# A B2B Software Demo Script: Make Vendors Show the Same Work

<a id="main-content"></a>

B2B Strategy

October 28, 2026  ·  7 min read  ·  Victor Valentine Romo

On this page

1. [Start With One Workflow You Can Explain](<#start-with-one-workflow-you-can-explain>)
2. [Send the Same Scenario Sheet Before the Call](<#send-the-same-scenario-sheet-before-the-call>)
3. [Assign People Before Assigning Scores](<#assign-people-before-assigning-scores>)
4. [Label Every Finding by What Happened](<#label-every-finding-by-what-happened>)
5. [Run the Work From Input Through Export](<#run-the-work-from-input-through-export>)
6. [Compare Two Fictional Demonstrations Without Manufacturing a Winner](<#compare-two-fictional-demonstrations-without-manufacturing-a-winner>)
7. [Keep Follow-Up Work Small and Comparable](<#keep-follow-up-work-small-and-comparable>)
8. [Finish With a Buyer-Owned Evidence Summary](<#finish-with-a-buyer-owned-evidence-summary>)
9. [Sources](<#sources>)

A software demonstration needs a job to perform.

Give three suppliers three different conversations and you leave with three different sets of [impressions](</articles/google-search-console-advanced-techniques/>). One shows reporting. Another shows intake. The third spends the session describing an integration nobody opens.

Write the work before booking the screen share. Give each supplier the same starting records, user role, scenarios, and expected evidence. Leave room for different implementations. Keep the underlying question fixed.

[GitLab's solutions architects handbook](<https://handbook.gitlab.com/handbook/solutions-architects/>) describes demonstrations built around technical discovery and specific customer problems. That documents GitLab's practice. The buyer script below is a proposed evaluation method, with fictional examples throughout.

<a id="start-with-one-workflow-you-can-explain"></a>

## Start With One Workflow You Can Explain

Consider a fictional service team choosing software for incoming maintenance requests. A coordinator receives a request, checks required information, [assigns a specialist](</articles/dispatch-system-client-management/>), and prepares a weekly handoff file.

The team needs to understand what happens between receipt and handoff. It has not authorized a purchase or a production trial.

Create three synthetic records. Request R-101 contains an equipment identifier, service location, problem description, and requested date. Request R-102 omits the service location. Request R-103 contains complete information but needs reassignment after its specialist becomes unavailable.

These identifiers and conditions are invented. Use no customer names, addresses, messages, or real exports. A demonstration can begin with a small file created for this purpose.

Set the user role as coordinator. Allow that role to create requests, edit missing information, and propose assignments. Whether it can override an assignment is a question for the demonstration.

Define the expected result: a visible request history, an understood exception path, and a handoff file containing the agreed fields. State which [decisions need human approval](</principles/>). Avoid declaring that software must follow your current screen sequence.

Bring this workflow to an [operator tech stack review](</articles/operator-tech-stack-tools/>) as one bounded requirement. A list of tools cannot substitute for observing the work those tools would carry.

<a id="send-the-same-scenario-sheet-before-the-call"></a>

## Send the Same Scenario Sheet Before the Call

Give suppliers preparation time and the same information. Ask each to identify its proposed environment, product version, subscription tier, enabled modules, and custom configuration. [Record unknown items as unknown](</articles/null-blank-zero-business-data/>).

Ask which scenarios require setup before the demonstration. Preparation is useful evidence when its boundaries remain visible. A configured prototype can show a possible approach. Keep the work required to create it in the notes.

Use this scenario sheet as the agenda:

| Scenario | Starting condition | Ask the presenter to show | Evidence still required |
| --- | --- | --- | --- |
| Normal path | Complete R-101, coordinator role | Create, assign, update, and locate the request again | Behavior with the buyer's approved configuration and permissions |
| Missing input | R-102 lacks service location | Attempt assignment, then correct the missing field | Which controls are standard, configured, or bypassable |
| Exception | R-103 loses its assigned specialist | Reassign, preserve the reason, and show the visible history | Approval rules and behavior for concurrent changes |
| Export | All three synthetic requests exist | Produce [the weekly handoff file](</articles/csv-data-contract-business-handoffs/>) and open it | Field definitions, record selection, and repeatability |
| Unresolved capability | A required step cannot be shown | Explain the boundary and name the next evidence | Documentation, a scoped test, or confirmation of unavailability |

The table describes a proposed buyer exercise. It establishes no product behavior until a supplier performs a step and the buyer records the conditions.

If a supplier cannot use the provided file, ask why. Record its alternative starting data and how that changes the comparison. An unexplained substitution leaves the scenario incomplete.

<a id="assign-people-before-assigning-scores"></a>

## Assign People Before Assigning Scores

Name a coordinator, a workflow observer, and a decision [approver](</articles/consulting-deliverable-acceptance-criteria/>). One person may cover several roles in a small team. Make the responsibilities explicit anyway.

[Atlassian's DACI framework](<https://www.atlassian.com/team-playbook/plays/daci>) distinguishes the driver coordinating a decision from its approver, contributors, and informed participants. Borrow that separation where it fits your purchasing process. The framework does not establish your organization's approval authority.

Let the coordinator keep the session on the agreed scenarios. Let the workflow observer ask how a request reaches the next state. Ask the approver which unresolved conditions would prevent further evaluation.

Select evidence labels before watching. Otherwise, an attractive interface can influence what the team later calls sufficient proof.

Do not invent weights to force a winner. First identify requirements that remain unobserved. Decide whether those gaps need another demonstration, a bounded test, documentation, or no further work.

<a id="label-every-finding-by-what-happened"></a>

## Label Every Finding by What Happened

Use four evidence labels. Attach the environment and scenario to every entry.

**Observed behavior:** The presenter performs the step and exposes the relevant result. For example, the buyer sees R-102 blocked from assignment until its location is supplied. This observation applies to that demonstrated environment.

**Vendor assertion:** The presenter [describes a capability without showing it](</articles/ai-vendor-due-diligence-small-business/>). Record the statement and the evidence requested. Keep an assertion separate even when it sounds plausible.

**Configured prototype:** The presenter shows a workflow assembled for the demonstration. Note the custom fields, rules, extensions, or sample data involved. Ask who would build, approve, and maintain the equivalent configuration.

**Unavailable capability:** The required behavior is confirmed unavailable for the stated configuration. Distinguish that answer from a presenter who cannot show it during the session. “Not demonstrated” remains unresolved until evidence clarifies it.

These labels describe evidence status. They do not rank suppliers automatically. A finding can carry two labels: observed behavior in a configured prototype. Record both the result and its setup dependency.

Capture the result before discussing its appeal. “Coordinator changed the assignee; history displayed the previous assignee and reason” is inspectable. “Great workflow” provides no next step.

<a id="run-the-work-from-input-through-export"></a>

## Run the Work From Input Through Export

Open with a short instruction: “Please use the supplied records and coordinator role. Show each transition and identify any preconfigured steps.” This is proposed wording for the buyer, not a quotation from a completed evaluation.

Start at the actual entry point. Ask the presenter to create R-101 rather than beginning with a polished final dashboard. [Observe which fields are supplied](</articles/data-lineage-small-business-reports/>), calculated, selected, or inherited.

When the request reaches assignment, ask what the coordinator can change. Record any switch to [an administrator account](</articles/small-team-access-review-evidence/>). A permission change alters the conditions under which the result was shown.

For R-102, try the incomplete action before correcting the data. Ask where the missing field becomes visible and which user receives the next task. Follow the correction back into the original workflow.

For R-103, keep the original request open while changing the assignee. Ask where the prior assignment and reason remain visible. If history cannot be shown, record that gap without assuming the information is lost.

End with the export. Open the generated file and locate all three identifiers. Compare field names, current status, assigned role, and exception information against the requested handoff. Note any transformation outside the demonstrated product.

Treat screens, exports, and notes as evaluation material. Agree on [recording and retention](</articles/business-record-retention-schedule/>) before capturing the session. Keep the synthetic records separate from later approved business data.

<a id="compare-two-fictional-demonstrations-without-manufacturing-a-winner"></a>

## Compare Two Fictional Demonstrations Without Manufacturing a Winner

Suppose two fictional suppliers perform this exercise. Call them Supplier A and Supplier B. Neither represents a real product or an evaluation I conducted.

Supplier A performs the normal path using its demonstration coordinator account. Its presenter says missing-field checks can be configured but does not show one. Reassignment appears in the request history. The export contains the current assignee but omits the reassignment reason.

Supplier B shows a prototype configured for the supplied workflow. Its missing-field rule blocks assignment. The presenter changes to an administrator account for reassignment. The export includes a reason column, but the presenter cannot establish whether it contains the complete history.

Write the comparison as questions requiring evidence:

| Requirement | Supplier A evidence | Supplier B evidence | Buyer follow-up |
| --- | --- | --- | --- |
| Missing location | Vendor assertion | Observed in configured prototype | A: show the control. B: identify setup and maintenance ownership |
| Reassignment | Observed with coordinator account | Observed after administrator switch | B: establish the intended coordinator permission |
| Exception export | Reason omitted from sample | Reason present, history scope unknown | Both: define required history and repeat the export |
| Setup ownership | Not established | Custom setup shown, ownership unknown | Obtain a bounded configuration plan from each |

A has a missing control demonstration. B has a permission question. Both have unresolved export requirements. Those are different gaps. A total score would conceal them unless the buyer first decides their importance and acceptance evidence.

Ask the workflow owner whether the export needs current state, complete change history, or both. The original requirement may be incomplete. Clarify it once and send the same revision to both suppliers.

<a id="keep-follow-up-work-small-and-comparable"></a>

## Keep Follow-Up Work Small and Comparable

Create one follow-up row per unresolved question. Name the buyer owner, supplier owner, required artifact, and next review point. Keep proposed dates separate from accepted commitments.

Use the [mutual action plan](</articles/mutual-action-plan-b2b-services/>) to coordinate that buying work. It can record the next evidence without turning a demonstration into a purchase milestone.

When documentation arrives, identify what it establishes. A permissions page may describe a supported setting. It does not show that your proposed configuration uses it correctly. Leave any remaining test question open.

If a supplier proposes [custom development](</articles/zapier-vs-custom-automation-b2b/>), describe the proposed work separately. Ask what exists today, what remains to be built, and how acceptance would be established. A roadmap statement stays an assertion.

Repeat only the affected scenarios when practical. If a permission change also affects export access, include that dependency. Record which earlier observations remain relevant and which need to be made again.

<a id="finish-with-a-buyer-owned-evidence-summary"></a>

## Finish With a Buyer-Owned Evidence Summary

Write a short summary the buyer can carry into a [business case decision memo](</articles/b2b-business-case-decision-memo/>). Include the workflow, tested scenarios, environment details, observed results, unresolved assertions, prototype dependencies, and confirmed limitations.

Attach an evidence location to each conclusion. Keep the supplier's explanation beside your team's interpretation when they differ. Ask the workflow owner to confirm that the summary represents what was shown.

State the next decision narrowly. The team might authorize a limited evaluation, request another artifact, pause, or remove a supplier from consideration. [Record the person authorized](</articles/decision-journal-business-operators/>) to make that decision.

A demo is neither a production test nor an authorization to purchase. Its evidence remains bounded by the records, roles, configuration, and conditions observed.

Keep the questions the screen did not answer.

<a id="sources"></a>

## Sources

- [GitLab: Solutions Architects Handbook](<https://handbook.gitlab.com/handbook/solutions-architects/>). GitLab's connection between technical discovery and demonstrations. Checked 2026.10.28.
- [Atlassian Team Playbook: DACI Decision-Making Framework](<https://www.atlassian.com/team-playbook/plays/daci>). Coordination, approval, contributor, and informed roles. Checked 2026.10.28.
