The timeline looks orderly, but the time is not spent evenly or predictably. Two stages dominate, and they dominate for different reasons.
Design is the first place time is lost. A protocol is the rulebook for a trial: who may join, what is measured, how often, and how the results will be judged. When a protocol is too complicated or poorly matched to the patients who actually exist, it has to be amended — changed after the trial has already started. Each amendment means re-approval, retraining of sites, and re-consenting of participants, and a single amendment can cost weeks to months. A trial with several amendments can lose a year before it has enrolled anyone.
Recruitment is usually the single largest delay. Trials routinely fail to enroll on schedule, and a large share of sites enroll far fewer participants than planned. Every month of delay in recruitment pushes every later stage back by the same amount, because the trial cannot be analyzed until enough participants have completed it.
Running the trial adds its own friction. Data arrives from many sites in inconsistent formats and must be checked for errors; safety information must be reviewed; participants miss visits or drop out, which reduces the usable data and can force the trial to run longer to compensate.
Finally, analysis and reporting add a block of months at the end. The statistical work is careful and reviewed, and the resulting reports and submission documents are long and must be internally consistent. None of this is wasted effort — it is what makes the result trustworthy — but it is slow, and it sits on the critical path.