CGCampbell lays out the tactical considerations so I will add in some strategic context.
The goals of warfare in Sun Tzu's day differed massively from those in our era. Wars in those days had more in common with street gangs squabbling over turf than what we consider valid war goals. All political organizations rested on family structure and individuals who combined military and political authority did so largely based on their position in the family structure.
The goal of warfare in the vast majority of cases was kill, capture or otherwise neutralize the (usually) patriarch of the opposing side and preferably his heirs and as any other members of his extended family you could.
The armies of the day, functioned more like vast vans of bodyguards all structured ultimately to protect the lives, and if not lives, then a surviving family structure of each side. This held true, even for leaders who led from the front.
History is replete with examples of armies going from near victory to total defeat when one guy, usually the patriarch, got taken out. This caused armies to implode because not only did they lose their immediate military direction, but also the entire political and social order the army sprang out of.
Killing the leadership was likewise important because the majority of soldiers were poorly trained, equipped, and easily replaceable, so wiping out "an entire army" but not the patriarch, his family and their clients, didn't actually accomplish much militarily (and hard to do in any case when an army could just scatter.) The enemy would just round up another mob of peasants and mercenaries working on promises the next year or so and strategically, nothing changed.
Another big problem faced by the armies of the past, that we have lost an intuitive feel for, is that armies were utterly temporary assemblages that began to disintegrate almost as soon as they formed.
Most civilizations could only fight for a few months out of the year when manpower losses would not destroy agricultural output. With rare exceptions, every army in human history up until World War 2 lost most soldiers to disease than enemy action (hurrah to DDT, estimated 20 million lives saved in WW2 alone!)
Strategy in the subject era boiled down to trying to force the enemy's army to offer battle as quickly as possible, while you had some but not too much advantage, so you could get a shot at the boss. That wasn't easy because the armies were not that large relative to the terrain they operated in and one army could avoid another if it so chose.
It seems almost comical today, but heralds in the past did ride between opposing armies in order to try to negotiate a field of battle; like two groups of kids trying to decide what lot to play baseball on.
This is why sieges played such a central role in the warfare of the times. An aggressor could lay siege to a city or other specific locale that the target had to have and that could, in theory, force him to give battle to relieve the city.
Unable to lay enough sieges, Henry the Fifth and his heirs, adopted a massive series of raids or Chevauchée that destroyed so much of the French kings agricultural lands, they had no choice but to come out and give battle. Basically, "come and fight where we can get a chance to kill you or we burn down the entire neighborhood!"
In Chapter 9 (The Army on the March), Verse 5, Sun Tzu is advising the aggressor, not a defender, in this case and dealing with the tricky problem of inducing an enemy army to cross a river to give battle, a crossing which the enemy knows will place them at risk while transiting.
Basically, his advice boils down to, "don't be tempted to defeat the enemy's bodyguard in detail as they cross because they'll just stop crossing and go away. Instead, step well back, then let them get all across, and then you'll get your chance to kill the big boss."
There were exceptions to the use of this tactic. In some cases, the destruction of the army was the point. The Battle of Bannockburn comes to mind. In it, the Scots wanted to destroy a relieving army so they could gain the castle Sterling. The odds of defeating the entire English army were slim and killing the King far more so. They just needed to kill enough of the relief force to make relief impossible and they would win. Fortunately, defeating a part of the English army in detail caused a rout, turning a wise, conservative, tactical action into a stunning strategic victory.
Very telling, if I recall correctly, Sun Tzu did not advocate these tactics with nomads but instead used bodies of water to hold them off until they got tired and rode back to the steppes. Nomads were decentralized, even under the great Khans, and killing one guy seldom caused wholesale collapse.
Sad to say, but a lot of what passed for large-scale chivalry back in the day, in most civilized cultures, often boiled down to politely cajoling one's enemy to stick his head in the noose before the clock ran out. More of a "hey, come out here where I can kill you before everyone sobers up and wanders away," than something out of Camelot.
附水迎客 敵必不得渡而與我戰
- "If you wait besides a river for the enemy, then the enemy will not cross to fight you."