🎯 The core idea
You do not have to charge the hill. Do a small amount of damage at range and the defending AI flips into attacker behaviour, walks down, and hands you every advantage it was holding. In Zerkovich's own controlled test that turned 786 casualties into 245.
- free costs nothing, just do it
- cheap risks one unit at most
- setup needs positioning first
- grind costs real battle time
- skip the thing to avoid
Why the high ground is worth this much trouble

Four advantages, all stacking, all in the first minute of the video.
| Advantage | What it does |
|---|---|
| Flank protection | Your flankers have to swing much wider to get round |
| Slower movement | Units running uphill move slower, so the defender gets more shooting time |
| Fatigue | Everything climbing tires faster, infantry worst |
| Damage | Up to 30% more melee damage from above, and up to 30% more on projectiles fired down |
The two damage figures are the ones people quote, and they scale with how much higher you actually are — they are not flat bonuses.
He gives "up to 30%" for both without a source and hedges it again later ("probably giving them that 30% boost"). Treat it as his rule of thumb rather than a confirmed constant.
The attacker–defender rule that makes it all work
This is the mechanic the whole counter-play hangs on, and it is worth understanding before the steps.

Every battle has an attacker and a defender, and the deployment screen labels them. If you attacked on the campaign map, you are the attacker.
What matters is that the AI plays the role, not the position. Zerkovich shows a Kislev army holding a hill with artillery and a mass of missile units against Greenskins with no artillery at all — and on Start Battle they walk down off it. They are the attacker, so they attack, and they give away flank protection, damage and fatigue to do it.
The inverse: a Vampire army with no ranged units at all sits perfectly still against Ogres who will shoot first, because the Vampires are the defender. And that is correct play — the attacker loses when the clock runs out, the defender wins.
So the hard case, and the only one this guide is about, is an AI defender sitting on high ground with missiles and artillery.
Bait them off the hill

The whole method is one sentence: damage them a little at range and they turn into the attacker.
Those three — magic, artillery, skirmish cavalry — are what he calls the main options. In the Slaanesh-against-Dwarfs example a single skirmish cavalry unit was enough: Slayers chased it, damaging the Slayers convinced the AI it was losing, and the whole Dwarf line advanced.
If you have none of the three
Out-climb them instead
No AI manipulation in this one. It costs time rather than trickery, and it only works when the map gives you somewhere to go.

The tooltip in that frame is the proof of why step four exists — the unit has arrived Winded after the climb. Rest first, fight second.
After that it is an ordinary battle with the roles reversed: use whatever choke points the new position gives you, put the missile units where they shoot downhill, front line engages, cavalry flanks.
Entirely map-dependent. He shows a Beastmen-versus-Dwarfs hill where no equal ground exists anywhere, and the method simply does not apply there.
The same battle, twice
Same map, same two armies: an Empire force from Reikland against Clan Mors holding the hill with missiles and artillery. He fights it once head-on as a control, then again properly.
Run 1 — straight up the hill

| Deployed | Losses | Remaining | |
|---|---|---|---|
| Player (Reikland) | 1374 | 786 | 588 |
| AI (Clan Mors) | 2129 | 1549 | 580 |
He wins, and calls it "a pyrrhic victory by all means." By the time the line reached contact, four or five units were down to roughly half health from the climb alone.
Run 2 — with the high ground taken

| Deployed | Losses | Remaining | |
|---|---|---|---|
| Player (Reikland) | 1374 | 245 | 1129 |
| AI (Clan Mors) | 2129 | 1578 | 551 |
Player casualties fall by 541 men — a 69% reduction — and survivors nearly double, from 588 to 1129.
The detail worth noticing is on the enemy side: their losses barely moved, 1549 against 1578. The entire benefit landed on the player. Taking the hill did not kill more Skaven, it stopped the Skaven killing you.
He says "and still the 1,500 losses for the Skaven," which rounds both figures. Enemy casualties were actually slightly higher in the tactical run, not identical. Also worth keeping in proportion: this is one trial each, not a controlled experiment.
Roll the line from one flank
The fallback when the map gives you no height and you would rather not bait.
Come in from one side of the enemy army rather than across its whole front. Leave a few units out wide on the far side so their line cannot rotate to face your main body. Then start at one edge and sweep, so only a few of their units fight your entire army at a time.
He is honest that it will not work every time, and that plenty of maps do not allow it.

Sources
Built from Zerkovich, How to beat one of the WORST BATTLE situations! — Total War Tactics, 12 minutes, and the top comments underneath it. Screenshots are frames from that video, including both post-battle casualty panels.
No patch number is stated anywhere in the video, so none of the mechanical claims here are pinned to a specific build. The "damage them a little" threshold is never quantified — commenters report both single-hit triggers and armies absorbing half a roster of magic damage without moving.
Frequently asked
Why does the AI leave a strong position when I damage it?
It switches from defender behaviour to attacker behaviour. The attacker loses when the battle clock runs out, so once the AI thinks it is attacking it has to come to you — and it abandons the hill to do it.
How much damage do I need to do to trigger it?
The video never quantifies it. A single spell, or one skirmish cavalry unit harassing the edge of their line, was enough in every example shown. Commenters report mixed results — some say it takes two hits, others report armies absorbing far more before moving.
Does it work with summons or a melee hero?
The video says yes and lists both as fallbacks. At least one commenter insists only ranged damage triggers the switch and that summons never do. Test it in your own battle before relying on it.
What if I do not want to use an AI quirk?
Find ground at the enemy’s height or above, march there staying outside their artillery range, stand still until your units are Fresh again, then fight the same battle from the better position. It costs several minutes of battle time and only works when the map has somewhere to go.
How much difference does taking the high ground actually make?
In the video’s controlled test the same army lost 786 men attacking uphill and 245 after taking the height — survivors went from 588 to 1129. Enemy casualties barely moved, 1549 against 1578, so the whole benefit lands on your side.
Read next
Ten Beginner Mistakes In Total War: Warhammer 3
Five battle mistakes and five campaign ones, regrouped so the numbering makes sense. Bracing, friendly fire, shields, charge windows, deployment lanes — and on the map, build order, Winds of Magic, recruitment stalls, sacking and unused heroes.
Five Campaign Habits That Sink Your Warhammer 3 Run
Income that stops rising, an army that stops getting stronger, and three wars you never chose. Zerkovich’s five campaign mistakes turned into checklists, with the tooltip numbers he never says out loud and the fog-of-war tricks that need no scout hero.