The client is a UAE-based B2B service provider — small niche, modest search volume, single-country targeting. They'd been running a Search campaign for four months at $17/day with steady lead flow but flat scaling. The question we wanted to answer was simple: if we take the same money and the same creative and put it into Performance Max instead, do we get more, fewer, or the same leads?
Everyone has an opinion. Very few have a controlled test on a small budget in a small market. We ran the test for eight weeks and pulled the tape.
The Experimental Setup
| Parameter | Search (control) | PMax (variant) |
|---|---|---|
| Daily budget | $17 | $17 |
| Bidding strategy | Maximize Conversions (target CPA off) | Maximize Conversions (target CPA off) |
| Ad copy | 15 headlines / 4 descriptions | Same 15 headlines / 4 descriptions in the asset group |
| Images / video | N/A | 10 stock images relevant to the service |
| Geo | UAE, presence only | UAE, presence only |
| Landing pages | Same 3 URLs | Same 3 URLs (no final URL expansion) |
| Conversion goal | Single primary — form submit + phone call | Same single primary |
PMax ran with URL expansion turned off — a critical control. Final URL expansion is the switch that lets PMax invent landing pages from your domain, and it introduces so much variability that any A/B comparison falls apart. We locked it to the same three URLs Search used.
Week-by-Week Results
| Week | Search conv. | Search CPA | PMax conv. | PMax CPA |
|---|---|---|---|---|
| 1 | 4 | $29.75 | 1 | $119 |
| 2 | 5 | $23.80 | 2 | $59.50 |
| 3 | 3 | $39.66 | 3 | $39.66 |
| 4 | 4 | $29.75 | 4 | $29.75 |
| 5 | 6 | $19.83 | 5 | $23.80 |
| 6 | 4 | $29.75 | 3 | $39.66 |
| 7 | 5 | $23.80 | 2 | $59.50 |
| 8 | 5 | $23.80 | 2 | $59.50 |
Search: 36 conversions total, $952 spend, $26.44 average CPA.
PMax: 22 conversions total, $952 spend, $43.27 average CPA.
The Learning Curve Was Not the Story
The common defense of a losing PMax test is “it needs more time to learn.” And weeks 1–4 do look like a learning curve — CPA fell from $119 to $29.75. Reasonable people would call the trend positive. Then weeks 5–8 came in and the pattern reversed. PMax found a plateau at $30–60 CPA and stayed there. Search stayed at $20–30 the entire time.
What we saw was not a learning curve that never finished — it was PMax spending its way into the same low-hanging fruit Search was already catching, then bleeding excess budget on Display and YouTube inventory at 3× the cost. In a small market with maybe 20–30 qualified searches per day, that inventory doesn't contain enough incremental buyers to justify itself.
What This Means for Small-Budget Decisions
| Situation | Better bet | Why |
|---|---|---|
| <$20/day, small market, B2B lead-gen | Search | Not enough non-Search inventory to justify PMax's spread |
| $50–150/day, e-commerce, product catalog | PMax + Search + separate brand Search | Catalog fuels Shopping side of PMax; the split works |
| Any budget, service business, no product feed | Search (with feed-less PMax as a small side test) | Feed-less PMax is 2–4× more expensive per lead in most tests |
| >$500/day, large market, both audiences known | Both, structured to avoid overlap | At this budget, PMax finds real net-new inventory Search misses |
The small-market rule that came out of this test is simple: PMax needs enough non-Search inventory to justify its channel spread. Small markets don't have that inventory. Below roughly $30/day in a market of 20–50 qualified daily searches, PMax cannot beat a well-run Search campaign on the same budget.
What Would Have Made PMax Win
We're not anti-PMax. We run PMax profitably on several e-commerce clients at 3–4× the budget. What would have changed the outcome of this specific test:
- Higher budget. $50/day would have given PMax more room to feed its Display/YouTube channels without starving the Search side. Below $20/day, spread is punishing.
- A product feed. PMax lives or dies on Shopping. Without a Merchant Center feed, PMax runs on Search + Display + YouTube only, and it costs more per conversion in almost every market we've tested.
- A larger market. This test was UAE-only. A pan-GCC or English-speaking Middle East target would have given PMax more inventory density to work with.
- Retargeting-heavy goals. PMax excels at retargeting assist. If the goal had been assisted conversions rather than direct leads, the numbers might have flipped.
How to Run This Test on Your Own Account
Copy every headline and description from your Search RSA into the PMax asset group. Turn URL expansion off. Any variance in creative kills the comparison.
Add your brand terms as negatives on both campaigns and run a separate small brand Search campaign. Otherwise PMax will hoover up your brand queries and look like it's winning.
PMax's learning phase is real. 8 weeks catches the phase-plus- plateau shape. 4 weeks catches only the phase and produces a false optimistic read.
Pull the PMax channel split. If >60% of PMax spend went to non-Search channels and <40% of PMax conversions came from them, your market probably can't support PMax at this budget level.
The Rule of Thumb We Now Use
Under $30/day in a small market with no product feed, run Search. At $30–100/day, run both if you can afford the setup — the comparison itself is more valuable than the outcome. Above $100/day with a feed, PMax earns its spot in the mix, and the question shifts from “which one” to “how to structure them so they don't overlap.”
More Google Ads campaign structure patterns in the Sterling Lab blog — including our full breakdown of when PMax wins, when Search wins, and how to combine them, and how to structure Google Ads across multiple countries.
