Planning a BI project
How to Develop a Data Strategy Roadmap
A data strategy roadmap sequences work into phases, each with a deliverable and a gate that decides when to start the next one. The four phases explained.
Sajagan Thirugnanam
·
Updated
A data strategy roadmap sequences data work into phases, each with a defined deliverable and a gate that decides whether to move to the next one. It is not a list of every initiative a company could run. A roadmap with no order attached is a wish list, and a wish list gets reprioritized every quarter with nothing finished.
For the structure each phase builds toward, see how to build a data strategy framework. For turning goals into the KPIs a roadmap should be built around, see how to create an analytics strategy.
Four phases, each with a gate
Each phase below has a stated deliverable and a gate: a condition that has to be true before starting the next phase. Moving to phase two before phase one's gate is met is the most common reason a roadmap stalls.
Phase | What it delivers | Gate to move on |
|---|---|---|
0. Assess | A source inventory and a data maturity gap list | An owner is assigned to every gap on the list |
1. Foundation | One platform, governance rules, one accurate report | The first report is trusted enough to replace a manual spreadsheet |
2. Expand | A shared semantic model reused across teams, row-level security | Two or more teams have self-serve reports live on it |
3. Scale | Broader self-service, predictive measures, wider automation | Phase 2's reports are stable with no recurring data-quality fixes |
Phase 0: assess before building anything
Assessment means listing every source system, who owns it, and where the current data breaks: missing fields, duplicate records, no shared customer identifier across systems. The deliverable is a gap list with an owner assigned to each gap, not a maturity score with no names attached to it.
Phase 1: build the foundation
This is where the platform and governance layers get built: one place data comes to rest, row-level security rules, and a first semantic model. The gate for this phase is concrete: the first report built on it has to be trusted enough that someone stops maintaining the spreadsheet it replaces. A foundation phase that never produces a report people actually use has not met its own gate, no matter how complete the underlying platform is.
A data quality measure is a practical way to track this phase's exit criteria directly in Power BI:
Rows Failing Validation % =
DIVIDE (
CALCULATE ( COUNTROWS ( Sales ), Sales[ValidationFlag] = FALSE () ),
COUNTROWS ( Sales )
)Watching this measure trend down across the foundation phase is a better signal that the platform is ready than a fixed calendar date.
Phase 2: expand to more teams
Once one report is trusted, the same semantic model extends to a second and third team, each with its own row-level security rule instead of its own copy of the data. Fabric deployment pipelines move a validated report from a development workspace through test and into production, so a change gets checked before it reaches the teams already depending on it.
The gate here is usage, not completion: two or more teams need self-serve reports live and in regular use before phase three starts. Building a report nobody opens does not meet this gate.
Phase 3: scale and automate
With the foundation stable and several teams self-serving, the roadmap can turn to predictive measures, wider automation of refresh schedules, and self-service extending past the teams that asked for it first. This phase depends entirely on phase 2 being stable. Starting it while phase 2's reports still need weekly data-quality fixes means building automation on top of a platform that is not ready for it.
Deciding when to move to the next phase
Use the gate column in the table above as a checklist, not the calendar. A roadmap that measures progress by dates instead of by the gates tends to move on schedule and stall in practice. Phase 2 quietly starts before phase 1's report is actually trusted. Data governance rules set in phase 1 are what make phase 2's row-level security expansion possible without rebuilding access control per team.
FAQs
What is a data strategy roadmap?
A sequenced plan with four phases, assess, build the foundation, expand to more teams, then scale, each with a deliverable and a gate that decides when the next phase starts.
Why is assessing current capabilities important?
Skipping the assessment phase means building a foundation on top of source data whose real problems, missing fields, duplicate records, no shared identifiers, are still unknown. Phase 1 work built on that foundation inherits every one of those problems.
How often should I review my data strategy roadmap?
Review it at each phase gate, not on a fixed calendar. If a phase's gate has not been met after several months, that is a signal to fix what is blocking it rather than moving to the next phase regardless. A related question, what a data strategy consulting engagement actually covers, is answered in our data strategy consulting guide.
Sources
Fabric deployment pipelines - Microsoft Learn
A data strategy roadmap sequences data work into phases, each with a defined deliverable and a gate that decides whether to move to the next one. It is not a list of every initiative a company could run. A roadmap with no order attached is a wish list, and a wish list gets reprioritized every quarter with nothing finished.
For the structure each phase builds toward, see how to build a data strategy framework. For turning goals into the KPIs a roadmap should be built around, see how to create an analytics strategy.
Four phases, each with a gate
Each phase below has a stated deliverable and a gate: a condition that has to be true before starting the next phase. Moving to phase two before phase one's gate is met is the most common reason a roadmap stalls.
Phase | What it delivers | Gate to move on |
|---|---|---|
0. Assess | A source inventory and a data maturity gap list | An owner is assigned to every gap on the list |
1. Foundation | One platform, governance rules, one accurate report | The first report is trusted enough to replace a manual spreadsheet |
2. Expand | A shared semantic model reused across teams, row-level security | Two or more teams have self-serve reports live on it |
3. Scale | Broader self-service, predictive measures, wider automation | Phase 2's reports are stable with no recurring data-quality fixes |
Phase 0: assess before building anything
Assessment means listing every source system, who owns it, and where the current data breaks: missing fields, duplicate records, no shared customer identifier across systems. The deliverable is a gap list with an owner assigned to each gap, not a maturity score with no names attached to it.
Phase 1: build the foundation
This is where the platform and governance layers get built: one place data comes to rest, row-level security rules, and a first semantic model. The gate for this phase is concrete: the first report built on it has to be trusted enough that someone stops maintaining the spreadsheet it replaces. A foundation phase that never produces a report people actually use has not met its own gate, no matter how complete the underlying platform is.
A data quality measure is a practical way to track this phase's exit criteria directly in Power BI:
Rows Failing Validation % =
DIVIDE (
CALCULATE ( COUNTROWS ( Sales ), Sales[ValidationFlag] = FALSE () ),
COUNTROWS ( Sales )
)Watching this measure trend down across the foundation phase is a better signal that the platform is ready than a fixed calendar date.
Phase 2: expand to more teams
Once one report is trusted, the same semantic model extends to a second and third team, each with its own row-level security rule instead of its own copy of the data. Fabric deployment pipelines move a validated report from a development workspace through test and into production, so a change gets checked before it reaches the teams already depending on it.
The gate here is usage, not completion: two or more teams need self-serve reports live and in regular use before phase three starts. Building a report nobody opens does not meet this gate.
Phase 3: scale and automate
With the foundation stable and several teams self-serving, the roadmap can turn to predictive measures, wider automation of refresh schedules, and self-service extending past the teams that asked for it first. This phase depends entirely on phase 2 being stable. Starting it while phase 2's reports still need weekly data-quality fixes means building automation on top of a platform that is not ready for it.
Deciding when to move to the next phase
Use the gate column in the table above as a checklist, not the calendar. A roadmap that measures progress by dates instead of by the gates tends to move on schedule and stall in practice. Phase 2 quietly starts before phase 1's report is actually trusted. Data governance rules set in phase 1 are what make phase 2's row-level security expansion possible without rebuilding access control per team.
FAQs
What is a data strategy roadmap?
A sequenced plan with four phases, assess, build the foundation, expand to more teams, then scale, each with a deliverable and a gate that decides when the next phase starts.
Why is assessing current capabilities important?
Skipping the assessment phase means building a foundation on top of source data whose real problems, missing fields, duplicate records, no shared identifiers, are still unknown. Phase 1 work built on that foundation inherits every one of those problems.
How often should I review my data strategy roadmap?
Review it at each phase gate, not on a fixed calendar. If a phase's gate has not been met after several months, that is a signal to fix what is blocking it rather than moving to the next phase regardless. A related question, what a data strategy consulting engagement actually covers, is answered in our data strategy consulting guide.
Sources
Fabric deployment pipelines - Microsoft Learn
Show us your report nobody trusts.
1 · A 30-minute call.
2 · We look at your current reports together.
3 · We tell you what we would do.
Show us your report nobody trusts.
1 · A 30-minute call.
2 · We look at your current reports together.
3 · We tell you what we would do.
Show us your report nobody trusts.
1 · A 30-minute call.
2 · We look at your current reports together.
3 · We tell you what we would do.
CaseWhen is a Berlin BI consultancy that builds reporting that leaders can trust, on the Microsoft stack: Power BI, Fabric and Azure.
Berlin, Germany
Free tools
CaseWhen is a Berlin BI consultancy that builds reporting that leaders can trust, on the Microsoft stack: Power BI, Fabric and Azure.
Berlin, Germany
Free tools
CaseWhen is a Berlin BI consultancy that builds reporting that leaders can trust, on the Microsoft stack: Power BI, Fabric and Azure.
Berlin, Germany