Most Performance Max guides explain what Google’s help docs already say. This one is different: I’m going to show you 90 days of real PMax spend from a D2C health-tech brand in India I work with — ₹1.63 crore between May and August 2026 — next to the same account’s Search campaigns, so you can see what PMax actually does when money is on the line.
What a PMax campaign is (the 60-second version)
Performance Max is Google’s goal-based campaign type: you hand it conversion goals, budget, and assets (text, images, video), and it buys across Search, Shopping, YouTube, Display, Gmail, Discover and Maps from a single campaign. You give up channel-level control; in exchange, Google’s bidding gets a bigger surface to find converters on. Whether that trade is worth it is an empirical question — so here’s the empirical answer from one real account.
Real data: PMax vs Search in the same account, same 90 days
| Metric (May–Aug 2026) | Performance Max | Search |
|---|---|---|
| Spend | ₹1.63 Cr | ₹1.08 Cr |
| CPC | ₹2.19 | ₹20.93 |
| CTR | 3.6% | 10.5% |
| Cost per conversion | ₹670 | ₹714 |
| ROAS | 1.50 | 1.46 |
Read that CPC row again: PMax clicks cost a tenth of Search clicks — because most of them come from Display, Discover and YouTube inventory, not the search results page. And yet cost per conversion lands within 7% of Search. That’s the whole PMax thesis in one table: individually weaker clicks, aggregated by a bidder that knows which ones to buy.
The more interesting finding: what happened as spend scaled
Month by month, same account:
| Month | PMax spend | PMax ROAS | Search spend | Search ROAS |
|---|---|---|---|---|
| May | ₹21.9 L | 1.52 | ₹5.9 L | 2.35 |
| June | ₹56.7 L | 1.55 | ₹26.9 L | 1.84 |
| July | ₹44.6 L | 1.57 | ₹53.5 L | 1.35 |
| Aug (partial) | ₹39.3 L | 1.34 | ₹21.4 L | 1.03 |
Search ROAS collapsed from 2.35 to ~1.0 as its spend scaled 9x — classic diminishing returns as you exhaust high-intent queries. PMax held between 1.5 and 1.57 while its spend grew 2.6x. This is the honest case for PMax in 2026: it’s not that it beats Search head-to-head at small budgets (it usually doesn’t), it’s that it degrades far more gracefully when you need volume.
The dos (earned the hard way)
- Feed it your own conversion definitions, ruthlessly. PMax optimizes to whatever you count. Count junk (page views, cheap micro-events) and it buys junk — I’ve watched a PMax campaign post six-figure “conversions” that were worthless engagements. Purchase or qualified lead only.
- One asset group per product theme, not per audience. Audience signals are hints, not targeting. Themes with dedicated creative are what change results.
- Give it real video. If you don’t, Google auto-generates a slideshow that runs on YouTube with your money. The auto-generated stuff is where CPMs go to die.
- Scale in 20–30% steps. The May→June jump above (2.6x) worked, but ROAS wobbles for ~a week after each step. Bigger jumps re-enter learning.
- Add campaign-level negative keywords (finally available to all accounts) — brand terms first, so PMax stops cannibalizing traffic you’d get free.
The don’ts
- Don’t run PMax without a brand exclusion and then celebrate its ROAS. A chunk of it is your own brand demand re-sold to you.
- Don’t judge it inside two weeks. The table above shows monthly stability, but week one after any change is noise.
- Don’t let it replace Search for high-intent queries. In this account Search at modest spend (May: ₹5.9L, ROAS 2.35) beat PMax comfortably. The right structure was Search capped where it’s efficient, PMax for incremental volume — not either/or.
- Don’t starve it. Below roughly 30 conversions/month per campaign, the bidder is guessing. Consolidate before you fragment.
Performance max ads examples: what the asset mix looked like
For the account above: 3 asset groups (hero product, category, offer), each with 5 headlines under 30 characters leading with the problem not the brand, 4–5 lifestyle + packshot images, and one 15-second vertical video per group. The offer-led asset group consistently pulled the highest conversion share — matching what the CTR data implies: PMax surfaces respond to direct-response creative, not brand films.
FAQ
Is PMax better than Search?
At small budgets, usually no — this account’s Search ROAS was 2.35 at ₹5.9L/month. At scale, PMax degraded far less: it held ~1.5 ROAS while Search fell to ~1.0. Use Search for efficiency, PMax for volume.
How much budget does a PMax campaign need?
Enough for ~30+ conversions a month at minimum. At this account’s ₹670 cost per conversion, that meant ~₹20,000+ per month as an absolute floor — and it performed best above ₹40L/month.
How long does PMax take to learn?
Expect noisy performance for 1–2 weeks after launch or any 20%+ budget change. Judge on 30-day windows.
Data note: figures from a D2C health-tech advertiser in India managed by the author, May–Aug 2026, anonymized. Past results in one account, not a projection of yours.
Comparing channels for your own budget? See the companion Facebook ads cost breakdown from the same 90-day window, or get in touch if you want a second opinion on your account structure.