2025 Gen Z Spending Habits Test AI Ad Targeting Claims

The 2025 Gen Z spending habits record starts with a cut, not a promise. PwC’s transaction analysis found Gen Z spending down 13% from January to April 2025, based on roughly one million consumer transactions across Yodlee, Yipit, and Spatial.ai panels covering July 2020 through July 2025.[1] Piper Sandler’s fall 2025 teen survey put self-reported annual teen spending at $2,213, down 6% year over year.[2] That is the benchmark any AI ad targeting claim has to clear before it gets used to justify more spend into a Gen-Z-skewed account.

The same record does not say Gen Z stopped buying. Numerator reports that 44% of Gen Z shoppers purchased on a social platform in the past month, and that Gen Z shoppers are 82% more likely than average to say social or digital ads influence purchases.[3] PwC’s holiday work, with a base of 1,000 Gen Z consumers, found 61% prefer in-store discovery and 43% expect social discovery for gifts.[1] So the contradiction is the point: the wallet is tighter, but discovery and influence are still active in places a standard online conversion setup can undervalue.

The dated baseline before the platform lift slide

Those signals are not a Gen Z stereotype. They are a media-buying constraint. If an account sells beauty, apparel, gifts, food, events, or other products that get discovered socially and bought later, the contraction signal does not mean “turn Gen Z off.” It means the platform needs to prove that its extra reach is finding incremental buyers rather than cheaper events from people who were already waiting for a deal.

JPMorgan’s holiday path-to-purchase data makes the online-only problem more concrete. Less than 45% of Gen Z in-store spend occurred within 10 miles of home, compared with more than 50% for other cohorts; more than 55% of Gen Z were omnichannel during peak events, compared with about 30% for others; and clothing store sales share rose 1.5% in Q4 2024 versus Q4 2023.[4] A buyer optimizing only to a site purchase event can easily tag that behavior as weak, scattered, or low-intent when it is actually moving through store, social, and event-driven moments.

Why the 2025 spend record weakens the growth-cohort shortcut

The lazy version of the argument jumps from future spending power to current media scale. That jump is not supported by the 2025 record. PwC’s 13% Gen Z spending drop is a dated transaction signal. Piper Sandler’s fall 2025 teen spending decline is a dated self-reported wallet signal. They do not measure the same population in the same way, but both point in the same operational direction: 2025 youth demand should not be priced as if it were expanding by default.[1][2]

This matters because broad AI targeting systems are often evaluated against account-level efficiency metrics that can hide who changed. If the campaign finds cheaper conversions by leaning into existing bargain hunters, returning customers, older shoppers, or easier online paths, the platform may still report improvement. The account may still be no closer to building Gen Z demand at the price the budget assumed.

There is also a category trap. Piper Sandler’s beauty wallet figure was $336, down 2% year over year, a smaller decline than the broader teen spend figure.[2] Qualtrics also identified beauty as a top-three splurge category, selected by 34% in its 2025 generational splurge research.[5] For a beauty brand, that is a usable reason to keep testing Gen Z demand. For a general retailer, it is not a license to treat all Gen Z spend as resilient.

Discount behavior is the other pressure point. PwC reported discount-code searches up 14%, browsing up 17%, and more than 79% waiting for discounts.[1] In an ad account, that can make algorithmic expansion look busy without being profitable. The system can find people who click, browse, save, compare, and return when the price moves. Whether that is good media depends on margin, repeat behavior, and whether the conversion window captures the eventual purchase.

Social buying is real demand, but it is not the same as a clean website conversion

Numerator’s 44% social-platform purchase figure is too large to treat social as only awareness.[3] It also creates a measurement problem. A shopper can see a creator post, open a product page inside a social app, compare on a marketplace, wait for a promo code, and buy in a store or on a retailer site. The demand was influenced by social, but the clean conversion may land somewhere else.

That is where platform-native reporting can get overconfident. A social campaign can optimize toward the easiest in-platform or web-tracked actions. A search campaign can harvest demand after social did the shaping. A PMax campaign can pool channels in a way that raises reported conversion volume while making it harder to separate discovery, capture, and store effects. None of those outcomes is automatically bad. They are bad only when the buyer accepts the platform’s reported lift as if it described Gen Z demand itself.

Creative testing has to be read through the same lens. If Gen Z is more socially influenced than average, a stale product-grid ad may underperform even when the product has demand. But a high-engagement social unit can also overstate buying intent if the measurement plan does not connect social interactions to later orders, store visits, or customer files. The useful question is not whether Gen Z likes social content. The useful question is which social signals can survive reconciliation with the order file.

In-store discovery breaks the neatest version of AI optimization

PwC’s 61% in-store discovery preference is the kind of number that should slow down any conversion-only read on Gen Z.[1] If a campaign is judged only on online purchases, it can penalize the channels and audiences that push store discovery. That is especially easy during peak shopping periods, when Gen Z’s omnichannel behavior is more pronounced than other cohorts in JPMorgan’s data.[4]

The consequence is not theoretical. A campaign can appear to have weak Gen Z efficiency because younger shoppers browse, compare, visit stores away from home, and convert later. Or it can appear to have strong efficiency because it captures the final online event after other channels did the work. Both errors push budget in the wrong direction if the advertiser has no offline import, store-sales match, holdout, geo read, or blended customer-file check.

Split payments, buy-now-pay-later use, gift cards, and household-funded purchases can add more fragmentation, but the materials here do not provide a 2025 Gen Z benchmark for those behaviors. Treat them as plausible measurement complications, not as sourced evidence. The sourced point is narrower and strong enough: Gen Z discovery and purchase paths are socially and physically distributed, while many campaign scorecards still reward the trackable online endpoint.

Where Advantage+, AI Max, and PMax claims need pressure

Meta says Advantage+ audience “could get” 14.8% lower cost per result for Awareness objectives and 9.7% lower cost per result for Traffic, Engagement, and Leads objectives.[6] That can be useful as a product claim. It is not a Gen Z spending benchmark, not an independent incrementality read, and not proof that Gen Z demand improved in 2025. The figure comes from Meta’s own materials; the official Help Center source is also difficult to audit from a crawl because of its rendered format, so it should be labeled as platform-claimed rather than treated as external evidence.

Google’s AI Max claim has the same boundary. Google says advertisers that activate AI Max in Search campaigns typically see 14% more conversions or conversion value at a similar CPA or ROAS, and 27% for campaigns mostly using exact and phrase keywords, by expanding into net-new queries.[7] Again, that is a platform claim about a Google product. It is not Gen Z-specific. It does not say whether the extra conversions came from younger buyers, from deal-seeking searches, from branded-adjacent demand, or from queries that would hold up under incrementality testing.

Performance Max belongs in the same measurement conversation even when a specific Gen Z lift claim is not on the table. PMax can be helpful when the advertiser has clean conversion values, meaningful offline imports, and enough customer data to keep the system honest. It can also flatten a messy path into one reported outcome. For Gen-Z-skewed accounts, the danger is not automation itself. The danger is allowing automation to define the demand picture using only the signals it can see.

This is also where stated demographics become weaker than many plans assume. AI audience systems can move toward the users and contexts that satisfy the objective, even when the media plan names a generation. The same site has a separate benchmark on how AI audience targeting found Boomer upsizers when the algorithm’s conversion logic diverged from the stated audience. That is not a reason to reject broad targeting. It is a reason to verify who the system actually found.

The baseline a buyer should test against

A Gen-Z-skewed AI campaign needs a baseline that includes more than platform-reported online conversions. At minimum, the account should separate current youth-demand evidence from long-term cohort rationale, and it should reconcile platform lift with sources the platform does not control.

• Start with a dated spend assumption: 2025 Gen Z spending is constrained, not automatically expanding.
• Segment by category where possible: beauty and other selective splurge categories should not be averaged into a generic Gen Z demand claim.
• Measure social commerce separately enough to see whether social influence becomes orders, store activity, or only engagement.
• Import or reconcile offline outcomes where in-store discovery and store purchase matter.
• Read discount-driven traffic against margin and incrementality, not only against CPA.
• Audit the age and customer mix the platform actually delivers, rather than assuming the campaign stayed inside the planning cohort.

The adjacent issue is signal decay. Financial strain can change how Gen Z browses, waits, compares, and converts, which makes behavioral signals noisier for automated bidding. That companion diagnosis sits in the Gen Z financial-strain benchmark. The spend baseline here is the prior question: before modeling signal decay, the buyer has to know whether the account is assuming too much current wallet strength.

Caveats that change the read

The age bands are not perfectly aligned. PwC and NielsenIQ/GfK/World Data Lab commonly frame Gen Z around 1997–2012, while Numerator describes Gen Z as born after 1995.[1][3] Piper Sandler’s teen survey is narrower than the full Gen Z cohort.[2] Cross-source comparisons are therefore directional. They are still useful for account planning because the important pattern repeats across different lenses: current spending is pressured, while social and store discovery remain active.

Long-range spending-power projections explain why platforms and brands keep chasing the cohort, but they should not be used as 2025 demand evidence. NielsenIQ, GfK, and World Data Lab project Gen Z spending power reaching $12 trillion by 2030.[8] Bank of America Institute frames Gen Z as having roughly $36 trillion in income over five years.[9] Those projections can justify staying close to the cohort. They do not erase a 2025 wallet contraction.

Excluded figures matter too. Stale ad-skipping claims, secondary stats without original-source verification, single-region ad-fatigue findings, and small qualitative AI studies should not be mixed into this benchmark as if they were US 2025 Gen Z spend evidence. The record is already complicated without padding it with numbers that cannot carry the account decision.

The verification line

The strongest 2025 read is narrow: Gen Z demand is not dead, but it is easy to misprice. Spending signals weakened, while social purchase behavior, in-store discovery, omnichannel shopping, discount seeking, and category-specific splurges still create reachable demand. That combination is exactly where AI ad targeting can both help and mislead.

Advantage+, AI Max, and PMax should be allowed to find demand a human planner would miss. They should not be allowed to replace the baseline. For Gen-Z-skewed accounts in 2025, platform lift is only useful after it is checked against dated spend pressure, social influence, offline discovery, customer mix, and the actual order file.

References

1. The Gen Z paradox: Spending less, expecting more — PwC
2. Piper Sandler Completes 50th Semi-Annual Teen Survey — Piper Sandler, Oct. 9, 2025
3. Gen Z Consumer Behavior: Brands, Retailers and Trends — Numerator
4. Holiday Shopping Trends 2025: Gen Z Drives Retail Evolution — JPMorgan
5. What is each generation most likely to splurge on in 2025? — Qualtrics, Aug. 2025
6. Advantage+ audience — Meta Business Help Center
7. Google AI Max for Search campaigns — Google, May 6, 2025
8. Spend Z — NielsenIQ, GfK, World Data Lab
9. Gen Z: A new economic force — Bank of America Institute