My subjective impression from the correlation studies that I have seen is that the remainder largely is a long list of purely negative factors - i.e. in the "nurture" stage there are many ways to screw up the development of children with a large impact, but limited ways to meaningfully improve it over the genetic 'baseline'.
I.e. the correlation of nutrition in IQ is mostly driven by malnutrition, not by differences between great nutrition and mediocre nutrition; the impact of parenting is heavily driven by the outlier cases of severe abuse or deprivation, not the differences between great parenting and mediocre parenting, the impact of peer groups is largely set by the minority of cases that result in severe addiction or gang membership, not the difference between lousy friends and great friends, etc - that's the pattern that I seem to see there.
Do you have a source for that claim? Might you be misquoting a different popular statistic that g accounts for roughly have of intelligence between individuals? Or perhaps the figure that intelligence is about 50% heritable?[0] It is important to note that heritability of a trait and the extent to which a trait is genetically caused are not the same. It is possible for a trait to be eg. 80% caused by genetics and only 60% heritable, even on the average. This can be the case where the trait is defined by multiple genes (and non-coding DNA) where the proportion of the different genes matters and where specific combinations of genes are necessary. Also heritability of intelligence has been reported as higher than 75%.
The article may be conflating genetics with other types of biological hereditability, (such as epigenetics?) however I think that is generally pretty common and fine in every lay conversations.
That leaves a ton of room for other factors, such as nutrition, health, and education, to create significant disparities.