A study that fails at design fails expensively
Every study has a set of assumptions that have to hold. Testing them costs weeks. Discovering they do not hold, after analysis, costs the programme.
Offering 2 of 5
One precise question, tested against the data before you fund it.
We take a single question and establish whether the data supports it: can the cohort be identified, is the outcome captured, is the denominator right, is the follow-up long enough, and what sample would you actually get. In writing, before money moves.
Feasibility & Cohort Scoping
Why this is where most engagements start
You find out the sample size before you commit
The most common quiet failure in real-world evidence is a cohort that turns out to be a tenth of the expected size. This finds that in week two.
It converts a vague question into a specification
Most questions arrive as a sentence. What a study needs is a cohort definition, an index date, an outcome definition and a comparator. Producing those is the work.
Sometimes the honest answer is no
And that is the finding you are paying for. A clean no in week three is worth more than a compromised yes in month six.
2
Feasibility & Cohort Scoping
We take a single question and establish whether the data supports it: can the cohort be identified, is the outcome captured, is the denominator right, is the follow-up long enough, and what sample would you actually get.
- Timeline
- Two to three weeks where access is established.
- Buy it when
- You have a study in mind and want to know whether it survives design before you fund it, or a payer or regulator has asked for evidence and you need to know if local data can produce it in time.
- Do not buy it when
- You do not yet know which market or which data type. Start with the landscape assessment, which is offering one.
Deliverables
What you actually receive.
Named, dated and agreed at scoping. Every one of these is a thing you can hold, circulate internally and be held to, not a description of an activity.
A written feasibility view
Is the question answerable, in which market, from which data type, and how fast. One page at the front, the reasoning behind it.
A draft cohort specification
Inclusion and exclusion criteria, index date definition, follow-up window and comparator, written so it can be executed by an analyst who was not in the room.
An expected sample estimate
With the assumptions that drive it stated explicitly, so you can see what would make the number move.
An outcome and endpoint assessment
Whether each endpoint you want is actually captured in the source, at what completeness, and what has to be inferred rather than observed.
A limitations statement
Written at the start rather than discovered in review. A limitation you predicted is a caveat; one you did not is a finding against you.
A go, no-go or redesign recommendation
And where the answer is redesign, the specific change that would make the question answerable.
How it runs
Four steps, and you see the direction before you see the deck.
Define the question
We turn a sentence into a testable specification. This is the step clients most often underestimate and most often thank us for.
Test against the source
We check cohort identifiability, outcome capture, denominator validity and follow-up length against the actual data structure.
Size it
An expected sample with its assumptions exposed, including the subgroups you will want later.
Report
A written feasibility view with a clear verdict, and a working session to go through it.
Alongside every engagement: weekly updates, methodology agreed before work starts, quality control at every phase, interim reporting rather than a single reveal, and milestones you accept rather than discover.
What it feeds into
Analysis & Insight
The cohort specification becomes the analysis specification. No redefinition, no drift, no argument about what was agreed.
Is this the one you need?
Describe the decision you are trying to make. If a cheaper offering answers it, we will tell you that instead.