Skip to content

For sponsors and CROs · AI feasibility

Which sites will actually enrol — before you sign them.

Feasibility runs on a questionnaire the site fills in from memory, optimistically, with no way for anyone to check the answer. That is why a third of the sites in a typical study enrol one patient or none.

The cost of being wrong

Every day of delay is priced in lost exclusivity.

Enrolment is the critical path on almost every trial, and the sites you pick determine it. Picking on self-report is the single largest avoidable risk in study start-up.

80%

of clinical trials miss their enrolment timelines

Industry estimate

$600k–8M

lost to the sponsor for every day a trial runs late

Industry estimate, USD, per day of delay

1 in 3

sites in a typical study enrol one patient or none

Industry estimate

Widely cited industry estimates, included to frame the problem. Not Accrual’s own performance data.

Site selection

Questionnaire feasibility, and what replaces it.

DimensionFeasibility questionnaireWith Accrual
Where the number comes fromThe site's own recollection of its populationAn AI read of the records the site actually holds
When it is producedOnce, during start-up, then never revisitedContinuously, and re-run on every protocol amendment
Competing studiesRarely disclosed, never quantifiedModelled as load on the same candidate pool
FalsifiableNo — nobody can check the estimateYes — the forecast is tested against what enrols
Coverage beyond academic centresThin, because those sites have no research staff to answerEqual, because no site staff are needed to produce it

Enrolment forecasting

A forecast you can argue with, because it shows its basis.

The forecast is not a black-box score. It is a count of candidates the AI actually found in the population, a rate derived from how quickly that site works through candidates, and an adjustment for the other studies competing for the same patients. Every input is inspectable.

Candidate count

How many patients in this site's population the model finds meeting the protocol's criteria today, with the criteria that most reduce the count identified so you can see which ones are costing you the study.

Realistic rate

Not the theoretical maximum. How quickly this site has historically converted candidates into consented patients, and what its current research load allows.

Competing load

Which other open studies draw on the same candidate pool, and how much of it they are likely to absorb over your enrolment window.

The boundary

Sponsors never receive patient data. Counts, distributions and forecasts only — with small counts suppressed rather than reported.

This is not a policy preference, it is what makes sites willing to participate at all. The boundary is written into every site agreement and described in full in the patient data notice.

Diversity mandates

You are now required to enrol where you have no infrastructure.

MHRA and FDA expectations alike now mean a trial population has to reflect the population that will take the drug. That cannot be done from the same twenty academic centres that have absorbed trial volume for decades.

The sites that serve those populations are district general hospitals and community trusts with no research staff to answer a feasibility questionnaire — which is exactly why questionnaire feasibility cannot find them, and why a read of the population can.

How those sites get connected

How we charge

Two ways, both tied to something you can measure.

Per qualified enrolled patient

£1,200–4,000

Charged when a patient Accrual surfaced is enrolled and qualifies under the terms of the study. If nobody enrols, nothing is charged — which puts the risk on us rather than on your budget.

Rate depends on indication, protocol complexity and volume. Agreed in the order form.

Per feasibility study

Fixed fee

A population-grounded feasibility answer across a defined set of candidate sites, delivered before you commit to site selection. Priced per study by scope — number of sites, geographies and protocol complexity.

Quoted on scope. Talk to us with a protocol and a site list.

Ranges reflect the commercial model, not a published rate card. Final pricing is agreed per engagement.

For sponsors

Bring us the study that is behind.

Send a protocol and the site list you are considering. We will tell you where the population evidence disagrees with the questionnaires.