top of page

Want Good Data Products? Ask Good Questions

Charles Sutton

Requirements Questions for Business Intelligence Projects


  1. What information do users need and why?

  2. Who are the people using this data?

  3. What are the expected outputs to make which decisions?

  4. When should the information be available?

  5. Where is the data for this request stored?


Information - why is the information needed for this request and how will it be used

1. What is the business problem to be solved?

2. What are the business questions answered by deliverable?

3. What is the expected data needed to answer the business question?


Audience - who are the people using this data and how will they use it

1. Who are the users and what are their roles?

2. What KPIs are users expecting to see?

3. What would the users do after they viewed the information?


Medium - what is the expected output of this data and in what form should it be delivered

1. Are users looking for at-a-glance information or wanting to do in-depth data analysis?

2. Does the user want to view the dashboard often or just be alerted to an anomaly?

3. Referencing other reports the department uses, what do they like/dislike about them?


Time - when should this information be available to make which decisions

1. What period do you need to see the information for?

2. Are there time periods that you'd like to compare against?

3. How frequently will this report be referenced and used?


Data – where is the data required for the report stored

1. What is the source data needed to answer business questions?

2. Is the data currently available or will a dataset need to be engineered?

3. How often is this data available?

4. What is the confidence level of data quality?

5. Does the data need to aggregated or individual level?


Information - why is the information needed and how will it be used


1. What is the business problem to be solved?

Don’t start any BI tool project without understanding the problem being solved by the tool. If a tool

doesn’t make a user’s life easier, then there is no reason to build it. The developer must know once

the tool is built what pain point of the user’s life will be alleviated.


2. What are the business questions answered by deliverable?

The business questions are what guide the developer on the views created in a report or dashboard.

A user has a situation with a question and they expect to get an answer from the BI tool. If the

developer doesn’t know what questions to answer, then they can’t create visuals that provide

answers the users need which results in an unused dashboard.


3. What is the expected data needed to answer the business question?

When the developer enters the data discovery phase, they will need a preliminary roadmap of the

source systems that capture the data and how this data moved through the data pipeline. Understanding the movement of data helps them build accurate measures later. If any anomalies

are found through data discovery then the developer will be prepared to educate the requester

about critical data points for analysis that may be incomplete or unavailable. This prevents any

surprises to stakeholders later in the design process and allows for alternative solution planning for

requested data views.


Audience - who are the people using this data and how will they use it


1. Who are the users and what are their roles?

A BI developer wants to understand the audience they are building a solution for. Later in

development a developer will want to know who they need to test usability against. At some point,

they will review those business questions asked earlier and will test their solution to determine if it’s

effective at helping end users do what they want to do.


2. What KPIs are users expecting to see?

This question collects the dimensions and measures a user wants to see in the business language.

Having stakeholders define key performance indicators forces them to not only think about the data

they want to see, but also the data points they want to compare against. If the user needs forecasts

or goal target data, then the developer also needs to add those data sources to the data model as

well. These KPI definitions will also be captured in the business glossary for reference during testing

and utilization.


3. What would the users do after they viewed the information?

Any BI project should be aligned with improving business outcomes by driving action by the user.

This question improves the chances that one the solution is deployed it will have the intended

impact on the business. Having requesters think about what action they expect to see data may also

change the data points they need to see to make the intended impact on the business.


Medium - what is the expected output of this data and in what format


1. Are users looking for at-a-glance information or wanting to do in-depth data analysis?

A developer will need to get an idea of what level of information a user will need to answer their

business question. Some users just need a snapshot of business performance while others need drill

features to explain root causes to performance. The more research features included in the report

influences how functionality is defined in a report. The developer will also need to test interaction

features during development to confirm features impact the reporting views as they were intended.


2. Does the user want to view the dashboard often or just be alerted to an anomaly?

Sometimes a user wants to spend the least amount of time in a report possible because they are

busy running the business. They only want to be alerted when processes or performance are

behaving unexpectedly. Alerting features need to be built into a report so a user is alerted to a

critical business issue. The developer will want to test during development that the alert feature is

working as intended.


3. Referencing other reports the department uses, what do they like/dislike about them?

Users will gladly tell you what reports they like or don’t like and why. The features they like from

other reports should be available in the design. Features that frustrate users in other reports should

be redesigned or not added. A simpler solution that delights the user will improve the chances a BI

tool will be used once deployed.


Time - when should this information be available to make which decisions


1. What period do you need to see the information for?

Users have a good idea of how they like to see the data. Depending on the user, they may only need

daily information while others may need to see a trend. The developer will need to plan for

providing the timelines needed to answer the user’s business question. Additionally, the size of the

data model can impact the latency of the report. A developer needs to figure out during data

discovery how the size of the data model will impact the responsiveness of the report. Depending

on size, alternative options may need to be discussed with the requester to make sure the BI tool

renders data quickly.


2. Are there time periods that you'd like to compare against?

It is common that a user wants to see how the business changes compared to previous periods.

Comparison periods can be yesterday, last month, or last year for example. To provide these views

the developer will have to build additional measures in the model so visualizations can provide the

requested views. They will also want to test comp periods with report features so filters provide

accurate information.


3. How frequently will this report be referenced and used?

Users hope to have the data when they start their day or when they are attending critical meetings.

They may work across different time zones so the developer has to plan to have data ingested into

the data model by time the users day or meeting starts. Sometimes the data warehouse goes

through its own processes and may not have data available when needed. If data is not going to be

available, then users need to be aware and a alternative solution should be provided if possible.

5 views0 comments

Recent Posts

See All

Comments


©2019 by Impacsis. Proudly created with Wix.com

bottom of page