Most marketers obsess over CPA.
That makes sense. It’s the only line item on your P&L you can sometimes move double digits overnight.
But CPA has a ceiling.
If your AOV is stuck at $80, your margin is capped no matter how good your ads are.
That’s the part many operators miss.
AOV is a capital efficiency lever, not a merchandising tactic.
When you raise AOV:
That is real leverage.
When I launched my current menswear brand, I started with:
Plenty of choice, yet AOV was ~$95.
A month later, it sits at ~$137.
What changed wasn’t ads. It was how the system encouraged larger carts.
I introduced tiered “buy more, save more” pricing, but only where the math worked.
At ~$30 CPA, discounts don’t work on single-item orders.
They do work on multi-item orders where:
That spread is where margin is created.
The offer lives everywhere:
Same economics. More surface area.
Another lever that materially moved AOV was bundling.
Not “frequently bought together” widgets. Not last-minute upsells.
Actual, pre-built bundles that made sense to buy as a unit.
I took existing products customers were already pairing manually and packaged them together:
No new SKUs. No new inventory risk. No new operational complexity.
Just clearer decision-making for the customer.
Why this works:
The customer isn’t asking, “Should I add another item?”
They’re asking, “Which setup do I want?”
That distinction matters.
From a unit-economics perspective, bundles are powerful because:
You’re not forcing higher spend. You’re designing for it.
Combined with tiered incentives and broader offer distribution, bundling becomes a structural AOV driver, not a campaign.
That’s the difference between optimizing a store and designing a system.
Instead of launching a new category, I expanded colorways on best sellers.
Why?
I even sold new colorways before inventory landed as preorders so demand tells me what deserves capital.
This isn’t about tactics.
It’s about understanding that AOV is a capital multiplier.
If you don’t design for it, growth will always feel harder than it should be.