A 5% chance is a reason to try more often
Wrong action produces information. Inaction produces nothing.
A 5% success rate sounds like a reason not to try. It is a reason to try many times.
If one attempt works with probability p, the chance that at least one of n attempts works is:
P = 1 - (1 - p)^n
At p = 5%, 14 attempts take you past even odds (51%). 45 attempts take you to 90%. The single-attempt number that scared you off says very little about a year of work.
Attempts teach
The formula assumes p stays fixed. It doesn't. Every failed attempt tells you something about why it failed, and the next one starts with better odds.
If each failure lifts p by a single point (5%, 6%, 7% and so on), you pass even odds on the 8th attempt, not the 14th.
This is the difference between wrong action and inaction. A wrong move returns information. Standing still returns nothing, so p never changes.
Scope sets the number of attempts
You don't choose n directly. You choose how big each attempt is, and your runway does the division. Two years of runway buys one two-year build, or twelve two-month ones.
Big scope is how smart people hide from iteration. The smarter you are, the better your justification for the two-year build.
The catch
Small attempts only compound if they search the same space. Twelve unrelated experiments are twelve first attempts, and nothing learned in one carries to the next.
Conviction picks the space. Iteration searches it.
In practice
Pick the direction on conviction. Cut the scope until you can afford many attempts inside it. Treat each failure as a paid lesson, and move.