The question most comparison posts dodge: these are not two ways to do the same job. One harvests buyers who are already looking for what you sell. The other goes and finds buyers who aren't looking yet. Picking "the better one" is the wrong frame — you pick the one that fits where your account actually is.
The one-line difference
Performance Max is a demand-capture machine. Demand Gen is a demand-creation machine. Google itself drew that line formally in 2025, casting Performance Max as end-to-end conversion automation and Demand Gen as controlled visual placement.
For a retailer, Performance Max runs mostly on your Merchant Center feed — smec, which tracks over €450M in annual European retail ad spend, reports that 74–97% of Performance Max cost goes to feed-based Shopping-style ads. Demand Gen, by contrast, is a visual-first format: product carousels, video, and image ads on YouTube, the Discover feed, and Gmail.
That difference drives everything else — the funnel stage, the creative, and, as you'll see, the amount of conversion data each one demands before it behaves.
What Performance Max actually does for an operating store
Performance Max serves one campaign across Search, Shopping, Display, YouTube, Discover, Gmail, and Maps, driven by your feed plus asset groups and a target. You supply the feed and a goal; Google decides placement and bid per auction.
For a store with real sales history, this is where the money is: it buys the auctions closest to purchase. The old "black box" complaint is stale — since 2025 you can read a channel performance report and full search terms, so you can see what share of conversions came from Search versus Display.
The honest watch-out is brand cannibalization. Performance Max chases the cheapest conversions, and your own branded searches are the cheapest of all, so it will happily claim orders you were getting anyway and report a glowing ROAS for it. The fix exists — account-level brand exclusions plus a dedicated brand Search campaign — and it is covered in depth in our Google Ads economics guide.
What Demand Gen actually does
Demand Gen runs on Google's visual surfaces — YouTube (in-feed, in-stream, and Shorts), the Discover feed, and the Gmail promotions tab. Its job is to put your product in front of people who weren't searching for it, so it is an upper- and mid-funnel play, not a closer.
The 2026 updates made it far more commerce-ready. Google now lets you attach a product feed, and rolled out Checkout Links that send a shopper from a YouTube ad straight to the cart — Google reports a roughly 6% average conversion lift from cutting that navigation friction. Treat that as a Google-reported average, not a promise.
Feed adoption is climbing fast. smec's data shows the share of Demand Gen campaigns using a product feed went from 16% in 2024 to 26% in 2025 to 35% in 2026 — roughly one in three by mid-2026.
The real decision lever: conversion volume
Here is the fact that actually decides the order of operations, and almost every comparison article skips it. The two campaign types have very different appetites for data.
For Search and Shopping, Target ROAS needs at least 15 conversions with values in the past 30 days to bid on — that is the floor Performance Max inherits for its feed-driven core. Demand Gen's Target ROAS floor is much higher: 50 conversions in the past 35 days, per the same Google documentation.
That gap is the whole argument for running Performance Max first. A store doing 120 orders a month clears the Performance Max floor comfortably but will starve a Demand Gen campaign's bidding engine. Practitioner budget guidance reflects this — Store Growers suggests Demand Gen wants around $100/day or 20x your target CPA to optimize well, a meaningfully bigger commitment than a feed-only Performance Max campaign needs.
So the sequence is: get Performance Max profitable and feeding you stable conversion volume first. Add Demand Gen when you have the order count and the budget to let it exit the learning phase cleanly.
The number that governs both: your break-even ROAS
Neither campaign should be judged against a blog's "good ROAS" benchmark. Both are governed by one number from your own P&L, and it's pure arithmetic:
Break-even ROAS = 1 ÷ contribution margin.
Say you run a POD store doing 420 orders a month at a $34 average order value, already spending $3,200/month on Google. On a $34 item that costs $15 to print and ship with about $1.50 in payment fees, your contribution is $34 − $15 − $1.50 = $17.50. That's a 51% margin, so your break-even is 1 ÷ 0.51 = 1.94x — call it 2.0x.
Add a profit buffer (practitioners commonly use 1.2–1.5x) and you set your Target ROAS around 2.0 × 1.3 = 2.6x, or 260%. That is the number you type into both campaigns — derived from your margin, not from a benchmark. Setting it far above what the account has historically earned doesn't force efficiency; Google's own documentation warns an over-high target "may limit the amount of traffic your ads may get."
For context on what clicks cost before they convert, LocaliQ's 2026 benchmarks put the retail-category search CPC at $4.14 and conversion rate at 4.01%, while WebFX pegs ecommerce conversion at 2–3% — ranges, not targets. Your break-even math is what makes a $4 click cheap or ruinous.
Running them together: the 2026 consensus
The strongest accounts don't choose. A Fospha and Google study of 25 retail brands over the 2025 holiday window found brands that put 10–20% of their Google budget into Demand Gen earned double the ROAS of brands spending under 5% on it, and retailers adding two channels to the mix saw 37% better ROAS than those adding none.
The catch is overlap. smec warns that feed-based Demand Gen can carry an uncomfortably high degree of overlap with Performance Max when both run on identical product data in the same account. Measure them on separate jobs: Demand Gen on new-customer reach and assisted conversions, Performance Max on last-click efficiency. Judge total business orders against total ad spend, not each campaign's self-reported ROAS.
If you're weighing how to divide money across your Google campaigns, our breakdown of the Performance Max budget split walks the allocation math, and if you're deciding whether to hand the account to an outside team, the Google Shopping ads agency guide covers what's worth paying for.
Where Victor fits
The hard part isn't knowing that break-even is 2.0x — it's holding both campaigns to it every day while the auction moves. That's the work Victor does. PodVector AI's Victor is an AI employee who runs your Google Ads account as a full operator: he adjusts budgets and bids, pauses the campaign that drifts below your break-even, adds keywords, and edits ads and audiences.
Because Victor also sees every order, every supplier cost from Printify, Printful, and Gelato, and every ad dollar together, he computes the true per-order profit on each product — so he's setting those Target ROAS numbers against real margin, not reported ROAS. He makes the same changes inside Shopify, Meta Ads, and Klaviyo, and delivers the reconciled report to your Google Drive. Put Victor on your Google Ads and let him hold the line on profit across both campaign types.
FAQs
Is Demand Gen replacing Performance Max?
No. Google defined them as distinct roles — Performance Max for full-funnel conversion automation, Demand Gen for demand creation on visual surfaces. They do different jobs, and the 2026 field consensus is to run both once you have the volume to feed Demand Gen.
Which one should a smaller store run first?
Performance Max, almost always. Its feed-driven core bids on as few as 15 conversions in 30 days, while Demand Gen's Target ROAS wants 50 in 35 days. A store without that order volume will watch a Demand Gen campaign thrash in learning while Performance Max is already profitable.
Does Demand Gen have lower CPCs than Performance Max?
Often, because it buys cheaper upper-funnel visual inventory rather than high-intent Search and Shopping auctions. But a cheaper click that doesn't convert is not cheaper acquisition — judge both on cost per order against your break-even ROAS, not on CPC.
Can I just run both with the same product feed?
You can, but mind the overlap. smec warns feed-based Demand Gen can heavily overlap Performance Max on identical data. Give each a distinct job and measure total business results, not per-campaign ROAS, so you don't pay twice for the same order.
What Target ROAS should I set on each?
The same number, derived from your margin: break-even ROAS (1 ÷ contribution margin) times a profit buffer of roughly 1.2–1.5. A 50% margin means a 2.0x break-even and a target near 2.6x. Setting it far above your account's demonstrated ROAS suppresses volume rather than improving efficiency.
How do I know Demand Gen is actually adding customers, not stealing credit?
Watch incrementality, not the campaign's own ROAS. Check whether total orders and new-customer rate rose after Demand Gen launched, the way the Fospha study measured added-channel lift — if blended business results are flat while Demand Gen reports wins, it's reshuffling credit, not creating demand.