Which recruitment metrics decide site selection in clinical trials?

Site selection has changed. Sponsors and contract research organizations (CROs) that once chose sites largely on the strength of the principal investigator, therapeutic experience, and long-standing relationships now score sites against hard performance data. The feasibility questionnaire (FQ) is still the front door, but the answers on it get cross-checked against a site's track record, third-party databases, and internal sponsor performance records before a study is awarded.

For research sites, that shift creates a straightforward challenge and a straightforward opportunity. The sites that consistently win the next study are the ones that measure their own recruitment performance the way sponsors do, present it clearly, and back it up with delivery. This piece walks through the recruitment metrics that carry the most weight in site selection, the ranges sponsors expect to see, and how a site should organize its own data to compete.

Why past performance carries the most weight

Sponsors treat prior enrollment behavior as the strongest available signal of future enrollment behavior. A site that delivered on its projections in a comparable study is materially more likely to deliver on the next; a site that under-delivered tends to keep under-delivering. Retrospective analyses across large trial portfolios have shown that in a typical Phase II or Phase III study, roughly one in ten sites enrolls no patients at all, close to four in ten under-enroll, and only a minority meets or exceeds targets. Every criterion sponsors apply during site selection exists to screen out that under-performing group.

That is why the phrase "delivery on commitment" has become a shorthand for the meta-metric of site selection. Individual numbers like enrollment rate or screen failure rate matter, but they matter most when read together as evidence that a site does what it said it would do. Sites that understand this frame prepare accordingly. They organize their historical numbers by therapeutic area and study phase, they present projections against actual outcomes, and they resist the pull to over-promise on any given FQ.

For a longer treatment of how sites can package performance data for sponsors, see Building sponsor trust: metrics that show site performance. The core idea is that trust in a research site is built on visible, verifiable numbers rather than reputation alone.

Delivery on commitment: projected versus actual enrollment

The single most decisive number a site can put in front of a sponsor is the ratio of actual enrollment to what was projected on the prior FQ. Sponsors compare that ratio across the sites they considered, and chronic over-promising is penalized more heavily than a conservative but met commitment. A site that projected fifteen participants and enrolled fourteen is remembered differently than one that projected thirty and enrolled twelve, even though both sites contributed similar volume.

The practical implication is that recruitment forecasting is itself a performance metric. Sites should build projections from patient database counts, historical conversion at each funnel stage, and known seasonality, not from a target that sounds attractive on the FQ. A one-page summary that shows projected versus actual for the last three to five completed studies, organized by therapeutic area, tends to carry more weight in business development conversations than any single top-line rate.

Funnel-level projections are only as trustworthy as the data feeding them. The related piece on pre-screening funnel metrics: from clinical trial participant to enrollment walks through which conversion rates matter at each stage of the funnel and how they roll up to a projection a sponsor can trust.

Enrollment rate and time to first patient

Two speed metrics sit at the top of the sponsor scorecard: the site-activation-to-first-patient cycle time, and the enrollment rate per site per month once the site is active. Both are read as leading indicators of overall performance. Industry analyses of large site portfolios consistently show that the faster a site screens a participant after activation, the more participants it ultimately enrolls, and the lower its protocol-deviation rate tends to be.

Enrollment rate itself is highly protocol-dependent. Oncology, central nervous system (CNS) trials, and rare-disease studies routinely run at fractions of the enrollment rates seen in more common conditions, and any benchmark quoted without a therapeutic-area qualifier is close to meaningless. What sites should track is their own rate against comparable prior studies and against the study-specific projection they made on the FQ. Sponsors read the trend line as much as the number.

Site activation timelines matter for the same reason. Study start-up (SSU) remains one of the longest and most variable phases of a clinical trial, with contracting and budget negotiation as the largest bottlenecks industry-wide. Sites that have documented workflows, standing regulatory binders, and pre-approved template language shave weeks off the cycle and demonstrate operational readiness before a sponsor ever visits.

The blog on time-to-first-patient (FPI): the most expensive phase of a clinical trial covers why sponsors treat this window as a leading indicator and what specifically slows it down at the site level.

What screen failure rate sponsors want to see

The screen failure rate (SFR) is the percentage of consented, screened participants who do not go on to enroll. Sponsors read it as a signal of pre-screening quality and inclusion/exclusion (I/E) targeting, and the range they consider acceptable varies widely by therapeutic area. Recent multi-year analyses have placed the all-therapeutic-area average around 36%, with CNS studies running considerably higher (in the mid-50% range), and oncology and rare disease often higher still.

The winning position is at or just below the study's therapeutic-area mean. Both extremes raise flags. An SFR that is far above the mean signals that the site is consenting participants who were never plausibly eligible, which imposes screening cost on the sponsor and drags on timeline. An SFR that is unusually low can look equally troubling, since it suggests the site may be applying eligibility criteria loosely at pre-screening or missing exclusions that will surface later as protocol deviations. Presenting SFR as a raw number without therapeutic-area context is a common self-inflicted wound; presenting it relative to the study mean or a peer benchmark is much stronger.

For a site-level operational treatment of how to bring SFR into the acceptable range, see top 5 ways to reduce screen failures at research sites.

Retention and dropout as quality signals

Retention is the percentage of enrolled participants who complete the study; dropout is its inverse. Sponsors care about both because a participant who drops out mid-study is, from a data perspective, expensive to have enrolled in the first place. The commonly cited industry average for retention sits close to 88%, with dropout rates in the range of 15% to 30% across most therapeutic areas and higher in long-duration or high-burden protocols. Recent multi-year analyses have shown dropout rates trending upward, which has pushed sponsors to weight retention more heavily in selection.

At the site level, retention is not just a downstream outcome of protocol design. Site behavior, including participant communication, visit scheduling flexibility, and coordinator continuity, moves the number substantially. Sites that track retention by study and can point to specific practices that improve it (proactive visit reminders, transportation support, coordinator follow-up cadence) demonstrate exactly the operational maturity sponsors are looking for.

The blog on 5 reasons patient enrollment and retention are failing clinical trials walks through the structural drivers that pull retention numbers down and where sites have leverage to change them.

Recruitment funnel conversion from referral to randomization

A single top-line rate rarely tells a sponsor what it needs to know. What sponsors increasingly ask for, and what more sophisticated site-selection platforms increasingly score, is the site's conversion at each stage of the recruitment funnel: referrals received, pre-screened, screened, consented, enrolled, and randomized. The widely used industry benchmark is roughly ten to one, meaning that identifying and reaching about ten patients at the top of the funnel typically yields one randomization.

The value of funnel-level tracking is that it isolates where performance breaks down. A site with a strong top of funnel but weak pre-screen-to-consent conversion has a different problem, and needs a different intervention, than a site whose consent-to-enrollment rate is fine but whose referral volume is too low. Presenting the funnel with the drop-off at each stage, ideally with recruitment-source attribution, is one of the most operationally credible artifacts a site can bring to a business development conversation.

Diversity of the enrolled population enters the picture here. Under the FDA Diversity Action Plan framework, sponsors are now required to describe upfront how they will achieve representation goals for pivotal trials, and they look for sites with a documented ability to recruit diverse participants. Capturing race, ethnicity, sex, and age at pre-screening (rather than reconstructing them at enrollment) is what makes it possible to present diversity as a live metric.

The blog on from external referral to randomization: a stage-by-stage look at the site eligibility funnel unpacks each stage and the conversion signals sponsors read at each one.

Presenting recruitment metrics for the next study award

Knowing which metrics matter is the first half of the problem. The second is having them documented, current, and packaged for the moments where sponsors evaluate sites, primarily the FQ, the site qualification visit (SQV), and the business development conversation that precedes both.

A standing site performance dashboard, updated continuously rather than reconstructed under FQ deadline, is the operational foundation. At minimum it should track the funnel per study (source, pre-screened, consented, screened, enrolled, randomized, completed), diversity fields captured at pre-screening, enrollment rate per month, projected versus actual enrollment for completed studies, SFR relative to study and therapeutic-area means, activation-to-first-patient cycle time, and retention. When an FQ arrives, the site pulls answers from a dashboard rather than assembling them from scratch, which is both faster and materially more accurate.

A one-page delivery-on-commitment sheet per therapeutic area is the highest-leverage business development artifact a site can build. It shows sponsors, at a glance, that the site knows its own numbers and delivers against them. Pair it with a quantified estimate of the pre-qualified patient database, and a newer site with a smaller trial history can compete credibly against more established sites on the criterion sponsors weight most heavily: patient access.

For sites that receive participants through DecenTrialz, the pre-screening data captured during participant matching becomes an additional input into the site's dashboard. Structured pre-screening records (with source attribution, diversity fields, and eligibility signals) make it easier to show sponsors the top of the funnel with the same rigor as the enrollment stage. Sites interested in bringing this data into their own performance reporting can learn more at decentrialz.com.

Compete on data, not just experience

Site selection is now a data-driven process, and it will keep moving in that direction as sponsor internal databases and third-party site-performance platforms mature. Sites that treat recruitment as a measured funnel, that document delivery on commitment by therapeutic area, and that walk into every SQV with current numbers in hand will win a disproportionate share of the studies they compete for. Sites that continue to lead with experience and reputation alone will find themselves scored against the numbers anyway, and often by algorithms that never asked to see them.

DecenTrialz supports research sites with AI-assisted participant matching and registered nurse-led pre-screening, delivering pre-screened participants along with the funnel and diversity data sites need to compete on performance. Final eligibility determination, informed consent, study walk-through, and enrollment are always handled by the research site team. To see how the platform fits into a site's recruitment workflow, visit decentrialz.com.