The Coeo Blog

The unseen dangers of Self-Service Power BI

Written by Andy Jones | 07-Jan-2020 12:15:00

“The more things change, the more they stay the same”.

Did you ever see a SQL Server Reporting Services (SSRS) setup that didn’t include a mass sprawl of reports with only vague knowledge within the business of the purpose of the majority?

The SSRS reporting estate might lack consistency, including similar outputs with slight variations. Report consumers never open most reports, with the original authors long departed the company. Nobody wants to be the person to break things, so nothing is deleted or edited. New requirements mean you add yet more reports to the confusion.

Let’s avoid repeating history with your self-service Power BI solution.

 

Power BI Self-Service Options

Firstly, what do we mean by self-service Power BI? You have three options:

  1. Business controlled – business users import, cleanse, model, visualise, publish and share.
  2. Hybrid - IT import, cleanse and model, i.e. create the semantic layer. Business users visualise, publish and share.
  3. IT controlled – IT publishes both the semantic layer and all reports. Business users slice and dice pre-configured Power BI apps.

The beauty of Power BI is there is no correct answer, and it’s a Swiss army knife. It can be the cornerstone of your enterprise BI strategy or used by a lone user to visualise their annual spend or fitness regime. There is a trade-off between IT providing a light-touch supporting role and fully implementing the whole solution. The latter option is tightly controlled, but requires far greater IT resources, the offset being full security and governance.

Is IT-controlled the best for you, or maybe you prefer the hybrid approach? Was Power BI sold in your organisation with the promise anyone can hook up to any data and publish visuals, and you are running with a full business-controlled solution?

Which option will you choose? What do you mean by self-service at your organisation?

 

No Governance

In your enterprise, you must implement proper governance to avoid creating data silos, data sprawl, poor performance and lax security. Without proper governance, your business users have unfettered access to source systems and Power BI workspaces. Users become enthused (not a bad thing!) with the fantastic functionality of Power BI and publish and share many reports far and wide. However, the problems you might encounter are:

  • Many refreshes – your Power BI resources are not infinite, especially when using Power BI premium. Many Power BI datasets being unnecessarily hydrated drains your resources and negatively impacts performance.
  • Load on source systems – similar to the previous point, but worthy of a dedicated bullet. Many Power BI duplicate queries of production systems could negatively impact business operations. You might encounter contention on CPU, memory or database locks.
  • Many versions of the truth – users will not interpret business metrics in the same way. An example is “Profit Year on Year % Change” for a supermarket in a particular geographical region. One report creator divides the profit difference over last years number – seems fair? Another report developer excludes stores opened or closed in the intervening year from both sides of the equation. Both reports include the “Profit Year on Year % Change” measure and are published throughout the business giving wildly different results. Very confusing for end-users and worse, might have serious implications when reporting financial numbers.
  • No audit – everyone is an admin of the tenant and workspace. Who is publishing outside your company, publishing to the public internet, deleting reports, dropping workspaces or downloading personally identifiable data from Power BI to their laptop?
     

Implement Governance

You implement governance to control self-service Power BI by:

  • Define – choose the best self-service strategy for your organisation. Document and communicate your reasoning.
  • Source permissions – implement the principle of least privilege for all data sources that feed your analytics solution.
  • Power BI workspace access – control who has access to your Power BI workspaces and at what privilege level. Who can publish to the workspace, refresh datasets or drop the workspace? See Power BI Security in Ten Steps.
  • Power BI admin portal – control and create a small group of Power BI admins. Define who can create workspaces, consume datasets, share externally, publish to web or export data outside the boundaries of Power BI.
  • A single version of the truth – Coeo recommends the best solution bespoke to you, using tools and techniques that might include: certified golden Power BI datasets, Power BI Premium, shared datasets across workspaces, Power BI Dataflows, common data model or Azure Analysis Services.
  • Data dictionary – based on your single version of the truth, create a data dictionary to communicate to your users.
  • Monitor and audit – people are diverting from your company standards? You need to know about it.

 

Coeo has developed an Azure Analytics Strategy series of workshops to cover the above points and many more. You will receive proven recommendations to deploy a secure, scalable, governed Power BI solution and unlock the huge potential in your data using Power BI. If you would like to hear more, please get in touch.