Performance Max New Customer Goal Won't Block Bad Queries

The new customer acquisition goal is usually sold as a fix for "low-value traffic" in Performance Max. That's half true. It reduces one kind of low-value traffic: people who would have bought from you anyway. It does nothing about the other kind, which is people searching for something you don't sell. Most of the pages that rank for this topic explain how to set the goal up: customer lists, conversion tags, the lookback window, value-more versus new-only. Very few point out that the goal works on a different level from query waste. This article is about that difference.

What the new customer goal actually changes

The goal changes how much Performance Max bids for a person, based on whether Google thinks that person has bought from you before. It doesn't change which searches your campaign is eligible for. Every query that could trigger your campaign before you switched it on can still trigger it afterwards.

There are two modes, and they behave differently:

• Value new customers more than existing customers. You add an extra value to a conversion from a new customer, for example $50 on top of the order value. Smart Bidding then bids more for people it predicts are new, because their conversions are worth more to it. Existing customers can still be served. They just get less weight.
• Only bid for new customers. The campaign tries to avoid bidding on people identified as existing customers. Everyone else stays in play.

How does Google decide who counts as "existing"? It uses the data you give it: Customer Match lists, conversion tags that pass a new-customer flag, and Google's own detection based on past conversions. If you upload a weak list, you get a weak split. If your customers buy logged out on a different device from the one they search on, Google may not match them at all.

Note what's missing from that list: the search term. The goal's decision is about the user, not the query. It's the same idea we cover in Performance Max audience signals are not targeting: telling Google about people shapes the bidding, but it doesn't fence off any searches.

The waste it does reduce

The goal is good at one thing: spending less on people who were going to buy from you anyway. In Performance Max that's a real cost, because the campaign likes to claim conversions from your existing customers.

Here's how that waste usually shows up:

• Repeat buyers searching your brand. A customer who has ordered three times types your store name, clicks the Performance Max ad, and buys. Performance Max records a conversion. You paid for a visit that would probably have come through organic search or a bookmark.
• Remarketing dressed up as prospecting. Performance Max serves on Display, YouTube and Gmail as well as Search. A lot of those conversions come from people who already know you. The goal lowers bids for them too.
• Subscription and replenishment customers. Someone reordering coffee pods every month doesn't need an ad to remember you.

If your Performance Max ROAS looks great but total revenue doesn't move when you raise the budget, this is often why. Switching on the goal, especially in new-only mode, is a fair test of how much of the campaign's "performance" was really existing demand.

Branded search is the clearest example, and it's only partly solved. The goal lowers bids for existing customers who search your brand. It does nothing about new people searching your brand, and those clicks can be cheap demand you would have won organically. If brand spend is your worry, measure it directly: see Performance Max branded search spend and how to measure it. Brand exclusions have gaps of their own, covered in Performance Max brand exclusions and what they miss.

The waste it leaves untouched: irrelevant queries

An irrelevant query from a new user is still an irrelevant query, and the goal treats it as a new customer you want. It has no way to tell "a new person who wants to buy boots" from "a new person who wants to repair boots they bought somewhere else".

Take a store that sells leather boots. Here are search terms that commonly turn up in a Performance Max campaign like that:

Only one row gets cheaper. Most of the bottom rows come from people who have never bought from you. That's exactly why they're irrelevant: they aren't in your market. So the goal sees them as the audience it's supposed to chase.

This is the gap in most write-ups on the goal. They talk about "low-value traffic" as if it all came from existing customers. In the Performance Max accounts we audit, the clicks that were never going to convert are mostly new people with the wrong intent. Returning buyers getting an ad they didn't need are the smaller part. Those are different problems, and they need different tools.

Why it can make query waste worse

With your budget unchanged, any money the goal takes away from existing customers gets spent somewhere else in the same campaign, and some of it lands on irrelevant queries. Performance Max doesn't hand back budget it no longer wants to spend on returning buyers. It moves that money to other users and other searches.

Here's a simple illustration with round numbers. These figures are illustrative. They aren't measured from any account.

Before the goal, a campaign spends $10,000 a month:

• $2,000 on existing customers (brand searches, remarketing-style conversions)
• $2,000 on irrelevant queries (the bottom rows of the table above)
• $6,000 on relevant prospecting

After switching on "only bid for new customers", existing-customer spend falls to $400. Identification is never perfect, so it doesn't reach zero. That frees $1,600. The budget is still $10,000, so the $1,600 goes to the remaining non-existing traffic. Before the switch, irrelevant queries were $2,000 of the $8,000 non-existing spend, or 25%. If the freed money spreads in that same proportion:

• $400 on existing customers
• $2,400 on irrelevant queries (+$400)
• $7,200 on relevant prospecting (+$1,200)

That adds up to $10,000. Relevant prospecting grew, which is the point of the goal. But the irrelevant line grew too, from $2,000 to $2,400. In practice the split won't be that even. If the irrelevant queries are cheap and high-volume, they can take more than their share.

Two more mechanics make this worse:

Higher bids for "new" users in value-more mode. If you add $50 of new-customer value, Smart Bidding sees every predicted new-customer conversion as worth more and is willing to pay more to get one. When the system wrongly expects a conversion from "boot repair near me", it now bids up a click that was never going to pay off. It's the same mistake as before, made with a bigger number.

Inflated value in your reports. The extra new-customer value feeds the bidding, and depending on how you've set up your columns, it can show up in your conversion value too. A campaign can look like it's beating its ROAS target while real revenue stays flat. Before you judge the goal, separate the new-customer value you added from actual order value, so you're reading the cash and not the bonus you assigned.

None of this makes the goal a bad idea. It means the goal moves money around inside a campaign that still has the same leaks.

How to test the goal honestly

Judge the goal on two numbers: real revenue from new customers, and spend on search terms that didn't convert. Don't judge it on ROAS alone. The first number tells you whether the goal works. The second tells you whether it moved money into waste.

A practical way to run the test:

1. Take a baseline before you switch. Export four to six weeks of search terms for the campaign. For Performance Max, Search and Shopping, these come from campaign_search_term_view in the API, or the search terms report in the interface. What the PMax search terms report actually shows explains what you get and what you don't.
2. Label the terms in three buckets: existing-customer demand (mostly brand), relevant prospecting, and irrelevant. Total the cost in each bucket.
3. Switch the goal on and change nothing else. No budget change, no new asset groups, no new feed segments. Otherwise you won't know which change did what.
4. After the learning period, export the same report and relabel. Compare cost per bucket. Did the brand bucket shrink? Where did that money go?
5. Check revenue from new customers in your own back end, not just in Google Ads. Your shop or CRM knows who's new. Google's "new customer" label is a prediction, not a fact.

Be honest about what this test can't see. Some Performance Max spend has no search term at all: Display, YouTube, Gmail and Discover placements don't report one. Even on Search, Google hides low-volume terms for privacy, so part of your spend sits in a bucket you can't label. Google Ads search terms hidden by privacy, and what to do covers how to reason about that gap. Your comparison is a comparison of the visible part. That's still useful, but don't treat it as the whole picture.

If CPA rises after the switch, expect that. New customers usually cost more to acquire than returning ones. The question is whether the extra cost is buying real new customers or paying for more "how to clean leather boots". CPA too high in Google Ads? Check search terms first walks through that diagnosis.

Pairing the goal with negative keywords in Performance Max

Use the goal to decide who is worth more. Use negative keywords to decide which searches you'll pay for at all. They don't overlap, and you need both.

The constraints in Performance Max are specific:

• A Performance Max campaign accepts up to 10,000 campaign-level negative keywords.
• It doesn't accept shared negative keyword lists. You can't build one "irrelevant" list and attach it to every Performance Max campaign the way you can in Search.
• Standard Shopping does accept negative keywords and shared lists, like Search. If you run Shopping alongside Performance Max, they're managed differently.
• The three negative match types are broad, phrase and exact. Broad negatives behave unpredictably across match types, so prefer phrase for patterns ("repair", "jobs", "wholesale") and exact for single terms you're sure of.

Some practical rules for a campaign running the new customer goal:

Block intent, not newness. "boot repair near me" should be a negative because of what the person wants, not because they're new. Don't try to recreate the goal with negatives, for example by blocking your brand name to "force" prospecting. That blocks new people searching your brand too, and the goal already handles existing customers more precisely.

Never block a term that converted recently. Under the new goal, some odd-looking terms will start bringing in genuinely new buyers. A term like "boots for wide feet" might look marginal and still convert. Check conversions before you block. Add negative keywords from search terms without guessing sets out a method.

Don't expect search themes to do the job. Search themes add reach. They don't remove it. Search themes vs negative keywords in Performance Max explains why that matters.

Watch the 10,000 limit if you work at scale. Large catalogues running Performance Max for months can produce a lot of junk terms. Phrase negatives that cover a pattern ("how to", "repair", "jobs") use that space far better than hundreds of exact terms. If your negatives don't seem to hold, check the usual causes in Performance Max negative keywords not working? Here's why.

Where PPCleaner fits, and where it doesn't

PPCleaner reads the search terms your Search, Shopping and Performance Max campaigns paid for, judges which ones were never going to convert, and adds them as negatives every night. For Performance Max, those are campaign-level negatives, because that's what the campaign accepts. It never blocks a term that converted in the last 90 days, so a new-customer term that's working stays live. Every block can be undone in one click for 90 days.

What it doesn't do matters just as much here:

• It doesn't touch the new customer goal, your bids, your budgets, your Customer Match lists or your conversion values. Whether you use value-more or new-only, and what value you assign, is your call.
• It can't act on spend with no search term. Display, Video and App report none, and neither do the non-Search placements inside Performance Max. If the goal shifts money into YouTube placements, negative keywords can't follow it there.
• It can't see terms Google hides for privacy. Those terms can't be read, so they can't be judged.

The goal and the negatives work on different problems. With the goal on and no negatives, you're telling Google to chase new people without telling it which searches don't count. That's how "boot repair near me" ends up with a higher bid than before.

A quick decision guide

Turn on the new customer goal if a large share of your Performance Max conversions come from people who already buy from you. Add negatives whatever you decide about the goal.

• Strong customer list, brand-heavy Performance Max, flat total revenue: the goal is worth testing, and new-only mode gives you the cleanest read.
• Weak or small customer list: the goal has little data to work with. Start with search-term cleanup and come back to the goal when your list is better.
• Value-more mode: pick a new-customer value you can justify from margin or repeat-purchase data, not a round number that makes ROAS look good. Then keep that value out of the revenue you report.
• Any mode: run the before/after search term comparison above. If the irrelevant bucket grew, the goal is working as designed and your negatives aren't keeping up.

The new customer goal answers "is this person worth more to me?" Negative keywords answer "should I pay for this search at all?" Most Performance Max campaigns need both answered, and neither tool can answer the other's question.