Workplace Communication
How to Give a Product Demo That Gets Useful Feedback
A complete fictional ten-minute volunteer-roster demo script, failed-preview fallback, classified feedback and accepted follow-up distinguish opinion from user-task evidence.
Give a product demo around one user task and two or three questions you need answered. Show the relevant journey, identify what is working versus simulated, and pause where feedback can change the design. Capture what people actually saw or said, then separate suggestions, unresolved questions, and approved next steps.
A smooth walkthrough proves that the presenter can show a prepared path. It does not establish that users can complete the task unaided, that the system handles failures, or that anyone has approved delivery. This guide from LandOffer.ai follows a fictional ten-minute demo through its script, interruption, feedback, and completed follow-up record.

Decide what feedback could change the work
Start with the purpose of this particular session. A concept demo might seek feedback on the problem and workflow. A review of implemented work might ask whether its behavior matches an agreed requirement. A release review needs its own evidence and authority. Calling all three a demo can leave people with conflicting expectations.
Write the user task before choosing which screens to show: “A workshop coordinator changes a volunteer shift and knows whether the replacement is only saved or actually communicated.” That is a journey with a meaningful outcome. “Show the dashboard, editor, and settings” is a tour of locations.
Then choose questions that are specific enough to answer. “Is this good?” encourages vague approval. “Where could a coordinator confuse a saved replacement with a notification already sent?” directs attention to a risk without telling people that the current design solves it.
Atlassian's Demo Trust play asks presenters to identify the feedback they need and show the user's experience rather than overwhelm the audience with every detail. Its leadership-review format is one working method, not a required meeting length or evidence that your particular product has been validated.
Tell the audience what they are seeing
Introduce the artifact and its limits before you start clicking. An interactive prototype can communicate navigation while using fixed responses. A development build may have working controls but a simulated external service. Successful navigation alone does not establish that its external effects or failure handling work. A screenshot can explain a state without demonstrating any behavior at all.
Use synthetic data appropriate for the task. Remove real personal information from examples and any fallback material. If the product would send messages, charge money, change access, or modify a shared system, use an isolated demonstration route with those effects disabled. Describe any unimplemented boundary instead of creating a surprise real action.
Agree who presents, who observes and takes notes, who can decide scope, and who can approve later delivery. A senior attendee's enthusiasm is not automatically a release approval. If the team does need a decision, state the specific decision and invite the authorized person to make it explicitly.
Prepare a fallback that preserves honesty. Label a screenshot or storyboard as such, and say which questions it can still answer. Switching to a still image does not establish that the interaction works. Record the interruption so later readers do not mistake the session for an uninterrupted successful run.
Worked asset: a ten-minute task-based demo script
The following team, prototype, session events, feedback, and documents are fictional teaching artifacts. No real product was tested and no messages were sent. Rosa is presenting a clickable prototype for a community workshop's volunteer-roster tool. Emil coordinates volunteers, Hana supervises workshop operations, and Tess records feedback. Product owner Dev can authorize design work; deployment owner Imani is absent and retains release authority.
The prototype uses invented volunteers and the Saturday pottery workshop, with one vacant afternoon shift. It can display prepared states for selecting a replacement and saving a draft. Its notification confirmation is a fixed mock state. It cannot send email, persist a real roster, prove permissions, or demonstrate delivery to a volunteer.
Rosa's complete run-of-show reads:
| Time | What Rosa shows and says | Feedback sought or boundary |
|---|---|---|
| 0:00–1:00 | “We are reviewing the replacement workflow, especially the difference between saving and notifying. This is a clickable prototype with synthetic names and no messages sent.” | Ask whether the two state distinctions and confirmation wording need revision; no delivery approval requested |
| 1:00–2:00 | Show Saturday's fictional pottery workshop and the vacant afternoon shift. “Emil needs to choose a replacement before contacting that person.” | Establish the task and starting state; do not imply the selected volunteer has accepted |
| 2:00–4:00 | Select fictional volunteer Nora and open the review panel. Read the displayed shift and name without describing every navigation control. | Pause: “What would you need to check before proceeding?” |
| 4:00–5:30 | Select Save draft; show the prepared draft state. “This step is intended to save a proposed roster change, not notify Nora.” | Pause: “What do you think has happened at this point?” Record interpretation before explaining the label |
| 5:30–7:00 | Open the separate notification preview and fixed mock confirmation. “This confirmation is simulated; it demonstrates proposed wording, not a sent message.” | Ask what receipt or status a coordinator would need to distinguish a request from an accepted replacement |
| 7:00–9:00 | Return to the draft/notification distinction and take feedback. Keep a separate list for new feature requests. | Ask for concrete examples behind unclear labels, missing checks, or proposed additions |
| 9:00–10:00 | Read back the observed concerns, requests and open questions. Dev states any accepted next design work. | Confirm owners and next review point; make no release or date promise on Imani's behalf |
Rosa also writes a specific exception route. If navigation fails, she will try the prepared starting state once. If that does not restore the path, she will stop clicking and show labeled storyboard frames for draft saved, notification preview, and proposed confirmation. Tess will mark the remaining portion as a storyboard discussion. Timing or interaction observations after that switch cannot be treated as prototype behavior.
The script gives the team a sequence and useful pauses without requiring a polished sales performance. It also protects the central question from disappearing into settings, colors, and unrelated requests.
Ask questions without supplying the answer
Before a pause, resist saying, “This makes it obvious that no notification has been sent.” That explanation can conceal the very misunderstanding you need to discover. Ask what the audience thinks happened, listen, and then clarify the actual intended behavior.
Watching a presenter is different from attempting a task. The GOV.UK guidance on moderated usability testing describes observing actual or likely users trying specific tasks, with neutral instructions that do not reveal the answer. That is a useful boundary here: a guided stakeholder demo may generate research questions, but it cannot substitute for observing independent task attempts.
If a stakeholder says, “Users will never notice that,” ask for the experience or constraint behind the concern. Record it as a concern unless the session supplies relevant user evidence. If they say, “I would prefer a different color,” capture that preference without turning it into a universal usability finding.
Let people finish their feedback before defending the design. Clarify enough to preserve its meaning, name any contradiction that needs a decision, and continue. Give a parked request a visible disposition later so an attendee does not mistake your note for a promise to build it.

Worked asset: capture feedback and decide its disposition
In the fictional session, the prototype reaches the draft-saved state. Emil says, “So Nora has the message now?” Rosa records that interpretation before clarifying that no message has been sent. Within the fictional record, this statement establishes Emil's interpretation during the guided demo. It does not establish how other coordinators would understand the workflow or what caused his interpretation.
At minute six, opening the notification preview fails. Rosa's single reset attempt does not restore it. She switches to the labeled storyboard and explains the change. Tess notes that the later confirmation discussion is based on a still image. No attendee tries the task independently.
Hana asks whether the selected volunteer can decline. The prepared confirmation has no visible acceptance status. Dev asks for a bulk-replacement feature. Emil says the blue banner looks better than the gray one. Tess produces this completed feedback record:
| Entry and context | Classification | Interpretation or consequence | Disposition agreed in the discussion |
|---|---|---|---|
| Emil asks if Nora has received a message after Save draft | Observed statement in a guided demo | Possible confusion between draft storage and notification; cause not established | Rosa accepts revising the state wording and review panel for another design review |
| Preview navigation fails; one reset fails; presenter switches to still frames | Observed event within this fictional session | Remaining walkthrough does not demonstrate the interaction | Rosa accepts documenting the broken prototype link before sharing it again |
| Hana asks whether the volunteer can decline | Workflow question | Selection, notification, delivery and acceptance must not be conflated | Hana accepts describing the current operational process for Dev's scope discussion |
| Dev asks for bulk replacement | Feature request | A separate scenario could expand the current single-shift scope | Dev explicitly defers it; no estimate, implementation slot or delivery date assigned |
| Emil prefers the blue banner | Preference | Does not establish comprehension, contrast or accessibility | Retain in notes; no color change approved from this preference alone |
The table preserves the raw event separately from its meaning. GOV.UK's research-analysis guidance similarly distinguishes what observers saw or heard from interpretation, then moves from findings to possible actions. This demo is not a completed research study; the distinction still prevents a stakeholder opinion from becoming an invented user result.
Dev and the named owners then accept the following follow-up record:
Design work authorized: Rosa will prepare revised draft-state wording and a review-panel alternative by Thursday at 15:00 Pacific. She will also document and repair the prototype navigation link before the next showing. This commitment concerns the prototype, not a production defect fix.
Operational question: Hana will describe how coordinators currently handle a volunteer declining by Wednesday at noon Pacific. That description will inform a scope discussion; it does not approve a new acceptance feature.
Next review: Dev and Emil accept reviewing Rosa's alternatives Friday at 11:00 Pacific. Dev will decide whether to authorize a separate usability-research plan with likely coordinators. Participant recruitment and testing have not been approved or performed.
Scope and evidence: bulk replacement is deferred. Notification delivery, persistent roster changes, independent task completion, accessibility and permissions remain untested. The storyboard portion supplies feedback on proposed content only. Imani has granted no release approval.
At the cutoff of this record, the revisions, operational description and Friday review are future work. The useful completed artifact is the accepted plan and accurate evidence record, not an invented improvement in user success.
Close the loop without committing to every request
Send the recap to the people whose input or authority matters. Include the artifact version, synthetic-data scope, questions asked, interruption and fallback, feedback classifications, accepted work, deferred requests, and unresolved decisions. Link the team's current work items so owners do not have to reconcile several competing task lists.
If feedback conflicts, expose the conflict. “Hana wants volunteer acceptance visible; Dev wants to preserve the current review scope” needs a product decision, not a presenter secretly adding both. If a request belongs to another team, ask that team about feasibility before announcing its commitment.
When a follow-up is completed, report the actual result at the correct level. “The prototype label was revised” is a change. “The revision reduced confusion” requires suitable evidence. “The demo was well received” is neither task-completion evidence nor permission to deploy.
Prepare your next demo around one unanswered question
Choose one real user task, write the question that could change your next design decision, and name the artifact's limits. Build the route and fallback around that question. Reserve time to read back the feedback and confirm what was actually accepted.
For more practical workplace communication, visit LandOffer.ai. Before your next demo, remove one screen that does not help answer the question, and add one pause where people can reveal an interpretation you did not expect.


