Most investors don't even know about it because their was no record of it, until 2016 when it congestions start creating awareness to the minds of not just it developers but also we the supporters.
The shocking part is that due to how fast it was, their was no record keeping of it's stats from 2010 to 2016, but look at it today, it's now gaining more attention for the wrong reason due to congestions and more adoptions.
My question is how can the mempool be improved to speed up transaction even with it congestions, aside from high fee's?
You don't understand mempool. There is no single node nor a global node that represents all the transactions. Do you know what happen when a transaction is broadcast?
It gets rebroadcasted across all the nodes of the bitcoin network and before each node accept a transaction into their pool, they validate it and confirm if it was a authentic transaction(meet consensus rule of bitcoin transaction) and then rebroadcast the copy to another nodes until the transaction is broadcasted to all available nodes on the bitcoin network. This is where they stay until a miner take them, mine them before included in the next block and they can contain as much as many transactions depending on the size of the mempool.
The question is how should a bitcoin block size contain more transaction so that the we don't have to experience more transactions waiting on the mempool, that has been an unending argument by developers, everyone has different opinions regarding this but just mote that the each bitcoin block size can accommodate a maximum of 4 million weight units.