changes.
4. Edit and re-narrate only the changed segment when the wording changes but the screen does not.
5. Retire duplicate answers so agents and customers do not choose between conflicting versions.

One-question videos make maintenance smaller. A change to an export flow should update the export answer, not a long recording that also covers billing, permissions, and onboarding.

Current ScreenDub limits for support video

Sources

Browser capture, MP4/WebM/MOV video, or PDF/PPT/PPTX deck

Per-video cap

Free 6 min; Creator 10 min; Pro 12 min; Teams 15 min

Deck cap

Up to 100 slides and 650 MB per deck

Outputs

Standard MP4, optional subtitles, chapters, and step-by-step PDF guide

Languages

80+ narration languages from the same project

Free plan

One 8-minute narration allowance, 3 exports/video, 5 guides/month, watermark

Common questions

Start with the highest-volume how-to question in your queue. Create a free ScreenDub account, or read the Quickstart before recording the first answer.