When Google replaced Discovery campaigns with Demand Gen in 2024, most marketers treated it as "the new Discovery — basically the same thing." When Performance Max launched in 2021, the same thing happened — "it's just Smart Shopping plus more inventory."
Both descriptions are wrong, and the confusion is costing accounts money. Demand Gen and Performance Max overlap significantly. They also compete with each other for budget and inventory if you don't structure them carefully.
I've run both side by side on the same accounts, same products, same budget for 90 days. Here's what each is actually for, where they overlap, and the decision rules I use to pick between them (or run them together properly).
The TL;DR: Performance Max is best for bottom-funnel conversion when your conversion data is mature. Demand Gen is best for top and mid-funnel demand creation on YouTube, Discover, and Gmail with creative-led ads. They overlap on Display. Run both, not as competitors, but as a layered funnel — Demand Gen feeds awareness, PMax closes.
What Demand Gen actually is
Demand Gen runs ads across:
- YouTube (in-feed, in-stream, Shorts).
- Discover (the Google app feed).
- Gmail Promotions tab.
It's heavily creative-focused. Image and video are mandatory. Targeting is based on audience signals (custom segments, lookalikes, in-market audiences).
You can bid with:
- Maximize Conversions
- Maximize Clicks
- Target CPA
- Target ROAS
It's optimised for demand creation, not demand harvesting. Think of it as the channel for getting your brand in front of people who weren't searching for you yet.
What Performance Max actually is
Performance Max runs ads across:
- Search (the part most underestimated).
- Shopping (for e-commerce with a Merchant Centre feed).
- YouTube (overlap with Demand Gen here).
- Discover (overlap).
- Gmail (overlap).
- Display Network.
- Google Maps.
It's optimised for conversion. Smart Bidding is fully automated. Audience signals are advisory, not restrictive — Google decides who to show to.
The Search component is what separates it from Demand Gen. PMax can capture high-intent search queries Demand Gen never sees.
Where they overlap (and fight each other)
Both campaigns can serve ads on:
- YouTube
- Discover
- Gmail Promotions
- (Some) Display inventory
If you run both targeting the same audience, you'll see auction overlap. Google's official position is "they coexist fine." My actual finding is that they sometimes pull from the same budget bucket and cannibalise each other if you don't structure them carefully.
The right framing isn't "Demand Gen vs PMax." It's "what role does each play in my funnel, and how do I prevent overlap from inflating CPCs?"
The side-by-side test
I ran both on the same e-commerce account selling a niche fitness product. Same products, same targeting where possible, equivalent budget.
Setup
- Budget: £15k/month each.
- Duration: 90 days.
- Same conversion action, same value rules.
- Both at Target ROAS 400%.
Demand Gen results
- Impressions: Very high (mostly YouTube and Discover).
- CTR: Low (~0.9%), typical for top-funnel inventory.
- CPM: Cheap.
- Direct conversion rate: Low (~0.4%).
- Assisted conversions: High — Demand Gen showed up in many conversion paths even when it wasn't the last-click.
- CPA on direct conversions: £42.
- ROAS on last-click model: ~310% (under target).
- ROAS on data-driven attribution: ~520% (above target, because Demand Gen helped move people down the funnel).
Performance Max results
- Impressions: Lower than Demand Gen but more focused.
- CTR: Higher (~2.4%), mostly because of Search component.
- CPM: Higher.
- Direct conversion rate: ~3.2%.
- CPA: £18.
- ROAS on last-click model: ~620%.
- ROAS on data-driven attribution: ~580% (about the same, because PMax mostly closes deals others assisted).
Combined view
When attributed under data-driven attribution (which credits multi-touch journeys), the combination delivered:
- Total spend: £30k.
- Total revenue: £156k.
- Blended ROAS: ~520%.
When I ran just PMax with the £30k combined budget, total revenue was £142k. Slightly less.
When I ran just Demand Gen with the £30k combined budget, total revenue was £71k. Significantly less.
The combined funnel outperformed either campaign alone. Demand Gen warmed people up. PMax closed them. Removing Demand Gen reduced PMax efficiency over time as the audience pool warmed up by Demand Gen dried up.
When to use Demand Gen (and not PMax)
- Your audience doesn't search for you yet because they don't know they have the problem.
- You have strong visual creative (video, lifestyle photography, story-driven ads).
- You're trying to build a remarketing audience for future Search/PMax campaigns.
- You're targeting new geographies where brand awareness is low.
- Your conversion data is still maturing (under 30 conversions/month) and PMax can't learn yet.
When to use Performance Max (and not Demand Gen)
- You have lots of conversion data (50+ conversions/month per campaign minimum).
- Your product feed is clean (for Shopping inventory).
- The buying intent already exists and you're trying to capture it efficiently.
- You're a direct-response advertiser with measurable, fast-firing conversions.
- You need Search inventory at scale without managing keyword lists.
When to use both (most accounts)
- You're running an account with £20k+/month in spend.
- Your buyer journey takes more than a few days (research → consideration → purchase).
- You have both top-funnel creative AND bottom-funnel conversion data.
- You can manage the audience and asset overlap to prevent self-cannibalisation.
How to structure both running together without cannibalisation
Three rules:
1. Use audience signals strategically
In Demand Gen, target broader top-of-funnel audiences (lookalikes, in-market, custom segments built from competitor content).
In PMax, your audience signals should be converters and high-intent visitors — your customer match list, your cart abandoners, your remarketing pools.
This keeps each campaign focused on its part of the funnel.
2. Separate creative
Demand Gen creative: storytelling, brand-building, problem-introducing.
PMax creative: product-led, price/offer-driven, conversion-focused.
This isn't just stylistic. Google's AI uses your creative to decide where to show ads. Generic creative dilutes both.
3. Use account-level brand exclusions in PMax
Stop PMax from serving on your own brand searches. Let your dedicated brand Search campaign capture those. This forces PMax to find net-new buyers, not steal credit from organic and brand.
Use case: a multi-brand event ticketing business
A composite based on patterns from a couple of accounts I've seen.
A regional events business sold tickets across concerts, festivals, and theatre. Total Google Ads budget: ~£60k/month. They ran PMax + Search only, no Demand Gen.
Symptoms:
- Spend was efficient on existing demand (people searching "concert tickets [city]") but stagnant. Same audience week to week.
- New artist launches struggled — no awareness of the artist meant no search demand.
- They were spending heavily on Meta to fill the awareness gap, but cross-channel attribution was broken.
We added Demand Gen with three asset groups:
- Artist storytelling (video) for upcoming concerts 3-6 weeks out.
- Festival lineup reveals (carousel) on a 6-week pre-event cycle.
- Theatre evergreen (image + video) for the always-on theatre programme.
Results over 120 days:
- Demand Gen brought 38% more new converters into the funnel (people with no prior interaction with the site).
- PMax efficiency improved over time as the audience pool warmed up (CPA dropped 14%).
- Search impression share for upcoming-event branded queries (e.g., "[artist name] tickets [city]") rose 60% — people who saw Demand Gen later searched the artist directly.
- Total revenue rose 22% on a 17% budget increase. ROAS improved.
The Demand Gen "wasn't profitable on its own" by last-click attribution. By data-driven attribution and total account impact, it was the highest-leverage budget added that quarter.
Common mistakes
- Judging Demand Gen by last-click ROAS. It's a top-funnel channel. Use data-driven attribution.
- Running both with the same audience signals. Self-cannibalisation. Differentiate by funnel stage.
- Skipping video creative in Demand Gen. Video is where Demand Gen punches hardest. Image-only Demand Gen underperforms.
- Letting PMax serve on brand searches. Use brand exclusions in PMax. Capture brand in a dedicated campaign.
- Treating Discovery campaigns as a Demand Gen drop-in. Google migrated Discovery to Demand Gen automatically, but the optimisation behaviour is different. Re-evaluate, don't just inherit.
Bottom line
Demand Gen and Performance Max are not competitors. They're layers in a funnel.
- Demand Gen creates demand through visual, story-led creative on YouTube, Discover, and Gmail.
- Performance Max harvests demand across Search, Shopping, and everywhere else, optimised for conversion.
- The combination outperforms either alone for most account sizes above ~£20k/month.
- Separate them by audience signal, creative type, and funnel stage to prevent cannibalisation.
If you've been running only one and wondering why growth has plateaued, the missing layer is probably the other one.
Sources and further reading: