Clinical trials don't fail on science. They fail on staffing.

Ask why a clinical trial slipped and the first answers are usually scientific or regulatory: an amendment, a supply issue, a slow ethics review. Ask the people running the sites and the answer is more often about people. Who was in the coordinator seat. Whether the monitor knew the protocol. How long the vacancy sat open. Staffing is the variable that quietly sets the timeline, and it is the one least often designed for.

Timelines are made of handoffs

A study runs on a chain of handoffs: feasibility to start-up, start-up to activation, activation to enrolment, enrolment to data cleaning, cleaning to lock. Each handoff assumes a competent person is present and current. Remove one and the chain does not break loudly; it stretches. Queries age. Visit windows slip. Monitoring reports arrive later than the issues they describe.

• A vacant coordinator seat delays consent conversations and pushes first-patient-in.
• A new monitor without protocol depth raises findings late, after the deviations have accumulated.
• A thin start-up team leaves essential documents incomplete and activation waits.
• Turnover mid-study forces re-training that is charged to the schedule, not to the budget line.

Nobody records staffing as the reason a trial slipped, because it never arrives as a single event.

Why the seat stays empty

The usual explanation is talent scarcity. That is only partly right. Much of the shortage is a matching problem: capable people exist, and the screen cannot see them. A site coordinator with four years of source documentation and deviation judgement is filtered out of a monitoring requisition for not holding the monitoring title. An internationally trained pharmacist is filtered on an unfamiliar institution rather than on investigational product handling.

Meanwhile the roles themselves keep moving. Decentralised elements, risk-based monitoring and new data platforms change what competence looks like faster than job descriptions are rewritten, so the requisition describes a role that no longer exists in quite that form.

Treat staffing as a study risk

Risk-based approaches are standard for data and for safety. The same discipline applies to people, and it costs less to apply than the delay it prevents.

• Map the roles each milestone depends on, and name the single points of failure.
• Assess capability before activation, not after the first monitoring visit finds a pattern.
• Score readiness by domain, so a gap in one area is developed rather than disqualifying.
• Build a bench: identify and assess adjacent-skill candidates before a seat opens.
• Track vacancy days against milestone slippage so the cost becomes visible to the study team.
• Re-check capability after protocol amendments that change the work materially.

The list is unremarkable, which is the point. None of it requires new technology. It requires deciding that the staffing plan deserves the same rigour as the monitoring plan.

What good looks like at the site

A well-staffed site is recognisable before enrolment starts. Roles are defined against the protocol rather than against a generic job family. Someone has confirmed, in advance, that the coordinator can run the consent conversation this protocol demands and handle its deviation reporting rules. Training is logged as demonstrated capability, not just as completed modules. And the site knows who covers each critical function when the primary holder is on leave.

The practical conclusion

Protocols are reviewed by many people before they run. Staffing plans usually are not. Bringing the staffing plan the same scrutiny, defined roles, assessed capability, a named bench and vacancy days on the risk log, is one of the cheapest schedule protections available to a study team.