Direct answer: In one recent U.S. Parkinson’s program we ran, 19,715 people completed a web screen. Two rounds of screening cut that to 1,006 site referrals, which became 200 signed consent forms and 71 randomized patients, 38.9% of the trial’s total enrollment. Every stage and conversion rate is below, because sponsors can’t calibrate their own enrollment plans against numbers nobody will show them.

Why we’re publishing our funnel

Recruitment vendors treat their funnels like trade secrets, which leaves sponsors planning enrollment with no base rates at all. Then the plan misses, and everyone acts surprised. Nearly 85% of interventional trials miss their enrollment milestones, and we think the secrecy is part of the reason. So here’s a real one, end to end. The sponsor and protocol stay confidential; the math is what matters anyway.

The funnel

Read it top to bottom once more. Nearly twenty thousand interested people, and the job was saying no to 19,644 of them so that sites only ever met the 1,006 worth meeting. That’s the part of recruitment nobody budgets for.

What each stage teaches

The web screen’s job is to say no. We screened out 69% of applicants before a human ever got involved. That’s not lost volume, that’s protection: every one of those people would otherwise have burned a coordinator’s hour or a screener’s call. The funnel narrows on purpose.

The phone screen says no again. Of the people the web screen let through, the phone screen removed another two thirds. Two independent filters, two chances to catch a disqualifier before a site visit gets scheduled.

Speed keeps the middle of the funnel alive. Every qualified web screen got a phone call within 90 seconds, from a recognizable number. That’s most of why 63% of web-screen passes actually completed a phone screen instead of ghosting. Interest decays by the hour; we don’t give it an hour.

The site handoff is where trials usually bleed, so we staffed it. After a phone-screen pass, every patient got a dedicated Patient Advocate: answering questions, retrieving medical records (56.6% of referrals arrived with EMR already in hand), and keeping a 72-hour touchpoint cadence until the site visit. The average from completed call screen to signed ICF was 55 days. That’s 55 days of chances for a patient to quietly drop out, which is exactly why the touchpoints exist.

Concentration is normal. 75% of the sites we referred to produced at least one randomization from our referrals, and at 14 sites our referrals made up more than half of ALL their randomizations. But production was not evenly spread, it never is. Site-level tracking is what tells you where to route the next referral.

Where the funnel still leaked, honestly

Even with double-screening, 23.9% of on-site screens failed. Not zero. Some disqualifiers only surface in labs and in-person assessments, and no phone screen catches them. And a quarter of our referred sites never converted a referral to a randomization; some of that is site capacity, some is geography, some is us. If a vendor tells you their funnel has no leaks, they’re not showing you their funnel.

What this funnel added up to

This program delivered 38.9% of the trial’s total enrollment and 50.63% of its signed ICFs. Across our programs, starting at trial launch, we typically deliver 32% of a phase 2 or phase 3 trial’s total enrollment, and we’ve been as high as 47%. We’re paid on enrolled patients, so every number in this article is a number we had to hit to get paid.

Frequently asked questions