How to choose a real-world data partner: 15 questions for CROs and sponsors
Real-world data, feasibility and site selection vendors can sound alike. These are the questions that separate them, grouped by theme, with why each one matters and what a good answer sounds like.
- Check the data first: patient-level, across institutions, over time, de-identified.
- Then ask what you get back: data to analyze, or the finished answer.
- Make sure every number can be explained to a sponsor.
- Test the fit on a real deadline: days, not quarters.
- Match the commercial model to how often you actually need answers.
Every provider in this space uses real-world data. What differs is the kind of data, what you get back and who does the work. For a CRO preparing an RFP or a bid defense, those differences decide whether you walk in with an answer or a project.
Group 1The data
Start with what sits underneath the answer.
Is the data patient-level, or totals and averages?
Only patient-level data lets you count how many people meet a protocol, rather than how many have a diagnosis.
De-identified records for each patient, so counts can be made against your criteria.
Does it follow patients across every institution they visit?
Patients see many providers. Data from a sample of sites misses the rest of each patient’s care.
Coverage that links each patient’s care across institutions, with the gaps stated plainly.
Does it follow patients over time?
A single snapshot cannot show treatment history, which many criteria depend on.
A longitudinal view that can check criteria such as prior treatment lines.
How is privacy protected?
You need to know the data is de-identified before it reaches anyone.
Anonymized IDs instead of names, and counts rather than identities in what you receive.
Group 2The answer
Then ask what you actually get back.
Do we get data to analyze, or the finished answer?
This decides whether you need your own analysts, and how fast you get a result.
A clear statement of the deliverable: the counts, the ranked sites and the reasoning, or the raw data.
Can you count patients against our protocol’s criteria?
A disease total overstates the pool. Criteria can remove most of it.
Counts against the inclusion and exclusion criteria, not just the diagnosis.
Can you name the investigators, and their record on similar trials?
Institutions do not enroll patients; investigators do.
Named investigators, with their record on trials like yours.
Can every number be explained to a sponsor?
If you cannot explain a number in the room, it does not help you win.
A plain reason behind each site and each count, ready to repeat in a bid defense.
Group 3The fit with how you work
Make sure it works on a real deadline.
How long until we have a first answer?
RFPs and bid defenses run on days, not quarters.
A realistic turnaround measured in days for a defined deliverable.
Who does the work, your team or ours?
Mid-size teams rarely have spare analysts for a new dataset.
A clear split of who does what, ideally with the analysis done for you.
What does the result look like in front of a sponsor?
Your team should not have to rebuild it before the meeting.
Slides or tables in your own template, with the detail in an appendix.
Does it cover the countries in our plan?
Many trials run outside the US, where data is thinner.
An honest view of what each country’s data can support, and where it cannot.
Group 4The terms
Finally, the commercial questions.
Is it a license, or tied to each protocol?
A license is a standing commitment; a per-protocol deliverable is tied to the work in front of you.
A model that matches how often you actually need answers.
Can we see a real example first?
The best proof is a deliverable from a real program, with names removed.
A walk-through of a de-identified example, not a demo of a dashboard.
What happens to anything we share with you?
Protocols and trial history are confidential.
Clear confidentiality, and your material never used for anyone else.
For mid-size teamsWhat to weigh most
- Who does the analysis. If the answer is “your team”, budget the analyst time as well as the data.
- Time to a first answer. An RFP response cannot wait for a data onboarding project.
- What the sponsor sees. The result should go straight into your deck, with the reasoning attached.
- How you pay. A standing license suits steady, heavy use. Answers tied to each protocol suit work that comes in waves.
Choosing a real-world data partner, answered
How do you choose a real-world data provider for clinical trials?
Start with the decision you need to make, then ask four sets of questions: whether the data is patient-level and follows patients across institutions; whether you get data to analyze or the finished answer; whether it fits your deadlines and your sponsor deliverables; and how the commercial terms match how often you need answers.
What should a CRO look for in a feasibility or site selection vendor?
Counts against your protocol’s criteria rather than disease totals, named investigators with their record on similar trials, a reason behind every site that you can repeat to a sponsor, and a turnaround that fits RFP and bid defense deadlines.
Is it better to license real-world data or buy finished answers?
It depends on your team. Organizations with data science groups may want the raw data. Teams that need an answer for a specific protocol, without building a data team, usually get more from finished answers.
What is the difference between a real-world data vendor and a feasibility partner?
A real-world data vendor typically licenses data and tools for your team to analyze. A feasibility partner delivers the analysis itself: how many patients qualify, where they are treated, and which sites and investigators to choose.
What questions should a mid-size CRO ask before buying real-world data?
Who will do the analysis, how quickly you will have a first answer, whether the result is ready to show a sponsor, and whether the commercial model matches how many RFPs and bid defenses you actually handle.