Key takeaways
- Decide the single claim the demo must prove before writing any step.
- Structure the script around a user journey, not a tour of files or components.
- Show one prepared path and one honest edge case instead of a perfect happy path only.
- Rehearse the timing, the screen share, and the handoff between presenters.
- Keep the evidence behind the demo in the ticket so questions can be answered after the call.
Decide what the demo must prove
A demo without a claim becomes a screen tour. Before writing the script, state the one thing the audience should believe when the demo ends. It might be that a new workflow saves a manual step, or that a fix holds under a realistic input. Everything in the script should support that claim.
Write the claim in the ticket and share it with the audience in advance. A clear claim lets reviewers prepare the right questions and lets the pod leave out features that do not matter. It also makes a failed demo easier to diagnose because the intended outcome is explicit. If the team cannot state the claim in one sentence, the demo is not ready.
Build the script around a user journey
Start where the user starts, not where the code starts. Show the screen, the action, and the result in the order a real person would experience them. Name the data being used and why it is representative. A journey gives the audience a mental model that survives after the call.
Keep the number of screens small. If the journey needs many steps, show the two that carry the claim and summarize the rest. A demo that tries to show everything leaves no time for questions. The script should fit the agreed slot with room to spare.
Show one prepared path and one honest edge case
Prepare the main path so the demo is not derailed by a typo or a missing record. That is not deception; it is rehearsal. Then show one edge case that the team actually handled, such as an empty result or a rejected input. The edge case is where the audience learns whether the work is robust.
Do not invent a failure to look thorough. Use a case from the test suite or a real ticket. Explain what the system does and what the user sees. If the edge case is not handled yet, say so and name the plan. An honest gap is more credible than a polished illusion.
Rehearse the timing and the handoffs
Rehearse with the same screen share, browser, and data the demo will use. Time each section and cut anything that does not support the claim. Check the transition between presenters so the story does not stall. A short rehearsal catches most of the awkward moments.
Prepare a fallback for the most likely technical failure. A recorded clip or a set of screenshots can keep the demo moving if the environment fails. Tell the audience when you are using the fallback. Transparency protects credibility better than pretending the live system is working.
Keep evidence behind the demo
Link the demo script to the ticket, the test results, and the design note. If someone asks how a number was produced, the answer should be one click away. This is especially important for a distributed audience that cannot look over your shoulder. The demo is a summary, and the evidence is the record.
Capture the questions and answers after the call. Assign owners to the open ones and add them to the same ticket. A question that is answered only in a meeting tends to be asked again. Writing it down also gives the pod a checklist for the next demo.
Run the demo
Open with the claim and the user journey in one sentence. Then show the path. Pause after the result so the audience can see it before moving on. Avoid narrating every click; explain the decision and the outcome instead.
Leave the last few minutes for questions and for the evidence. If a question needs a number you do not have, say so and follow up in writing. Do not guess in front of the audience. The credibility of the next demo depends on how this one handles the unknown.
Limitations
A demo is a sample, not proof of production readiness. It can show a happy path and a handled edge case, but it cannot prove reliability under load or across every data shape. The pod should say what was not tested. Reviewers should ask for the evidence behind the claim.
A distributed demo also loses some of the room's informal signals. It is harder to see confusion or a side conversation. Ask for written questions as well as spoken ones. A short follow-up note with the recording and the evidence closes that gap.
Make the demo accessible to a remote audience
Use a font size and window size that remain readable on a laptop screen. Describe what is on the screen before clicking, because remote viewers may see the video a second late. Turn on captions if the platform supports them, and speak at a measured pace. A demo that only works in the room excludes part of the audience.
Share the script or an outline in advance for people who cannot attend live. Record the session and add the link to the ticket. Keep the recording short and label the sections so a new reviewer can jump to the claim. Accessibility is part of whether the demo does its job, not an extra step after it.
Close the loop with a written summary
Send a short summary after the demo: the claim, what was shown, what was not shown, and the open questions with owners. Link the recording and the evidence. This gives the audience a durable record and gives the pod a checklist for the next review. It also prevents the same questions from returning in the next meeting.
Ask for one piece of feedback on the demo itself, not only on the feature. Was the claim clear, was the pace right, was anything missing? A team that reviews its demos improves faster than one that only reviews the code. Keep the feedback short and act on one item at a time.
Questions about assessing Philippine developers
How long should a demo be?
Short enough to leave time for questions. Ten to fifteen minutes of scripted content is usually enough for one claim and one edge case.
Should we show unfinished work?
You can, if you label it clearly and explain what remains. An honest gap with a plan is more useful than hiding it until the next review.
What if the live environment fails?
Use a prepared recording or screenshots and say that you are doing so. Then follow up with a live check once the environment is stable.
