A short scoping exercise to identify the decision and who owns it
| IN BRIEF Before starting a BI project, we review the decision, data, users, owner, budget and security requirements with you. |
Before you start a BI project, five well-chosen questions often do more for you than a long demo. They help you check that you know which decision to improve, which data to use, who will look after the report, how much work is realistic and which security rules apply. We use them to ensure that a technically sound project produces something people want to use.
Bring at least one sponsor, a person who knows the sources and a future user together to answer these questions. Their answers may differ. The exercise makes those differences visible early, while they are still easy to resolve.
A one-hour workshop can save several weeks of misdirected work. Use it to surface what you still need to know before you can decide on the budget, schedule or architecture. A complete specification can come later.

Objectives, data, users, resources and security give you five ways into the project.
Question 1. Which decision do we want to improve?
Try completing this sentence: “When we see this deviation, we will decide to…” If no one can finish it, the need is probably too broad. Return to a specific meeting, action or responsibility to give the project an observable outcome.
“Get better visibility” describes an intention. Describe the actual decision, who makes it, when and what the current problem costs. For example, every Monday the logistics manager needs to identify items at risk of running out and prioritise supplier orders.
What action will the report change? How frequently is the decision made? What is the cost of getting it late or wrong? What observable result will define success for the pilot?
| A GOOD TEST If the report disappeared tomorrow, which meeting, decision or action would be affected? |
Question 2. What data is available and reliable?
We need to know where data lives, how it is produced and which limitations could affect what the report shows. A known flaw can be addressed or disclosed. An undocumented assumption can undermine confidence as soon as figures diverge.
List systems, files and owners. Check shared identifiers, how far the data goes back, refresh frequency, missing values and permissions. Data you can see in the ERP may be unavailable for automated extraction or at the granularity you need.
Take a sample and reconcile a few totals with the users. Document discrepancies so people can see how they affect the result. Start with a clear understanding of the data limitations.
Question 3. Who will use the report and how?
“Management” is too broad to define a user. Who will actually open the report, when and with what question? A summary read before a committee meeting differs from an operational view checked every morning. The situation determines the level of detail and the navigation.
Distinguish sponsors, data experts, report creators and readers. Identify their decisions, devices, export requirements and level of detail. A director checking a monthly summary and a supervisor acting on the shop floor every hour need different pages, schedules and permissions.
Who validates the designs and definitions? Who participates in acceptance testing? Who should see which rows or entities? What routine will keep the report in use after delivery? What support or training will be available?
Question 4. What budget, schedule and owner do we have?
The budget includes scoping, data preparation, the model, reports, licences, infrastructure, training and maintenance. Set a scope you can deliver in the time available. A successful pilot that supports one process gives you a usable result to build on.
Name a person responsible for the BI product after delivery. They decide which changes to make, monitor refreshes and coordinate incidents. Without an owner, requests pile up and the report deteriorates.
Question 5. What are the security and compliance requirements?
Map sensitive personal, financial, commercial and industrial data. Define roles, retention, external sharing, hosting location and how to revoke access. In Switzerland, the revised Federal Act on Data Protection has applied since 1 September 2023. The GDPR may also apply to some activities.
Row-level security can limit which data a user sees in a report. Test it. Visual filters and hidden pages do not provide access control. Check exports, subscriptions and service accounts too.

Move ahead when all five answers are specific enough to keep the pilot manageable.
The scoping checklist
| Question | Minimum answer | Warning sign |
| Decision | User, frequency, action and outcome | “Build a dashboard” |
| Data | Sources, owners, sample and quality | Access or identifiers unknown |
| Users | Roles, testers, permissions and routine | Users consulted at the end |
| Resources | Scope, budget, timeline and owner | Nobody responsible after delivery |
| Security | Sensitivity, access, hosting and retention | Filters treated as access control |
When should you postpone the start?
If the sponsor is unavailable, no one understands the sources or the scope changes every week, a few preparatory workshops can save time. We prefer to resolve those issues before delivering a solution that has no owner.
Consider postponing when the sponsor cannot name a decision or measurable result, the main source is inaccessible and no alternative can be tested, nobody validates definitions or owns the product after delivery, the proposed architecture cannot protect sensitive data, or the requested scope clearly exceeds available resources.
Give each blocker a short next action: get an export, name an owner, define three KPIs or test permissions. Start the project when you know enough to run a sensible pilot.
What we want you to leave the workshop with
Keep it to one page. Record the problem, decision, users, sources, three to five KPIs, constraints, pilot scope, owners, schedule and acceptance criteria. List the unknowns and assign someone to resolve each one. Use that page as a shared starting point for assessing additional requests.
Datavizin offers a Power BI audit and support for an existing environment and an initial assessment with the Data Starter Pack to see whether the proposed use case is a good fit.
| NEXT STEP Bring together the sponsor, a user and the source owner. Answer the five questions in an hour and write down what remains unknown. Those answers will shape the work. |