How to Improve ROAS: What Moves the Number | Polaris
The real drivers behind ROAS, why margin and creative matter more than any single tactic, and how to test more ad variants efficiently.
Improving ROAS (return on ad spend) means increasing the revenue you get back per dollar spent, and that comes from three places working together: getting more people to stop and buy from the same spend (creative), spending that dollar on an audience more likely to convert (targeting), and getting more revenue per buyer (average order value and margin). There's no universal ROAS benchmark that applies across products and margins — a "good" ROAS for a low-margin product is different from a high-margin one — so the useful question is which of these three is your account's weakest link right now.
Why does ROAS differ so much between products?
ROAS is a function of price, margin, and how competitive the auction is for your audience, on top of how good your creative and offer are. A product with strong margin can sustain a lower ROAS and stay profitable, while a thin-margin product needs a much higher one just to break even. That's why comparing your ROAS to a number you saw quoted for a different store or niche is rarely useful — comparing it to your own trend over time is.
What's the fastest lever to pull?
Creative is usually the fastest-moving lever available to you directly, because it's the thing you can change today without renegotiating margin or waiting on platform algorithm shifts. A new hook, a different proof point, or a fresh format can meaningfully change how much of your audience stops and engages, which flows straight through to ROAS. Stale creative that's been running for weeks tends to drag ROAS down as the same audience sees it repeatedly and tunes it out.
Does testing more variants actually help?
Generally yes — accounts that consistently test new angles and formats find winning creative faster than accounts running one or two ads long-term, and winning creative is what moves ROAS the most reliably. The limiting factor for most teams isn't lack of ideas, it's the time and cost of producing enough distinct variants to actually learn something from testing.
How do I test more creative without slowing everything down?
Polaris is built for exactly this gap. You connect it to Claude once as an MCP tool, authorize it, and then just ask Claude in plain chat to make an ad — attach a product photo or paste a script, pick settings from tappable option cards, and a live panel renders the finished ad right in the conversation, typically in 20 to 180 seconds. Because batches render in parallel and download as a ZIP, you can test several hooks, formats, or angles against each other in one pass instead of committing budget to a single untested creative. Failed renders auto-refund, so testing volume doesn't carry wasted cost.
How should I measure whether ROAS is actually improving?
Track ROAS against your own baseline over a meaningful window, not a single day, and watch hook rate on new variants as an early signal before revenue data settles. The ROAS calculator can help frame what ROAS you need at your current margin and spend level, and the ROAS glossary entry covers how the metric is calculated.