Brian Geier | Driftless Data Partners
Turn data into decisions you can trust.
I help leaders work through data and AI questions that are holding up the business: which investments are useful, why a workflow still needs so much manual work, or what the numbers mean for the next decision.
We start with the specific problem, agree what needs to change, and how to carry the work forward.
I refined this approach over more than a decade leading analytics teams through complex, cross-functional business problems.
Where I can help
Three situations worth a closer look.
None of these means a team is doing something wrong. Capable data, finance, and operations leaders run into them too, often because the work crosses teams, systems, and definitions that no one person owns.
AI activity is growing, but useful results are hard to show.
Pilots and tools add up, yet it is hard to say which ones deserve more investment.
This might look like
- Someone still checks AI answers by hand before anyone acts on them.
- Two tools give different answers to the same question about performance.
- The next phase needs funding, but there is no agreed way to judge the last one.
Where I come in
Agree what a useful result looks like, how to check it, and what is worth continuing.
The same workflow keeps needing manual correction.
Skilled people spend their time matching, fixing, and handing off work instead of using the result.
This might look like
- Contact and account records get matched by hand before a campaign or forecast.
- A recurring report needs the same corrections every cycle.
- Work waits at the handoff between two teams or systems.
Where I come in
Find where the work breaks down, decide what should change, and agree who owns the fix.
The reports are clear, but the next action is not.
Leaders can see what happened but do not have enough evidence to choose what to do next.
This might look like
- A hiring or spending decision rests on assumptions no one has tested.
- The same numbers support two different courses of action.
- Results moved, and it is unclear which change caused it.
Where I come in
Lay out the assumptions behind the decision, test the ones that matter, and show what the evidence supports.
Shared business logic
How a semantic layer connects business data and AI
A semantic layer is a shared model of your business definitions, calculations, relationships, and context. Connected reporting and AI tools can use it, so they work from the same rules instead of each keeping its own version.
1. Leadership agrees
What counts as an active customer
And how retention is calculated, who owns each definition, and how changes get approved.
2. The semantic layer holds it
Defined once, in one place
The calculation, the data it draws on, and the business context are written down once instead of rebuilt in every dashboard, spreadsheet, or prompt.
3. Connected tools reuse it
Dashboards and AI use the same rules
When someone asks about retention in a connected dashboard or AI assistant, both start from the agreed definition.
The Results
I build the systems that find the money.
Before Driftless, I led data and analytics work inside growing companies where better decisions produced measurable business results.
That experience is the foundation of how I work today.
“Brian built the trusted data foundation that became a strategic advantage for the business. Whether providing data for the diligence process, redesigning pricing, preparing board-level analysis, or managing GTM performance, his work enabled decisions that once took weeks to happen in days or hours.”— Chris Wong, VP of FP&A at Recurly
Priced risk with precision
$50M+ in incremental LTV created annually
DISH
I built a credit-based pricing model that repriced the margin-destroying cohorts: fewer bad-LTV signups, higher LTV on the rest.
Churn down, revenue retained
$5M/yr retained
Sling TV
I drove the experimentation framework that isolated which product changes actually cut churn: 4% lower, worth $5M a year.
An engine that turned data into deals
$3.5M+/yr net-new ARR
Recurly
My team built an engine that showed each prospect what they were leaving on the table: a vague pitch became a hard number that closed deals.
The right lapsed customers, won back
+25% lift in win-back subscribers at Naked Wines
Naked Wines
I directed a predictive model that re-targeted only the lapsed buyers worth winning back: proven against a holdout control.
How this works
Each engagement is available on its own. Where we start depends on what is already clear.
Focused assessment
For when the problem is real but the cause or the right approach is not yet clear. We look closely at the specific workflow, decision, or AI question and agree what needs to change, who owns it, and what comes first.
Scoped to the question in front of you.
See how the assessment worksAlignment & execution
For when the priority is clear but progress depends on agreeing definitions, owners, and next steps across teams. I lead that work and help move the first improvements forward.
Scoped to the situation.
See how alignment & execution worksFractional data leadership
For ongoing, executive-level ownership of data priorities, tradeoffs, and delivery. I work alongside your existing executives, data leaders, and implementation teams.
A cadence matched to the business.
See how fractional leadership worksNot sure which fits? That is what the first conversation is for.
Pick a Time