A sandbox needs a service description
An implementation proposal that includes a sandbox can still leave important work unpriced. The word may refer to a temporary demonstration account, a copy of your configured application or a separate environment with limited capacity. Ask the provider to describe exactly what it will supply and what your team may do there.
Start with the intended activities: configuration, migration rehearsal, user practice, integration testing and later change review. State when each activity begins and ends. Some can share an environment; others may interfere with one another. A training session is difficult to run while another team is repeatedly deleting and reloading the data.
Microsoft's environment-planning guidance connects the number and type of environments to overlapping project work, cost and timing. Use that principle to question the proposal's assumptions. You do not need an elaborate environment portfolio for a small implementation, but you do need somewhere suitable for the work you have agreed.
References: Microsoft: Plan an environment strategy
Identify ownership and the access route
Record whether the environment belongs to your organization, the implementation partner or the software vendor. Identify the administrator and the route for granting and removing access. Company staff should know whom to contact if the consultant who created the environment becomes unavailable.
Ask whether your normal identity and permission model can be used. Training every person as an administrator may hide restrictions that appear at launch. Where a test environment cannot reproduce production roles, list the limitation and arrange another way to verify the affected workflow.
Specify whether external accountants, temporary testers and support engineers may enter, and who approves each group. Avoid shared credentials as a shortcut for a temporary project. The environment may contain sensitive information even when nobody calls it production.
Agree what data may be copied
Choose the smallest dataset that supports the planned work. Synthetic examples may be enough for training, while a migration rehearsal may require a carefully controlled copy of historical records. Have the appropriate company owner approve the data and the people who can see it.
Include attachments, customer contacts and bank-related details in that decision. Removing names from a spreadsheet does not necessarily remove them from attached invoices or free-text notes. Ask who performs the preparation and how the team verifies that the resulting records still support the test.
Define refresh procedures before the first copy. A refresh may replace configuration, erase test results or reintroduce confidential fields that were removed earlier. Record what survives, what is reset and who confirms the environment is ready for use afterward.
Agree how completed work is preserved before a reset. A migration mapping or approved configuration should not exist only inside an environment scheduled for replacement. Ask the provider to identify the export or backup it will retain, the person who checks it and the practical steps for restoring the relevant work if the next rehearsal fails.
Control connections that can reach the outside world
List every outbound connection in the environment, including email, payment services, document delivery and connected business systems. Ask the implementation team to demonstrate how test activity is prevented from reaching real recipients or creating real financial events.
Use clearly identifiable test destinations and records. A copied customer email address can look harmless until a training invoice is sent to it. Before a rehearsal, have someone check the configured destinations against the approved list rather than relying on the environment's name.
Keep the connection rules with the environment record. If a consultant reconnects a service to investigate a problem, that change needs an owner and a reversal step. Temporary connectivity should not quietly become the normal setup for every later training session.
Make capacity and version differences explicit
Ask which application version, features, integrations and capacity will be available. The test environment does not have to be identical in every respect, but the differences should be relevant to the conclusions you plan to draw. A small training instance may show navigation well while providing little evidence about a large import.
Microsoft's product-specific guidance notes that a basic acceptance-testing environment may be unsuitable for performance testing. That is a useful example of why an included sandbox should not automatically be treated as proof of production capacity. Confirm the requirements for the actual product and workload.
Agree who updates the environment and when. An unexpected update during acceptance work can invalidate a comparison; an environment left far behind production can conceal problems with the next release. Include a simple notice and scheduling process in the service description.
References: Microsoft: Product-specific environment guidance
Price the full period of use
Separate subscription charges from the labor needed to create, configure, refresh and support the environment. Ask whether storage, additional users, integrations or higher-capacity periods create extra charges. Obtain the unit and billing period for each item before calculating the project budget.
For illustration, assume a sandbox costs $240 per month for five months and two approved refreshes cost $300 each. The direct total is $1,800: $1,200 for access plus $600 for refresh work. Those assumed amounts are a budgeting example, not a vendor price estimate.
Ask what happens if the project extends. A temporary entitlement may expire before the team finishes its final rehearsal. Record the extension route, notice needed and commercial basis so a scheduling delay does not unexpectedly remove the place where remaining work must be validated.
Choose what remains after launch
Decide whether the environment will support future training and change review or be retired after implementation. If it stays, assign a budget owner, maintenance owner and data-refresh policy. If it closes, preserve the configuration notes, test evidence and other records needed to support the live system.
Ask the provider to describe the shutdown steps and confirm who performs them. These should address access, integrations, stored files and any ongoing charges. Retiring a test environment is a defined project action, not simply the moment when staff stop opening its URL.
The completed scope should let another administrator explain what the environment is for, who controls it and what it costs. That clarity makes the sandbox useful during the project and prevents it from becoming an unmanaged leftover afterward.
Sources
Requirements and product details can change. Check the linked guidance for the facts that apply to your situation.
Related reading
Need help with this work?
General information, not personalized accounting, tax, legal, or investment advice. Have the appropriate qualified professional review decisions that depend on your circumstances.