What decides the split? Keynes started from the marginal propensity to consume (MPC), the share of an extra dollar of income that gets spent. Later theories added foresight. Under the permanent income and life-cycle hypotheses, households try to keep their spending steady over time, so they save most of a one-time windfall and spend most of a lasting raise. In practice many households hold little cash and cannot borrow easily, so their spending follows their current income closely. People also save more when the future looks risky.
Put numbers on it. A household takes home $5,000 a month and spends $4,600, so it saves $400, and its saving rate is 400 ÷ 5,000 = 8%. One month it receives a one-time payment of $1,200 and spends $300 of it. Its MPC out of the payment is 300 ÷ 1,200 = 0.25. For that month its income is $6,200, its spending $4,900 and its saving $1,300, a saving rate of 1,300 ÷ 6,200 = 21%. Close the shops so that its usual spending falls as well, and the result is April 2020.
Saving is easy to misread in three ways. It is not only money put in a bank: repaying the principal of a loan counts too, because it is income that was not spent. It is not investment in the economist's sense, which means building new capital; the Loanable Funds page shows how the two are connected. And what is prudent for one household is not always good for everyone at once. If all households cut spending together, firms sell less and incomes fall, an effect known as the paradox of thrift. The 2020 spike was not that paradox, because incomes were being supported while spending was blocked. How consumption feeds total spending is on the Aggregate Demand page.