Elevation is often measured in terms of "[unit of length] above sea level." Considering that the oceans aren't a uniform height, how is this measured?
It's actually an inordinately difficult problem. Neither the land nor the sea level is stable. A tide gauge along a coast needs 30 to 40 years of data to filter out the very noisy tidal and other weather signals. By the time one has homed in upon a precise average, one is already within the time-frame of land uplift or subsidence, and that is before we even think about the complexities of climate-change generated sea level rise, or quasi-decadal oceanic oscillations. Then there is the question of Earth's non-homogeneous gravity which affects local sea levels. For example if we correctly measure the elevation of a Himalayan peak relative to average sea level at the Gurarat coast, west India, we get an elevation of 'h1'. We can then measure the same peak relative to sea level in the Ganges delta, east India, which we will call 'h2'. We can make all the usual corrections for Earth's curvature etc., yet h1 will be different to h2 - though both are 'correct'.
There is no stable bench mark on 'God's Earth', so maybe the answer is to get off the planet completely. Satellite orbits are not perfect, but they are predictable, so satellite-based measurements of elevation are the best available.
This question is more about surveying than earth science.
Every country with an ocean coastline determines it's own standard for where sea level is. A mean point between the lowest tide mark and highest tide mark is generally taken to be the average sea level.
A surveying datum station is place at that point and all elevations above sea level, for those countries are referenced back to the national standard datum.