I have presented and posted on this topic for years, and it remains one of the most important realities in clinical research. Not all sites are the same. Not all patient databases are meaningful. And access to patients does not automatically translate into enrolled and retained study participants.
Here, I take a deeper look at what realistically separates consistently high-enrolling sites from the rest, and why sponsors and CROs need to look beyond site size, database counts, technology platforms, and optimistic feasibility projections.
The Feasibility Mirage
There is a persistent assumption in clinical research that patient access is the same thing as patient enrollment. It is not. A site may have thousands of patients in an EHR. A health system may have millions of patient records. A large multi-site company may have broad geographic reach and impressive database counts. Those things matter, but they do not, by themselves, predict whether the site can enroll the right patients, at the right pace, under the requirements of a specific protocol.
The evidence is clear that recruitment remains one of the most persistent causes of trial delay. A systematic review of online patient recruitment noted that around 80% of trials fail to meet initial enrollment targets or timelines, with consequences that are scientific, economic, and ethical in nature. [1] Tufts CSDD research, as reported in Applied Clinical Trials, has also shown that a substantial percentage of activated sites fail to reach target enrollment, reinforcing what many of us have seen repeatedly in the field: activation does not equal performance. [2]
That distinction is at the heart of one of the most important challenges in clinical research today. The industry does not simply need more sites. It needs better insight into which sites can actually execute.
Patient Access Is Not Patient Convertibility
When sponsors and CROs evaluate sites, the first question is often some version of, ‘How many patients do you have?’ That is a reasonable starting point, but it is an incomplete question. A better question is: How many of those patients are actually eligible, reachable, interested, logistically able, and willing to participate in clinical research?
There is a meaningful difference between theoretical patients and enrollable patients. Theoretical patients appear in a database. Potentially eligible patients may meet broad diagnosis criteria. Pre-screenable patients have enough available information to evaluate key inclusion and exclusion criteria. Reachable patients have current contact information and a relationship strong enough to respond. Research-inclined patients have previously indicated a willingness to consider clinical research. Enrollable patients meet criteria, consent, randomize, and remain in the study. That is the real funnel.
This is not simply a philosophical distinction. Research on site selection increasingly supports the need to combine real-world patient data with site-level recruitment history and performance data. One 2024 machine-learning study found that incorporating both site-level recruitment data and real-world patient data improved site ranking compared with common industry baselines. [3] That finding supports what many experienced feasibility professionals already know: patient counts alone are not enough. A database may help identify possible patients, but past site behavior and operational execution are essential to understanding enrollment potential.
Too often, feasibility stops at the first or second step and mistakes that for enrollment potential. That is how a large integrated database of ‘patients’ can become what I call a feasibility mirage. The records may be real. The diagnoses may be real. The population estimate may even be directionally useful. But unless those patients are reachable, eligible, interested, and willing to participate, the database is not enrollment intelligence. It is raw inventory.
High-Enrolling Sites Are Built Differently
Consistently high-enrolling sites are rarely successful because of one factor. They usually enroll well because they have built a system. They understand their patient population from lived operational experience, not simply database counts. They have a research culture, not a research hobby. They maintain a warm database of patients in their community who have already shown some inclination toward research participation.
They know which patients have participated before, which patients declined and why, which patients need transportation support, which patients prefer text messages, which patients require evening visits, and which patient populations are already overcommitted to competing studies. They may increasingly use CTMS platforms, EHR tools, recruitment technology, analytics, and AI to identify, track, and communicate more efficiently, but those tools are not the reason they succeed. The core was already there: relationships, discipline, trust, process, and follow-through.
That point matters because patient willingness is not created by technology. A review of barriers and facilitators to participation in clinical trials found that participation is influenced by patient, physician, and system-level factors, including trust, communication, understanding of the trial, perceived benefit and risk, logistics, and the role of the healthcare professional in introducing or supporting the opportunity. [4] These are human and operational realities before they are technical ones.
High-performing sites have coordinators who function as operational quarterbacks, not simply task processors. They also have investigators who do more than sign documents and attend initiation visits. The investigator actively supports recruitment, talks about research with patients, reviews potential subjects, and helps create confidence in the study. These sites begin preparing before activation, track the recruitment funnel, understand where patients are dropping out, communicate specifics to sponsors and CROs, and retain patients because they treat participation as a relationship, not a transaction.
In other words, high-enrolling sites are not merely sites with access to patients or access to better software. They are sites with the operational ability to convert appropriate patient access into enrolled and retained participants. Technology can help them do that more efficiently, but it does not create the underlying capability.
Under-Enrolling Sites Often Start With the Wrong Assumption
Sites that consistently miss enrollment expectations often begin with a flawed premise. They believe access equals enrollment. They may say, ‘We have 2,000 patients with this condition,’ or ‘Our physicians see these patients every week.’ That may be true, but it does not answer the harder questions.
How many of those patients meet the major inclusion and exclusion criteria? How many are currently reachable? How many have previously expressed interest in clinical research? How many are already in competing studies? How many would accept the visit schedule, procedures, washout, placebo risk, or travel burden? How many can realistically be contacted in the first two weeks after activation? How many can be screened without overwhelming the coordinator? How many are likely to randomize? How many are likely to complete the study?
When those questions are not answered honestly, feasibility becomes optimistic forecasting. And optimistic forecasting is one of the fastest ways to create enrollment failure. Adding a new CTMS, EHR search tool, AI model, or patient matching platform may improve visibility, but it will not fix unrealistic assumptions, weak patient relationships, insufficient staffing, poor follow-up, or lack of site discipline.
This aligns with broader research on why trials fail. A review of factors associated with failed clinical trials noted that slow recruitment may result from inadequate staffing, lack of prioritization of the trial over day-to-day operations, investigator competition from other trials, and failure to tailor the protocol to practices at the study center. [5] Those are not database problems. They are operational problems.
The Best Sites Are Often Conservative During Feasibility
One of the more important differences between high-performing and underperforming sites is that strong sites are often more conservative during feasibility. That may seem counterintuitive. Sponsors want enrollment. CROs want enrollment. Sites want awards. But mature research sites understand that overpromising creates problems for everyone.
It damages credibility, strains staff, frustrates sponsors, and can turn a study into rescue mode before the site has had a fair chance to succeed. Strong sites protect their reputation by giving realistic numbers. They ask whether the protocol fits their actual patient population, look at key exclusions, evaluate competing studies, assess visit burden, and consider whether patients will accept the procedures, the washout, the placebo risk, the stipend, and the schedule.
This becomes more important as protocols become more complex. Tufts CSDD and others have reported increasing protocol complexity and rising participation burden, including more procedures, endpoints, eligibility criteria, and operational workload for sites. [6] As complexity increases, enrollment becomes less about how many patients carry a diagnosis code and more about whether real people can realistically participate in the study as designed.
Strong sites also evaluate coordinator capacity, know whether the investigator will truly support recruitment, and understand the difference between a best-case estimate and a likely-case estimate. Underperforming sites often give the number they think the sponsor wants to hear. High-performing sites give the number they believe they can achieve. That difference matters.
Sadly, too often, those sites best equipped to meet enrollment expectations are passed over during the initial wave of site selection because they honestly responded to narrowly designed feasibility questionnaires, and are eliminated by the algorithms that assume every site will over project their ability to enroll.
Dedicated Independent Sites Still Have a Critical Role
There is a place for large multi-site research companies. They can bring scale, infrastructure, geographic coverage, shared systems, centralized services, technology investment, and operational efficiencies. Those advantages are real. But scale and technology do not automatically equal lower sponsor cost, better recruitment, better retention, or better execution at the individual site level.
There is also a strong and necessary place for dedicated, independent research sites. The key word is dedicated. The independent site that succeeds today cannot simply be a medical practice with clinical research attached as a side function. It must be genuinely committed to research as a core operating model.
The best independent sites build real patient relationships. They continuously grow a database of people in their community who are already inclined to participate in research. They know the local referral patterns, understand the culture of their community, know which patients need extra education, recognize which populations are skeptical, understand which outreach channels work, and can anticipate which barriers are likely to prevent participation.
The evidence on patient participation supports the value of this type of local knowledge. Barriers to participation often include lack of understanding, logistical burden, concerns about risk, mistrust, insufficient provider communication, and structural obstacles such as travel or access. [4] A site that knows its community, has earned patient trust, and communicates clearly is better positioned to address those barriers than a site relying only on database output.
These sites can and should use technology, including CTMS, EHR tools, patient engagement platforms, and AI-enabled support where appropriate. But their advantage does not begin with the technology. It begins with community trust, patient knowledge, operational discipline, and research commitment. The tools simply allow them to move faster, communicate better, track more precisely, and scale what already works.
They can also be agile in ways larger organizations sometimes cannot. They can move quickly, adapt, communicate directly, and maintain a human connection with patients that no centralized database can replicate. The future is not large versus small. The future is verified execution capability versus assumed access.
Technology Is an Amplifier, Not a Substitute
Technology is essential. Used well, CTMS platforms, EHR tools, recruitment technology, analytics, and AI can help sites identify potential patients, search records, segment databases, automate reminders, track outreach, monitor funnel performance, and reduce coordinator burden. These tools are increasingly important, and sites that refuse to modernize may find themselves at a disadvantage.
The evidence supports the promise of better data and better tools. Site selection models using real-world patient data and site-level recruitment history can improve site ranking. [3] CTTI recommendations also support using real-world data to plan eligibility criteria and enhance recruitment, while emphasizing that data must be fit for purpose and used with attention to patient and site needs. [7]
But technology does not create trust. It does not make a patient willing, explain the study with empathy, overcome fear, make a burdensome protocol attractive, or replace the relationship between a patient and a site team. It also does not make an unrealistic feasibility projection more realistic.
Technology helps strong sites become stronger. It gives disciplined sites better tools and allows high-performing teams to move faster and track more accurately. But technology placed on top of a weak recruitment model does not magically produce enrollment. A database can identify possible patients. AI may help prioritize those patients. A CTMS may help organize the workflow. An EHR may help find records faster. But the site must still convert interest into consent, consent into screening, screening into enrollment, and enrollment into retention.
The technology is not the engine. It is the accelerator. The engine is the site’s culture, relationships, staff capability, investigator engagement, patient trust, and operational discipline.
Coordinator Strength Matters More Than PI Prestige
A respected investigator helps. A high-volume practice helps. A well-known institution helps. But in daily study execution, the coordinator and site operations team often determine whether patients move from identification to consent, from consent to screening, from screening to randomization, and from randomization to completion.
Strong coordinators manage the funnel. They pre-screen, educate, schedule, follow up, remind, document, anticipate barriers, and communicate with sponsors and CROs. They know when a patient is becoming uncertain, when the visit schedule is becoming a burden, when the protocol is creating friction, and when recruitment tactics are not working.
This is consistent with evidence showing that staffing, prioritization, dedicated support, competing trials, and site practices affect recruitment and retention. [5] It also aligns with what sponsors and CROs see operationally: the best feasibility answer on paper means very little if the site does not have the coordinator capacity and internal discipline to execute.
Technology can help coordinators manage that work more efficiently, but it cannot replace their judgment, persistence, empathy, and operational control. A site with a famous PI but an overloaded coordinator may underperform. A dedicated site with a disciplined coordinator, engaged investigator, strong patient relationships, and appropriate technology may outperform expectations. That is why site selection must look beyond logos, database size, investigator reputation, and software platforms.
High-Enrolling Sites Manage the Funnel
Weak enrollment discussions often focus only on the final number. Strong enrollment management focuses on every step before that number. How many patients were identified? How many were contacted? How many were reached? How many were interested? How many were pre-screened? How many consented? How many failed screening? Why did they fail? How many randomized? How many are at risk of dropping out?
A high-performing site can explain where the funnel is working and where it is breaking. An underperforming site often cannot. ‘We are working on it’ is not recruitment intelligence. ‘Patients are not interested’ is not enough. ‘The criteria are too tight’ may be true, but it is incomplete unless the site can explain which criteria are causing the screen failures and how many patients were affected.
This is one area where technology can provide real value. A good CTMS, recruitment platform, EHR workflow, or AI-supported matching process can help make the funnel more visible. But visibility is not the same as execution. Seeing the problem faster only matters if the site has the discipline and capacity to act on what it sees.
Sponsors and CROs need visibility into the funnel early enough to act. Without that, recruitment becomes guesswork.
Retention Is Part of Enrollment Performance
A site that enrolls quickly but loses patients is not truly high-performing. Sponsors do not just need consented patients. They need patients who remain in the study, complete visits, follow the protocol, and produce usable data. Retention begins before consent, when the site explains the study clearly, sets realistic expectations, identifies barriers, and helps the patient understand what participation will actually require.
A review on retention strategies emphasized that high retention is important to the validity and credibility of randomized controlled clinical trials. [8] That is an important reminder that enrollment volume without retention and data quality is not true site performance.
High-quality sites think about retention from the beginning. They provide flexible scheduling when possible, communicate clearly, follow up consistently, make patients feel respected, respond quickly to questions, and recognize when a participant is becoming frustrated or overwhelmed. They understand that the patient experience is not separate from trial execution. It is trial execution.
Here again, technology can support the work. Automated reminders, visit tracking, patient communication tools, and AI-supported workflow prompts may reduce missed visits and improve responsiveness. But patients stay engaged primarily because they trust the site, understand the commitment, feel respected, and believe their participation matters.
Better Questions Lead to Better Site Selection
Sponsors and CROs should still ask how many patients a site has, but they should not stop there. They should ask how many patients the site has identified against the most important inclusion and exclusion criteria, how many are already in the site’s research database and have agreed to be contacted, what actual enrollment versus target looked like in similar studies, what the screen failure rate was, and why patients failed.
They should also ask how many patients can be contacted in the first two weeks after activation, who owns recruitment at the site, how often the investigator will review potential patients, what competing studies are currently open, which recruitment channels work best in that community, what the main reasons are that patients would decline this specific study, what protocol elements concern the site, and what the realistic best-case, likely-case, and worst-case enrollment estimates are.
It is also fair to ask what technology the site uses. Does the site have a CTMS? Can it search its EHR effectively? Does it maintain a research-inclined patient database? Does it use AI or analytics to support identification and outreach? Those are useful questions. But they should not be mistaken for the whole answer, nor should it be assumed a “No” answer to any of them is disqualifying. The more important question is whether the site has the people, processes, discipline, and patient relationships to turn those tools into enrolled and retained participants.
The quality of the answer matters. Mature sites give specifics. Weak sites give generalities.
What This Means for Feasibility
The industry does not need another static list of sites. It needs better real-time intelligence. It needs to know which sites have actual patient relationships, research-inclined databases, current coordinator capacity, competing studies, activation speed, honest communication, retention discipline, clean data, realistic feasibility judgment, and the practical ability to use technology in support of execution.
That kind of intelligence cannot be fully captured by a database alone. It requires relationships, experience, judgment, and current site-level knowledge. Technology can support that intelligence, organize it, and help analyze it. But the insight still depends on understanding the real difference between patient access and patient convertibility.
The Bottom Line
The difference between consistently high-enrolling sites and sites that consistently miss enrollment expectations is not simply size, brand, database volume, investigator reputation, or access to newer technology. It is the difference between a living recruitment system and theoretical patient access.
High-enrolling sites convert real community relationships, disciplined feasibility, operational readiness, investigator engagement, coordinator strength, appropriate technology, and patient trust into enrolled and retained participants. Underperforming sites often rely on diagnosis counts, passive recruitment, optimistic projections, and hope.
There is a place for large multi-site companies. There is also a place for dedicated, independent research sites. But in both models, the winners will be the sites and networks that can convert real relationships, real community trust, and real recruitment intelligence into enrolled and retained patients.
CTMS platforms, EHR tools, analytics, and AI will continue to play an increasing role in clinical research. The best sites will use them well. But those tools do not create high-performing sites. They make already capable sites more efficient, more visible, and more scalable.
That is where feasibility becomes more than a spreadsheet. That is where site selection becomes strategy. And that is where clinical research has the best chance to move faster, perform better, and deliver more for the patients waiting on new medicines and devices.
Endnotes
[1] Mette Brøgger-Mikkelsen, Zarqa Ali, John R. Zibert, Anders Daniel Andersen, and Simon Francis Thomsen, ‘Online Patient Recruitment in Clinical Trials: Systematic Review and Meta-Analysis,’ JMIR, 2020. The paper notes that around 80% of trials fail to meet initial enrollment targets and timelines. https://pmc.ncbi.nlm.nih.gov/articles/PMC7673977/
[2] Kenneth A. Getz, ‘Enrollment Performance: Weighing the Facts,’ Applied Clinical Trials, reporting Tufts CSDD findings that 41% of activated investigative sites were unable to achieve target enrollment. https://www.appliedclinicaltrialsonline.com/view/enrollment-performance-weighing-facts
[3] L. Hulstaert et al., ‘Enhancing Site Selection Strategies in Clinical Trial Recruitment Using Real-World Data and Machine Learning,’ PLOS ONE, 2024. The study incorporated site-level recruitment data and real-world patient data and found improved site ranking compared with common industry baselines. https://pmc.ncbi.nlm.nih.gov/articles/PMC10927105/
[4] Edgardo Rodríguez-Torres, Margarita M. González-Pérez, and Carmen Díaz-Pérez, ‘Barriers and Facilitators to the Participation of Subjects in Clinical Trials: An Overview of Reviews,’ Contemporary Clinical Trials Communications, 2021. The review summarizes patient, physician, and system-level factors influencing participation. https://pubmed.ncbi.nlm.nih.gov/34401599/
[5] David B. Fogel, ‘Factors Associated with Clinical Trials That Fail and Opportunities for Improving the Likelihood of Success: A Review,’ Contemporary Clinical Trials Communications, 2018. The review discusses slow recruitment associated with inadequate staffing, lack of trial prioritization, competing trials, and protocol/site-practice fit. https://pmc.ncbi.nlm.nih.gov/articles/PMC6092479/
[6] Tufts Center for the Study of Drug Development, Impact Reports and related research on protocol complexity and rising participation burden, including more procedures, endpoints, eligibility criteria, and operational workload. https://csdd.tufts.edu/publications/impact-reports
[7] Clinical Trials Transformation Initiative, ‘Recommendations: Use of Real-World Data to Plan Eligibility Criteria and Enhance Recruitment.’ CTTI emphasizes fit-for-purpose data, cross-functional planning, and attention to patient and site needs. https://ctti-clinicaltrials.org/wp-content/uploads/2021/06/CTTI_RWD_Recs.pdf
[8] S. Poongothai et al., ‘Strategies for Participant Retention in Long Term Clinical Trials,’ Perspectives in Clinical Research, 2022. The review emphasizes that high retention is important to the validity and credibility of randomized controlled clinical trials. https://pmc.ncbi.nlm.nih.gov/articles/PMC10003583/
