Run Vendor Demos Against Your Work Instead of Their Presentation
Prepared with AI assistance by Alactic Editorial. Examples identified as hypothetical are illustrative, not customer results.
A vendor creates a record, moves it through a workflow, and opens an attractive dashboard. Everything looks straightforward. Your team still does not know what happens when the record already exists, the assigned person is unavailable, or a user lacks permission to make the change.
The solution is to give vendors the same short brief before they demonstrate: your starting conditions, representative tasks, expected outcomes, and one meaningful exception. During the session, label each capability demonstrated, stated, or unverified. Record what happened, not what you inferred from the presentation.
This makes the demo a useful screening step before a pilot. It does not turn a staged demonstration into proof of production reliability. The aim is narrower: identify which requirements have visible support, which depend on claims, and which still need evidence.
1. Choose the work that should drive the demonstration
Start with a decision your team needs to make, not a catalogue of features. Instead of asking whether the product has assignment rules, ask whether a new enquiry can reach an accountable person, retain its context, and remain visible when the normal assignment cannot happen.
Select tasks using three tests:
- Frequency: Does this happen often enough to affect daily work?
- Consequence: Would failure cause lost work, inappropriate access, or expensive rework?
- Uncertainty: Is there something about the proposed solution that you cannot establish from documentation?
A familiar, frequent task establishes the normal path. A handoff exposes how information moves between people. An exception reveals whether the workflow has a usable recovery route. Avoid trying to cover every requirement in one meeting; a long checklist often produces shallow answers rather than better evidence.
Describe outcomes without prescribing exact interface steps. “The next person can see the enquiry and its previous handling” is a useful requirement. “Click this menu and use this particular tab” may unnecessarily rule out a different but workable design.
2. Send a brief that allows preparation without surrendering control
Share the script in advance. Surprise tests can reveal unfamiliarity, but they can also waste the session on sample-data preparation. You want the vendor ready to demonstrate your workflow, not merely ready to improvise.
Ask the vendor to identify prerequisites before the meeting: required configuration, user roles, additional products, external dependencies, and the relevant plan or workspace availability. A prepared demonstration is acceptable; hidden preparation is the problem.
Specify the starting state and the evidence you want at the end. If the task begins with an existing customer record, ask to see that record before the task starts. If the result should appear in a queue, ask to inspect the queue afterward. Otherwise, a presenter may show a plausible final screen without establishing how it was reached.
Leave a bounded space for the vendor to propose a better route. The outcome can remain fixed even when the implementation changes. Record the alternative and its operational cost rather than rejecting it simply because it differs from your current process.
3. Use this filled hypothetical demo brief
Fictional worked example: The organisation, records, timings, thresholds, and observations below are hypothetical. They illustrate a completed evaluation brief, not the capabilities or results of any actual product.
Decision and session conditions
Buyer: A fictional equipment-services business selecting software for enquiry intake and team handoffs.
Decision: Decide whether the proposed setup has enough evidence to justify a limited pilot for the intake team.
Workflow boundary: From receiving an enquiry to assigning responsibility and handing the work to a colleague. Quoting, billing, and live customer messaging are outside this demonstration.
Participants: An intake operator performs or narrates the work; a team lead checks the result; a buyer-side recorder logs evidence; the vendor explains configuration and dependencies.
Hypothetical session allocation: Five minutes for setup disclosure, twenty-five for tasks, ten for the exception, and ten for unresolved questions.
Starting data: An existing fictional customer, Northbank Repairs, with an open enquiry about equipment servicing; a second, new enquiry from the same customer; one available operator, one unavailable operator, and one restricted-role user. All contact details are synthetic.
Setup disclosure requested: Identify the demonstrated environment, relevant plan, enabled components, prepared configuration, and any steps performed before the meeting. Confirm separately whether the same setup would be available in the buyer’s intended workspace.
Task A: Receive and associate an enquiry
Buyer instruction: Enter the new enquiry from Northbank Repairs and associate it with the existing customer without losing the enquiry’s separate description.
Expected observable result: The operator can find the existing customer, associate the enquiry, and retrieve both the customer context and the new request. Any manual matching is made explicit.
Filled evidence note: The presenter searched for the customer, selected the existing record, and saved a new enquiry. The buyer requested a fresh search; both enquiries were visible under the same customer.
Label: Demonstrated for manual association in the shown environment. Stated for automatic duplicate detection, which the presenter described but did not show. Automatic matching is not required for this pilot decision.
Task B: Assign and hand off the work
Buyer instruction: Assign the new enquiry to the available operator, add a handoff note, and show what the receiving operator can see.
Expected observable result: The receiving operator can identify ownership, read the request, and see the handoff note without asking the sender to repeat it.
Filled evidence note: Ownership and the note were visible after switching to the receiving operator’s view. The presenter refreshed that view manually. No notification was observed.
Label: Demonstrated for ownership and visible handoff context. Stated for an optional notification. Notification timing and configuration remain unverified.
Exception: The intended operator is unavailable
Buyer instruction: Repeat the assignment with the intended operator unavailable. Show how the team detects the unresolved work and how an authorised person recovers it.
Expected observable result: The enquiry remains findable, its ownership state is understandable, and an authorised team lead can recover it without re-entering the request.
Filled evidence note: The presenter could not reproduce the unavailable-operator condition in the prepared environment. They described a shared queue as the recovery route, but did not show the enquiry entering that queue.
Label: Stated for the proposed shared-queue route; unverified for exception detection and recovery. Do not mark the exception as passed.
Access check and decision record
Buyer instruction: Using the restricted role, attempt to change ownership. Then show the authorised team lead making the change.
Filled evidence note: The demonstration had no restricted-role account available. An administrator’s view of role settings was shown, but the requested behaviour was not exercised.
Label: Unverified for enforcement of the requested restriction. The settings screen is supporting context, not evidence that the attempted action is blocked.
Hypothetical decision: Keep the vendor under consideration, but defer the pilot decision until a follow-up shows exception recovery and the restricted-role attempt. The buyer’s evaluation lead owns the open-items list; the vendor’s solution contact owns the requested evidence.
4. Keep evidence labels narrower than the sales claim
The labels describe your evidence, not the vendor’s honesty or the product’s overall quality.
- Demonstrated: You observed the specified behaviour under recorded conditions. Include the environment, role, input, and result where relevant.
- Stated: The vendor explicitly said the capability exists, but you did not observe it. Preserve the conditions attached to the statement.
- Unverified: Your requirement remains unresolved. There may be no answer, insufficient evidence, or an answer that does not cover the requested condition.
A single requirement can contain a stated capability and an unverified outcome. In the hypothetical brief, the vendor stated that a shared queue was available; whether it received the failed assignment remained unverified.
Split broad requirements into smaller claims. Seeing a note in another user’s view does not demonstrate a notification. Seeing an administrator change a record does not demonstrate that a restricted user cannot change it. Seeing a successful save does not establish behaviour under every input or load condition.
Keep the result separate from the evidence label. A restriction that visibly fails is still demonstrated behaviour, but it does not meet the requirement. Add a plain outcome such as “meets,” “does not meet,” or “not resolved.”
5. Interrupt the tour at the points where evidence can disappear
Agree that the buyer can pause the demonstration to inspect a result. This is not an invitation to interrogate every click. It is protection against skipping the transitions that matter.
Useful interventions include:
- “Before we move on, can we inspect the saved result?”
- “Which user role is active now?”
- “Was that configured before this session?”
- “Is that available in the proposed setup, or does it depend on something else?”
- “You described an automatic action. Can we observe it, or should we record it as stated?”
Assign one buyer to facilitate and another to record. When the facilitator also takes detailed notes, gaps often appear precisely when the demonstration becomes complicated. Capture concise observations rather than a transcript: starting condition, action, visible result, limitation, and follow-up.
If recording the session is permitted by participants and organisational practice, use it as a reference, not as a substitute for the evidence log. A recording without indexed requirements is difficult to compare across vendors.
6. Recover when the demonstration cannot complete the task
An incomplete demonstration is not automatically a product failure. The cause might be missing sample data, an unavailable role, an unconfigured dependency, or a genuine capability gap. Record the cause as stated until it is established.
Offer a controlled restart when setup is the obstacle. Return to a known starting state and repeat the same task. If the presenter substitutes a different workflow, ask whether it still satisfies the original outcome and what extra work it creates.
Do not let a workaround disappear into a positive score. A manual review queue might be acceptable for low-volume work but burdensome for frequent exceptions. Record who performs the extra step, how unresolved items become visible, and what the team must monitor.
When evidence cannot be produced, create a follow-up with a precise request: “Show the unavailable-operator case from input to recovered assignment using the proposed configuration.” A general promise to send more information leaves the original uncertainty intact.
7. Compare vendors without hiding critical gaps in an average
Use the same core tasks and outcomes across vendors, while allowing different implementation paths. Compare results at the requirement level rather than counting impressive screens.
For each requirement, retain its importance, observed outcome, evidence label, dependencies, and unresolved question. Treat a mandatory requirement with no evidence as an open gate, not a neutral score that can be offset by attractive optional features.
Consistency has a trade-off: a fixed script can miss a useful capability you did not anticipate. Allow a separate segment for the vendor’s proposed improvements. Keep that segment distinct so it does not replace the buyer’s core test.
Also distinguish demo evidence from pilot questions. A demo may establish that a handoff is possible. A pilot can investigate whether your people complete it consistently with realistic data and interruptions. Neither should silently stand in for the other.
8. End with a bounded next decision
Before closing the session, read back the unresolved requirements and ask the vendor to correct factual misunderstandings. Do not negotiate an ambiguous item into a pass merely to finish the meeting.
Your next action is to prepare one shared brief and one evidence log before the next demo. Identify a normal task, a handoff, an exception, and any decision-blocking access requirement. Give every unresolved item an owner and a requested form of evidence.
The measurable output is simple: every priority requirement should leave the session with an evidence label, an outcome, and either a decision or a specific follow-up. You may still have unanswered questions. You should no longer have to guess which questions they are.
Put the next conversation to work.
Explore Alactic Communications for WhatsApp conversations, campaigns and AI-assisted workflows. Choose the product and plan that fit your team.
Explore Communications