Metrics & KPIs

AOV vs ROAS: The Lever Behind the Ratio

In short: AOV measures how much a customer spends per order; ROAS measures how much revenue came back per dollar of ad spend, and AOV is one of the levers that moves ROAS without changing traffic or conversion rate at all. Raising average order value through upsells, bundles, or free-shipping thresholds can lift ROAS even if the same number of people convert at the same rate. AOV is an average and can be skewed by a handful of large orders, so it needs to be read alongside order count or median, not alone. Track AOV to find the lever, track ROAS to see whether pulling it worked.

By the AdFlint research team · Fact-checked against current Google and Meta platform behavior · Last reviewed July 2026

AOV

Total revenue divided by the number of orders in a period, the average amount a customer spends per transaction.

It is the lever connecting conversion rate to ROAS: raising it improves returns without winning a single additional customer. Being an average, a handful of large orders can pull it well above what a typical buyer spends. The common misreading is tracking it without the median or product mix behind it, then crediting a promotion-driven swing to a creative change.

Full definition

ROAS

Revenue attributed to advertising divided by the ad spend that produced it, expressed as a ratio or multiple of spend.

It compares topline revenue against media cost and ignores everything else: margin, shipping, returns, fulfillment, and the fixed cost of operating. It also inherits every weakness of the attribution feeding its numerator. The most common misreading is taking a strong platform figure as proof of profit, when the same customers may well have purchased without ever seeing the ad.

Full definition

Side by side.

The differences that actually change what happens in your account.

	AOV	ROAS
What it measures	Revenue per order - total revenue divided by number of orders.	Revenue per ad dollar - total attributed revenue divided by ad spend.
What moves it	Product mix, upsells, bundles, minimum-spend free shipping thresholds, pricing changes.	Conversion rate, AOV, cost per click, and auction competition - AOV is one of several inputs.
Does spend affect it directly	No - ad spend doesn't appear in the AOV formula at all.	Yes - it's the denominator of the ROAS formula.
Sensitivity to a handful of large orders	High - a few big-ticket orders can pull the average well above what a typical customer spends.	Lower directly, though a large order can still show up as a ROAS spike for whichever campaign gets credited.
Where it's reported	Ecommerce platform or analytics tool such as Shopify or GA4, not the ad platform.	Native column inside every major ad platform.
What a rising number alone tells you	Customers are spending more per order, for reasons that could include ads, merchandising, or pricing.	Return improved, but not why - could be more conversions, higher AOV, or cheaper clicks.
Failure mode	Read without order count or median, so a promotion-driven spike gets wrongly credited to a creative change.	Read without decomposing which input moved it, so the wrong lever gets pulled next.

What actually separates them.

AOV has no ad spend in its formula at all, so it can move for reasons completely unrelated to advertising - pricing, merchandising, shipping thresholds - while still changing ROAS as a side effect.

Raising AOV improves ROAS without acquiring a single additional customer, which makes it one of the few levers that improves ad efficiency on the demand side rather than the media-buying side.

AOV is an average and can be pulled sharply by a few large orders, so a jump in AOV doesn't necessarily mean the typical customer is spending more - checking the median or order distribution is the only way to tell.

ROAS is reported natively inside the ad platform; AOV lives in the ecommerce or analytics stack, so connecting a ROAS change back to an AOV change requires pulling data from two different places.

A ROAS improvement can come from more conversions, a lower cost per click, or a higher AOV, and only decomposing the number against AOV tells you which of those actually happened.

Which one should you use?

Use AOV when

You are trying to explain why ROAS moved and want to check whether order value, not conversion rate or cost per click, was the driver.
You are testing upsells, bundles, or a free-shipping minimum and need to see whether the average transaction size actually increased.
You suspect a ROAS spike came from a handful of large orders rather than a broad improvement in campaign performance.
You are looking for a lever to improve ROAS that doesn't require acquiring more customers or lowering cost per click.

Use ROAS when

You want the single number that reflects overall ad efficiency, regardless of which underlying input moved it.
You are comparing campaigns or setting a bid strategy, where the platform needs one blended efficiency number, not a breakdown.
You've already checked AOV and want to confirm whether a merchandising or pricing change actually flowed through to ad returns.
You are reporting overall campaign performance and AOV is a supporting detail, not the headline number.

Common questions.

Can raising AOV improve my ROAS without touching my ad campaigns at all?

Yes. Since ROAS is revenue divided by ad spend and AOV is one of the drivers of that revenue, anything that raises what a customer spends per order - a bundle, an upsell, a free-shipping threshold - can lift ROAS even if the campaigns themselves, their targeting, and their spend never change.

Why did my ROAS spike for one day?

Check whether a small number of unusually large orders pulled up your AOV for that day - a single big-ticket purchase attributed to a campaign can move a daily ROAS number a lot more than it would move a monthly one. Looking at order count alongside the spike usually reveals whether it's a real trend or a one-off order skewing the average.

Is a high AOV always good for ROAS?

Generally yes for the ratio itself, but a higher AOV built on heavy discounting or free-shipping giveaways can raise revenue while quietly shrinking margin, so the ROAS improvement doesn't always translate into more profit. This is the same gap that shows up between ROAS and POAS.

Should I track median order value instead of average?

It's worth tracking both. AOV, the mean, is what most reporting defaults to and is required to explain ROAS mathematically, but the median tells you what a typical customer actually spends and won't get dragged around by a few outlier orders the way the average will.

Or stop choosing between them.

AdFlint picks the setting, writes the ads, and keeps optimizing inside the Google and Meta accounts you already own.

Try AdFlint free See how it works

Related comparisons

All comparisons