‍Inside Modern Commerce #1 - Why ROAS Is a Constraint, Not the Goal

For years, return on ad spend (ROAS) has been one of the defining metrics in retail $media. Brands use it to evaluate campaign performance. Retailers use it to demonstrate value. Agencies optimize around it every day.

Naturally, many assume that the objective of an advertising system should be simple: maximize ROAS. But it isn't. In fact, if your bidding system is designed to maximize ROAS alone, it will often make the wrong decisions.

The best commerce platforms don't optimize for the highest ROAS. They optimize for the greatest business value while respecting a target ROAS.

That distinction may sound subtle, but it fundamentally changes how modern bidding systems are designed.

In retail media, ROAS should function as an efficiency constraint rather than the final optimization objective. The real goal is to maximize total business value while staying within the advertiser’s budget and target ROAS.

What Does It Mean to Treat ROAS as a Constraint?

Imagine three campaigns.

If the objective were simply to maximize ROAS, the first campaign would always be considered the winner. But no retailer or advertiser would choose to invest just one dollar to generate ten dollars in sales if there were an opportunity to generate thousands of dollars in incremental revenue at a slightly lower return. The problem is that ROAS measures efficiency, not scale. Optimizing only for efficiency often means leaving valuable growth opportunities on the table.

Commerce Is a Constrained Optimization Problem

This is where intelligent bidding systems think differently from humans.

Rather than asking: How do we maximize ROAS?

they ask: How do we maximize total value while staying above the advertiser's required ROAS?

Instead of treating ROAS as the objective, modern bidding systems treat it as a constraint.

The objective becomes maximizing total sales or business value.

The constraints include:

• available campaign budget
• desired ROAS
• campaign pacing
• auction opportunities
• marketplace conditions

Every auction becomes part of a much larger optimization problem.

Why Spending Less Isn't Always Better

One of the biggest misconceptions in retail media is that a campaign with an unspent budget is simply being "efficient."  Often, it isn't.

Imagine a campaign with a daily budget of $10,000 and a target ROAS of 6x.

If the system spends only $2,000 because it's waiting for only the very best opportunities, the campaign may achieve excellent ROAS, but it also leaves most of its budget unused.

The advertiser misses potential sales. The retailer loses monetization opportunities. The marketplace becomes less competitive. 

An intelligent bidding system continuously balances these tradeoffs.

If spending is too slow, bids may gradually increase to capture additional opportunities while still respecting the target ROAS. If spending is too fast, bids may decrease, or the system may intentionally skip certain auctions, to avoid exhausting the budget early in the day.

The goal isn't simply to spend more or less, but to spend wisely over time.

Every Bid Reflects an Opportunity Cost

Choosing to win one auction also means choosing not to participate in another.

That makes budget allocation one of the most important problems in commerce infrastructure.

Modern systems don't ask:

"Is this click worth paying for?"

They ask:

"Is this the best opportunity available given the remaining budget, expected future demand, and the advertiser's objectives?"

This perspective turns bidding into a resource allocation problem rather than a pricing problem.

Why AI Matters

This is also why AI has become such an important part of commerce infrastructure.

An AI-powered bidding system isn't simply increasing or decreasing bids.

It continuously estimates the expected value of every opportunity by combining signals such as:

• likelihood of conversion
• expected order value
• campaign performance
• budget utilization
• marketplace dynamics

These predictions allow the system to decide not only how much to bid, but also whether participating in a given auction is worthwhile at all.

The quality of those decisions determines long-term marketplace performance.

A Better Way to Think About Commerce Media

As commerce becomes increasingly dynamic, with AI shopping assistants, personalized recommendations, multiple discovery surfaces, and real-time shopper behavior, the role of bidding systems is changing.

The challenge is no longer maximizing a single metric, but balancing multiple objectives simultaneously.

ROAS remains an important measure of success, but it shouldn't be mistaken for the objective itself.

The most advanced commerce platforms don't optimize for ROAS. They optimize for sustainable marketplace growth, using ROAS as one of the guardrails that keeps those decisions aligned with advertiser goals.

Interested in learning more? Talk to a Topsorter.

FAQ:

Is a higher ROAS always better?

Not necessarily. A higher ROAS indicates stronger efficiency, but it does not show the total scale of sales or business value generated.

Why might a campaign fail to spend its full budget?

A campaign may underspend when its target ROAS is too restrictive for the available inventory, competition, and predicted conversion opportunities.

What is the goal of retail media autobidding?

The goal is to maximize total expected business value while respecting campaign constraints such as budget, pacing, and target ROAS.