Data science consulting
I start from the decision, pick the analysis that answers it, and say plainly how sure we can be. If you want, I run the first analysis myself.
Book an intro call
Always included
Architect-lead
Add when you want it
Hands-on build
You choose
Which parts
What I build
How a project runs
The question
A clear question and the data needed to answer it.
The analysis
An analysis matched to the question, such as cohorts, funnels or forecasts.
The findings
Findings with the uncertainty stated, not hidden.
The recommendation
A recommendation tied to the decision, and what to measure next.
How it works
Plan the analysis, or plan and run it
First, always
Architect-lead
I frame the question, choose the analysis, and direct whoever runs it, your team or a vendor.
Question framing for the decisions you pick, the analysis plan and success measures, and a review of the findings
Then, added on top
Architect-lead plus hands-on build
Everything in part one, and I run the analyses you choose myself. Costs more, because it is more of my time.
Completed analyses, not only a plan, with a write-up your team can act on and repeat
Hands-on work always comes with the architect-lead work. You can take the analysis plan on its own.
Proof
Analyses I have done
1
Questions answered across a subscription business
Lifecycle, funnel and cohort questions answered from customer data, alongside a dozen dashboards built for the same business.
2
A forecast the team can test its plans against
A three-year forecast built from conversion, retention and acquisition data, with planned features entered as expected lifts and the model rerun. Used for annual and long-range planning.
3
Comparing products on one set of measures
Built at multiple companies. A standard set of top-line KPIs across products, so each is judged the same way and the next level of detail is there when something looks off.
4
Testing the assumptions before building
User research and other data used to test the forecast assumptions and the market appetite for what was being built.
Fit
Who this suits
Strong fit
You have a decision to make and data to look at, and need someone to find what the data can and cannot tell you.
Weak fit
You want a conclusion fixed in advance. If you are not sure what is wrong, start with the Bottleneck Diagnostic.
Questions