Performance Max New Customer Acquisition Goal: What Bid Higher vs New Only Really Changed for My CPA

Left to its own devices, a Performance Max campaign optimising for conversions will happily sell to people who already bought from you. That's not a bug existing customers are the cheapest, easiest conversions in the account, so any bidder chasing volume drifts straight toward them. The problem is that reacquiring a customer you already own isn't growth, and you're paying media dollars for a sale that would often have happened anyway. The performance max new customer acquisition goal exists to break that habit, and in my experience it's one of the few PMax settings that changes account economics rather than just shuffling metrics around.

But there's a catch that trips up almost everyone the first time: the goal isn't one switch, it's two very different behaviours hiding behind the same panel. "Bid higher for new customers" and "New customers only" sound like siblings. They are not. One nudges the auction, the other slams a door and choosing between them is the whole decision. This is a targeting choice as much as a bidding one, which is why I file new customer acquisition goal performance max work under targeting rather than creative. If you want the wider context on how this campaign type hands levers to automation, my primer on what Performance Max is sets the stage I'm building on here.

What follows isn't a settings screenshot tour. It's what I saw when I actually ran both modes on live accounts: which one I reach for by default, the specific situation where I flipped a campaign to the hard filter to stop bleeding budget on people we already had, and the part everyone actually wants what happened to CPA when I measured it properly instead of glancing at the blended number.

What the new customer acquisition goal actually does

Before the two modes make any sense, you have to know what the goal is optimising against. The new customer acquisition goal in Performance Max works by first drawing a line between "new" and "existing" customers, then changing how the campaign bids on each side of that line. Google draws the line two ways: with your first-party data a Customer Match list of people who've already bought and, on top of or instead of that, with its own autodetection, which looks back over your account's purchase history and treats anyone who hasn't converted in that window as new. The stronger and cleaner your first-party seed, the sharper that line is, and the less the campaign guesses.

This is also where the goal quietly depends on your measurement. It can only tell a new customer from a returning one if your conversion setup is passing that signal the new-customer parameter on the purchase event, plus enough historical conversion data to build the returning-customer audience. If that plumbing isn't right, the goal is drawing its line with a blindfold on, which is why I don't touch this setting until conversion tracking is verified and firing a clean purchase event. Get that wrong and every downstream number including the CPA figures I'll get to is measuring noise. This dependency is the single biggest reason a new customer acquisition goal performance max campaign underdelivers despite being switched on.

One detail worth flagging up front, because it catches people out: Google's own documentation is explicit that the value-based modes require at least one Purchase conversion goal, while the hard-filter mode technically doesn't but strongly recommends one for accurate new-vs-returning reporting. The full breakdown of the modes and their requirements lives in Google's customer lifecycle goals help page, and it's the reference I check before enabling any of this on a lead-gen account where "purchase" doesn't map cleanly.

Bid higher vs new only: the difference that actually matters

Here's the distinction the panel makes look trivial and that in practice changes everything about how the campaign spends.

"Bid higher for new customers" a multiplier, not a filter

This mode leaves the campaign free to serve to everyone, but adds extra conversion value to new customers so smart bidding pushes harder to win them. You're telling the auction "a new customer is worth more to me than a repeat one" say an extra fixed value on top of the purchase and the bidder re-weights accordingly. Crucially, it never stops bidding on existing customers; it just makes them relatively less attractive. This is Google's recommended default, and it's mine too for most accounts, because it tilts the campaign toward acquisition without amputating reach. Nothing gets locked out the machine still has the full auction to work with, it's just leaning.

"New customers only" a hard filter with a cost

This mode does what it says: the campaign is optimised to bid exclusively for people the system classifies as new. It doesn't lean, it excludes. And that exclusion is exactly why Google, and I, treat it as the more aggressive choice you reach for deliberately, not by default. Two things follow directly from it. First, the reason to use it: it genuinely stops the campaign from spending on customers you already own no more paying to "acquire" a buyer who'd have come back regardless. Second, the cost: because it filters the auction down to a slice of demand, it serves less, and Google is candid that it limits reach. There's a structural consequence too since this mode won't touch your existing base, you need a separate campaign to keep reaching those customers, or you simply stop marketing to them. I don't flip this switch without deciding what happens to the existing-customer audience it's now ignoring.

The mental model I use: "bid higher" changes the weighting of the same auction; "new only" changes the size of the auction. That's why they behave nothing alike in the reports, and why comparing their CPAs naïvely leads people to the wrong conclusion which is the next section.

When I used it to stop paying for existing customers and the real CPA impact

The account that made this concrete was a repeat-purchase ecommerce brand with a loyal base. Their PMax campaign looked healthy on a blended view, but when I split conversions into new versus returning, a large share of spend was going to people who'd bought before the campaign was, in effect, buying back its own customers at a premium. The brief was growth, not retention, so we were paying acquisition prices for retention outcomes. That's the exact scenario the performance max new customer acquisition goal is built for.

I didn't start with the hard filter. I opened on "bid higher for new customers," because it's the reversible, lower-risk move: keep the reach, tilt the weighting. The effect on the numbers was the one people find counterintuitive at first. The campaign's blended CPA drifted up, because new customers are more expensive to win than repeat buyers, so as the mix shifted toward them the average cost per conversion rose. But the new-customer CPA the metric that actually mapped to the client's growth goal improved relative to what we'd been implicitly paying, because the budget was now working to find fresh demand instead of harvesting the base. The headline number went the "wrong" way while the number that mattered went the right way. If you only watch blended CPA, you conclude the setting hurt you. It didn't; it re-pointed you.

We only moved to "new customers only" on a second campaign, deliberately carved out, once we'd confirmed the base was being served elsewhere. There, the hard filter did stop the existing-customer spend almost entirely that was the point and new-customer CPA on that campaign was cleaner still, at the price of noticeably lower volume and a campaign that would sit idle if we ever turned off the existing-customer coverage. The trade was explicit: we accepted less reach in exchange for spend that was almost purely acquisition. On accounts without that separate coverage, I'd have left it on "bid higher" and not touched the filter at all.

The bidding side of this matters as much as the goal itself, because the value-based modes only behave when the bid strategy can act on value. On a campaign where I'm adding value to new customers, I want a value-based strategy carrying that signal which ties directly into how I choose among Performance Max bidding strategies: the acquisition goal sets the priority, the bid strategy is what executes it, and a mismatch between the two is a common reason the goal appears to do nothing. Set a target too tight on top of a fresh acquisition tilt and you'll strangle the very serving you were trying to redirect.

The blunt takeaway from measuring it: the performance max new customer acquisition goal rarely lowers your reported CPA, and expecting it to is how clients get spooked. What it does is change who that spend buys. Judge it on new-customer cost and acquisition volume, not on the blended average, or you'll kill a setting that's quietly doing its job.

Where the goal goes wrong on real accounts

Almost every time a new customer acquisition goal performance max setup "doesn't work," it's one of a short list of configuration problems rather than the goal itself failing.

The first is missing measurement: no purchase conversion goal on a value-based mode, or a conversion tag that never got the new-customer parameter, so Google can't classify new versus returning and the whole thing runs blind. The second is a thin or absent first-party seed relying entirely on autodetection when a Customer Match list would sharpen the line dramatically; the goal is only as good as the data telling it who's already a customer. The third is the reach surprise: someone flips "new customers only," serving drops, and they panic instead of realising that's the mode doing exactly what it promises if reach matters more than purity, "bid higher" is the correct choice and Google explicitly recommends stepping back to it. The fourth, and the one I care about most, is the reporting error I covered above: reading blended CPA and concluding the goal hurt performance, when the honest comparison is new-customer CPA before and after.

There's a structural point underneath all four. The acquisition goal decides who you bid for, but it can't fix a campaign that's structured to spend on the wrong inventory in the first place if the underlying product feed and asset groups aren't organised so budget follows what converts, a cleaner audience just gets pointed at the same mess. That structural layer is a separate discipline I break down in my PMax product group strategy, and it's the foundation the acquisition goal sits on top of. Fix the measurement, feed a real customer list, choose the mode that matches whether you can afford the reach hit, and read the right CPA do those four and the goal stops being a mystery and starts being a lever.