It is the most honest execution benchmark because it includes everything: spread, impact, timing drift, and opportunity cost on the unfilled remainder. Beating vwap while the market ran away from you is not a good outcome and implementation shortfall says so.
Worked example: decision price 50.00, you buy 10,000 shares at an average of 50.06 and abandon 2,000 shares that would have cost 50.20. Shortfall is 10,000 x 0.06 plus 2,000 x 0.20, which is $600 + $400 = $1,000, or 20 basis points on the intended $500,000 position.
Algorithms optimising this quantity trade more urgently when volatility is high, because timing risk dominates, and more patiently when it is low. That behaviour falls straight out of the objective rather than being a rule someone added.
Related: execution-algorithm, market-impact, slippage, transaction-cost-modelling