ROAS Optimization by Campaign Type: What to Change in Prospecting, Retargeting, and Brand
ROAS optimization works when you tune each campaign type against the job it does, not against one blended target. Prospecting finds new buyers, retargeting closes people who are already in motion, and brand search collects demand that other campaigns created. Hold all three to the same ROAS number and you will starve the first, over-reward the second, and misread the third. This post assumes your tracking is already sound. If it is not, start with how to calculate your true return on ad spend, then come back here for the levers.
Why does one ROAS target fail across campaign types?
One target fails because the campaign types sit at different points in the buying journey, so the same dollar of spend does different work. Retargeting reaches people who already visited your store, which means high reported ROAS and a lot of revenue that would have arrived anyway. Prospecting reaches strangers, which means lower ROAS on paper and most of your future customer base. Brand search sits at the very end and mostly reflects demand you paid for elsewhere.
The reporting layer makes this worse. Platforms count conversions their own way, and since Meta's March 2026 change, click-through attribution only counts conversions after a link click, while engage-through and view-through are counted separately. Google also notes that its conversion counts can differ from other reporting tools because of timing, count settings, and attribution settings. Our walkthrough of Meta's 2026 click definition change covers what moved in your dashboards.
What to do: split your account view into three buckets before you touch a bid, each with its own success metric and expectation.
How do you optimize ROAS in prospecting campaigns?
Judge prospecting on new customer revenue and first-click contribution, then optimize creative and audience first, because those move cold-traffic results more than bids do. A prospecting campaign with a modest ROAS can be your best campaign if it consistently introduces customers who buy again.
What to measure
- New customer revenue and new customer acquisition cost, not total revenue.
- First-click credit by campaign, so you can see which campaigns start journeys.
- Repeat purchase rate of the cohort that arrived through each campaign, read after a few weeks.
What to change
- Creative: Rotate concepts, not just edits. Test distinct angles (problem, proof, offer) and let cold audiences tell you which one earns the click. Retire creative that gets cheap clicks but no new-customer orders.
- Audience: Keep prospecting audiences broad enough for the platform to learn, and exclude existing customers and recent site visitors so you are not paying to reach people you already own. Google's new customer acquisition goal can use your own customer lists, though Google says lists under 1,000 members may weaken it.
- Budget: Raise budget in steps, not jumps, and watch new customer cost at each step.
- Bid: Set targets you can defend against first-order margin plus a realistic repeat rate. A cost cap or ROAS floor set from blended results will usually choke cold traffic.
- Exclusions: Exclude purchasers (by list, not by hope) and keep prospecting out of your brand terms and retargeting pools.
How do you optimize ROAS in retargeting campaigns?
Judge retargeting on incremental lift, meaning the orders it caused rather than the orders it touched. Retargeting looks great in every platform dashboard because it targets people with high purchase intent, so reported ROAS is the least trustworthy number in the account. We cover why that credit inflates in why Meta's default attribution fails marketers.
Dot counts show the idea, not real data.
What to measure
- Click-based revenue against platform-reported revenue for the same campaign, side by side in AdBeacon's Optimize report. A wide gap points to view-through and engage-through credit.
- Holdout or geo-based lift where you can run it, even a simple one on a slice of the audience.
- Overlap with email and organic: if a Klaviyo flow reaches the same shoppers, the credit is being shared.
What to change
- Creative: Match the message to the stage. Cart abandoners need a reason to finish (shipping, reviews, a stock note), while product viewers need a reason to come back.
- Audience: Shorten windows and segment by depth. Someone who viewed a product yesterday is a different job than someone who added to cart a month ago. Google's conversion lag reporting shows how many days it takes to count conversions, which helps you set the window.
- Budget: Cap it as a share of total spend. Retargeting can only spend what prospecting fills the pool with, and pushing frequency past that point buys annoyance.
- Bid: Do not chase the highest ROAS. Loosen the target until incremental orders stop appearing, then hold.
- Exclusions: Exclude recent purchasers for a sensible post-purchase window, and exclude subscribers already in a live email flow when you can.
If you only change one thing, change the question from "what is the ROAS" to "what happens if I cut this by a quarter for two weeks."
How do you optimize ROAS in brand search campaigns?
Judge brand search on whether it captures demand other campaigns created, not on its ROAS, which will almost always look enormous. The useful question is whether paid brand clicks are cannibalizing organic listings, or protecting you from a competitor bidding on your name.
What to measure
- Brand search volume and branded revenue trend against prospecting spend. If prospecting rises and brand follows, brand is harvesting.
- Impression share and the share of brand clicks that would have gone to your organic listing.
- First-click and last-click credit side by side. Brand search tends to win last click and lose first click.
What to change
- Creative: Keep ads simple and aligned with the landing page. Use sitelinks for the pages shoppers want next.
- Budget: Size it to demand. If you are hitting the budget cap on brand terms, you are losing your own customers. If you are far below it, the budget is not the lever.
- Bid: Aim for the lowest cost that holds the top position when a competitor is present. Test pausing in a controlled window if organic coverage is strong.
- Exclusions: Add negative keywords for support, careers, and login queries, which waste spend without producing sales.
How do you set ROAS expectations by campaign type?
Set them in order of intent: brand search highest, retargeting next, prospecting lowest, then judge each against its own job instead of against the others. The numbers themselves depend on your margins, so anchor them to your breakeven and your repeat rate, not to a benchmark from someone else's account.
A practical scorecard to review weekly:
New customer revenue, new customer cost, first-click share, and four-week cohort repeat rate.
Click-based ROAS, lift from any holdout, frequency, and overlap with email.
Branded revenue trend, impression share, and the cost per brand click against organic coverage.
Rebalance in small moves and give each change a full week before you judge it.
One brand's 90-day example: Cacao Cardamom, a chocolatier working with Best Practice Media, reported Meta ROAS up 644%, net profit up 420%, and CAC down 52% in a 90-day comparison measured with AdBeacon. Read the Cacao Cardamom case study. Results varied by account; these are one brand's 90-day results, not a guarantee.
One brand's 90-day results with Best Practice Media, measured with AdBeacon. Results vary by account.
The fastest way to improve ROAS is to stop asking one number to describe three different jobs. Measure prospecting by the customers it adds, retargeting by the sales it truly causes, and brand by the demand it protects. When your attribution is click-based and tied to store orders, those three views get easier to defend in front of a client or a CFO. If you want to see how that looks on your own data, book a demo with AdBeacon.
FAQ
Should prospecting ROAS be lower than retargeting ROAS?
Usually yes, because prospecting reaches people with no prior intent while retargeting reaches people already close to buying. Compare them on new customer cost and incremental lift, not on one shared ROAS target.
How do I know if retargeting is actually incremental?
Run a holdout, even a small one, and compare orders between the exposed and held-out groups. Without a test, compare click-based revenue with platform-reported revenue and treat the gap as a warning sign.
Is a very high brand search ROAS a good sign?
Not by itself. It mostly shows that shoppers who already chose you clicked an ad on the way in, so check whether the revenue would have arrived through organic results.
Can I use one ROAS target for the whole account?
You can as a ceiling for overall profitability, but it should not steer individual campaign decisions. Set separate expectations per campaign type and tie them to your breakeven.