A normal distribution has kurtosis 3, so excess kurtosis of 0. Daily equity index returns commonly show excess kurtosis between 3 and 10; individual small caps and crypto pairs go much higher.
The practical meaning is that your estimate of volatility is dominated by a handful of days. Remove the ten worst days from a decade of S&P 500 returns and the standard deviation drops noticeably, which tells you the risk you are actually carrying lives in events you have barely sampled.
Because the fourth moment converges very slowly, a measured kurtosis from five years of data is itself unreliable. Use it as a qualitative flag that the tail is heavy, not as a parameter to plug into a formula.
Related: skewness, fat-tails, normal-distribution, tail-risk