Make Managed Self-Service BI Work for You with Power BI and Purview (the BI Mullet Approach)

Description

When designed well, a managed self-service BI approach strikes the perfect balance between centralized data discipline and flexible self-service reporting. This session will demonstrate how smart Power BI implementation (including centrally maintained semantic models, report templates, and administration) saves time, money, and headaches when paired with Purview's Unified Catalog.

Key Takeaways

• Centralized Semantic Models
• Power BI themes and templates
• Unified Catalog in Microsoft Purview
• …but Enterprise BI isn’t always an option
• Discouraged business users
• Reporting projects slow way down
• Prioritization becomes based on

My Notes

Action Items

• [ ]

Resources & Links

Slides

Make Managed Self-Service
work for you with Power BI
and Purview
(The BI Mullet Approach)
Lenore Flower | Data Consultant & Trainer (MCT)
Session Agenda:
Why choose managed self service BI
What does managed self-service BI requires
How to make managed self-service work for you
• Centralized Semantic Models
• Power BI themes and templates
• Unified Catalog in Microsoft Purview
Who is best positioned to meet each of these needs?
(not just the most visible ones)
Data Integration Data Quality
Data Admin
Data Security Data Warehouse
Design, Build and Maintain
Project
Management
Change
Management
CI/CD
ETL
Semantic Model
Data
Storage
Report
Development
Your BI Haircut options
Enterprise BI
The BI Crew Cut
Self-Service BI Managed Self-Service BI
The BI Skullet
The BI Mullet
With Enterprise BI, the “data crew” designs, builds and
maintains everything
Data
Integrations
Data
Governance
Data
Admin
Data
Storage
Power
Query
Report
Development
Data Security
Semantic Model
Data Warehouse
CI/CD
…but Enterprise BI isn’t always an option
IT teams are increasingly petite
The Dream: A fully-stocked data team
Security
Admin
BI Project
manager
Power
BI/Fabric
Admin
Data
Engineer
DQ&G Data Quality &
Governance team
Data Viz &
Reporting
Specialists
Data Architect
…Data
team? What
data team?
The Reality?
IT teams are
increasingly petite
Moss
Roy
Jen
Even when Enterprise BI is an option, it’s not necessarily the best option
When the “dedicated” data team succumbs to “BI Island Syndrome…”
• Discouraged business users
• Reporting projects slow way down
• Prioritization becomes based on
“who you know”
Outdated
Outdated
Reporting
Reporting tools
tools
Frustrated end
users
Rogue BI
Team budget
cuts…
The Skullet: Self Service BI
Where most of us started with Power BI
Per
as of Dec
2024:you
“…analytical
JulyMicrosoft
24, 2015: “We’re
inviting
to disrupt the way you
responsibilities
handled
people
throughout
many
do business by are
bringing
youby
this
cloud-hosted,
business
areas
of the and
organization.
intelligence
analytics service that connects to your
data. We want you to experience the team’s mantra of,
The content creator has a lot of freedom and flexibility
to
create
content
for up,
individual
usage.”
‘five
seconds
to sign
five minutes
to wow!’”
07/24/2015 - Power BI is Generally Available today | Microsoft Power BI Blog | Microsoft Power BI
12/30/2024 - Power BI usage scenarios - Power BI | Microsoft Learn
How Self-Service BI goes wrong: who’s in charge?
Anarchy and data are not a good
combo
CI/CD
Data
Governance
Data
Security
Data
Admin
Data
Warehouse
Limited IT capacity & no
data team?
Anything that isn’t actively on
fire goes straight to the back
to the queue.
…even when you’ve got
that one guy who has been
thinking deeply about your
org’s data problems for a
while now.
Has anyone provided training or guidance? If not…
80% of your staff are
probably baffled…
…while 20% have gone full chaos demon just
seeing what they can whack together.
“Pure” Self-Service BI is hard to pull off in an office
Not unlike a skullet
Unless you’re Floki
“…. managed self-service BI is characterized by a
blended approach that emphasizes discipline at
the core and flexibility at the edge.
“The data architecture is usually maintained by a
single team of centralized BI experts, … “
while reporting responsibility belongs to creators
within departments or business units”
Power BI usage scenarios: Managed self-service BI - Power BI | Microsoft Learn
We can get the benefits of both
Self-Service BI
Managed Self Service BI
Enterprise BI
More control over data systems
More opportunity for innovation and “data kismet”
…while mitigating the risks of both
Data Chaos
(Potential)
Happy Middle
Ground
(Potential)
Data Island
What Makes Self Service BI Manageable?

1. Mullet Leadership
2. Centralized Tools
3. Ongoing alignment between the
   crewcut and skullet
4. ongoing capacity building
   Step 1: Mullet Leadership
   paves the way for successful data systems
   Define:
   In plain English
   Data stewardship
   Who’s responsible for validating a given data point
   and correcting it when the data is wrong?
   Data access rights
   Who should have access to different kinds of data?
   Data sensitivity rules - What data is considered sensitive

• Who gets to decide?
  Exception approval
  Who can approve access when someone feels
  they should be able to use a data point they
  wouldn’t normally see
  The trickiest data problems are people
  problems in disguise
  Some “Data Problems” that are really leadership gaps
  “We urgently need to limit access to this report, but we haven’t defined
  who should have access.”
  “This data is wrong, but no one is responsible for fixing it, so I just keep
  reporting it to IT.”
  “IT gave me Power BI access and now I’m supposed to build some urgent
  reports, but I have no idea how to use this tool.”
  “We all have a different number for net sales and honestly this meeting
  could have been a fist fight.”
  Create centralized reporting tools that shrink the report
  writing learning curve:
  Tasks required to build a Power BI Report “From Scratch”

1. Power Query (ETL, cleansing validation, etc.)
2. Semantic Model design (best practices, DAX, relationships…)
3. Build & evolve the report itself
   We want our report writers focused here
4. Format the report (& ensuring consistency with other reports)
   Use human-centric language in your
   Semantic Model Design
   ‘Sales’
   ‘dim_Sales’
   ‘Table19238’
   ‘Query1’
   Eliminate ambiguity wherever possible. Report writers
   shouldn’t need to guess which field to use.
   Hide what’s not helpful
   •Keys
   •Duplicate fields (remove when possible)
   rename, and/or describe
   •“Helper” Remove,
   disconnected
   tables
   Remove, rename,and/or describe
   Remove, rename,and/or describe
   Remove, rename,and/or describe
   Get Department-specific with hierarchies, folders and
   perspectives
   Finance Date
   Hierarchy:
   • FY2025
   • FY Q1 – 2025
   • Mar-2025
   • 03-01-2025
   Marketing Date
   Hierarchy
   -2025

• Q1 - 2025
  -March 2025
  -March 1st, 2025
  Demo
  Centralized semantic models: an opportunity to empower
  business users
  Include them: design
  “foundational” semantic model
  collaboratively, in detail*
  Manage expectations—this is a
  starting point that you’ll build on
  it together
  *Opportunity to make progress on
  data catalog
  Reduce the amount of staff time, training, and guesswork
  involved with report writing with themes, templates & style
  guides

1. Build a JSON theme
2. Create a custom “Thin” Power BI Report
3. Train using that report
   Reduce the amount of staff time, training, and guesswork
   involved with report writing with themes, templates & style
   guides
   •Pre-connected to semantic model
   •Pre-loaded with JSON theme, plus
   •Logo
   •Header
   •Hidden tabs can include: Style Guide, FAQ, “help” button, and/or
   links to submit new data requests)
   Demo
   Connect the dots using Purview’s Unified Catalog
   • Active Glossary Terms
   • Data Quality scores
   • Visibility with data lineage
   • Automated access request
   workflow
   Demo
   Build systems that support data requests
   (Without making your data team to manually direct them)
   Automatically route requests for:
   • New data or logic
   • Data quality corrections
   • Data access exceptions
   The tool isn’t important
   You just want something that
4. Minimizes guesswork and follow ups
5. Routes requests to the right people, and
6. Enables them to take action
   Guide your users - Get Specific
   Creating forms that collect actionable information
   •Who is requesting? Who needs access?
   •What should a field/measure be called?
   •When is it needed (and why?)
   •Where is the data now?
   (require screenshots & links, esp. for D365)
   •Why is it needed?
   This will minimize (but not eliminate) the need for meetings.
   Example: Daniel needs a new version of a customer name field
   from CE
   New data request form
   New dataPlease
   request
   form
   submit
   when new data
   Requestor
   needs are identified for the shared
   Field Name
   Data Source Link
   model
   1.semantic
   New data need
   (Select: measure, field, or table)
   Screenshot ProvidedData Steward
7. Hyperlink to existing data location
8. Attach screenshot with red box
   indicating specific field
9. Data Steward team (if known)
10. Please briefly describe business need
11. Anticipated teams who will use
    data (select all that apply
12. Additional notes
    Demo
    Sound off.
    The mic is all yours.
    Influence the product roadmap.
    Join the Fabric User Panel
    Join the SQL User Panel
    Share your feedback directly with our
    Fabric product group and researchers.
    Influence our SQL roadmap and ensure
    it meets your real-life needs
    https://aka.ms/JoinFabricUserPanel
    https://aka.ms/JoinSQLUserPanel