Most ecommerce accounts have a structural problem they don't see until it's too late. Automated campaigns do exactly what you ask: they find conversions at the lowest possible cost. But in doing so, they concentrate all spend at the bottom of the funnel, where people already know you and are ready to buy.
The result? You keep selling to the same customers, over and over. ROAS looks great on paper. New customer acquisition stalls. And when Q4 arrives, there's nobody left to convert because nobody funded the top of the funnel 6-8 weeks earlier. It's like harvesting a field and never planting the one next to it.
A recent piece on Search Engine Journal frames this the same way we see it across the accounts we manage: prospecting funded in October pays out during Q4. Prospecting funded during Q4 mostly pays out in January, at holiday prices. That's the worst version of the trade. And yet, that's exactly what most accounts do: delay prospecting until it's too expensive to matter.
Blended ROAS hides the real problem
The most dangerous thing about blended efficiency is that it reports roughly the same number whether you grew your customer base or harvested it. We've written about what ROAS doesn't tell you about real performance before, but in a Q4 context, the stakes are higher.
Say you're seeing a blended ROAS of 5x in October. Looks excellent. But if 80% of those conversions come from returning customers, your actual prospecting ROAS is 2x, hidden under a pretty average. The problem doesn't show up in reports. It shows up in January, when your new customer base is the same size it was in September and there's nowhere left to grow.
When you cut prospecting in November because the ROAS looks soft next to your harvest campaigns, you haven't saved money. You've moved your December problem forward and called it efficiency. And in January, acquisition costs peak while organic demand drops. It's the worst possible time to buy someone's attention.
What to do: set your prospecting budget as a fixed number, not a percentage of whatever's left over. Percentages are the first thing to disappear under Q4 pressure. A fixed number isn't negotiable, not for you, not for the algorithm.
Google Ads: Customer Match and the new customer acquisition goal
Google introduced a new customer acquisition goal in Performance Max that operates in two modes. The first, "bid higher for new customers," shifts bidding weight toward new customers without sacrificing campaign volume. It's the recommended starting point for most accounts because it doesn't suddenly cut off the flow of existing conversions.
The second mode, "new customers only," is for deliberately carved-out prospecting campaigns with separate funding and different performance metrics. Don't move everything to this mode right away. Shift the weighting, observe for two weeks, adjust again. If you jump straight to "new customers only" across all campaigns, you'll see a sharp volume drop that will scare you back to the old settings.
A critical detail many teams overlook: this goal works far better with an up-to-date Customer Match list than with conversion history alone. The difference is significant. Google can exclude existing customers more precisely when it knows them from your list, not just from the conversion pixel. Export your customer file, upload it to Google Ads, and label it in the acquisition panel of your Conversions Summary. Do this in October, not November, because the list needs a few days to process and match.
Meta: how to rebuild the top of the funnel
Meta retired the Existing Customer Budget Cap from Advantage+ Shopping (now renamed Advantage+ Sales). If you relied on that feature to control how much budget went to existing versus new customers, you need to rebuild the mechanism manually. Not ideal, but the alternative is letting the algorithm decide on its own, and in Q4 the algorithm will always choose the path with the lowest cost per conversion: existing customers.
The simple version: one manual sales campaign that excludes existing customer custom audiences. That's the equivalent of a 0% cap on existing customers. It works for smaller accounts where fine segmentation doesn't justify the added complexity.
For larger accounts, the recommended structure uses two ad sets: one targeting existing customers and one targeting broadly with customers excluded. Use ad set spending limits to maintain your desired split and to run different creative for each group. Existing customers respond to win-back messages and exclusive offers. Prospecting audiences respond to social proof, benefits, and differentiation. Running the same ad for both audiences is a compromise that costs more than it looks.
This structure gives you control over what Advantage+ used to handle automatically but now does without transparency. It's more work, but in Q4, control matters more than convenience.
The exclusion list you need ready now
Most accounts have exclusion lists that haven't been updated in months. In a normal period, the impact is small. In Q4, with increased budgets and maximum competition, every dollar spent on a customer who would have bought anyway is a dollar missing from prospecting.
What to exclude from automated campaigns, by priority:
- Purchasers from the last 30, 60, or 90 days (based on your product's repurchase cycle)
- Your full customer file (current, not the one from six months ago that excludes people who've since changed their email)
- Cart abandoners, if you're running a separate remarketing flow (to avoid duplicate bidding on the same user)
- Brand terms (run in a separate campaign with its own budget)
On brand terms: if a competitor bids on your brand and you don't, you can lose that sale to them. It happens, it's measurable, and brands tend to learn it expensively. Buy your brand terms defensively. But keep them in their own campaign so your prospecting numbers actually reflect prospecting, not inflated by conversions you would have gotten anyway from organic brand traffic.
What to do in October, specifically
Don't restructure your account in the last two weeks before Black Friday. The learning period is too short and you risk entering peak season with unstable campaigns that spend without direction. Everything below needs to happen now, in October:
- Export and upload your customer file as a Customer Match list in Google Ads. Give it a few days to process.
- Enable the new customer acquisition goal in bid-higher mode on Performance Max campaigns. Don't jump straight to "new customers only."
- Rebuild your Meta structure with two ad sets and spending limits that reflect how much you want to invest in acquisition versus retention.
- Audit your exclusions across both platforms. Look for stale lists, expired audiences, and at least one campaign nobody remembers building.
- Set your prospecting budget as a fixed amount, not a percentage. Write it down somewhere visible so it's not the first line cut when pressure builds.
The attribution discussion about which campaign produced which conversion? Save it for January, when you have complete data and the pressure has eased. Right now, structure is what matters. And the right structure is one that doesn't let the top of the funnel go dark when Q4 pressure builds.
