Open Google Ads to create your first campaign and you’re immediately asked a question you have no way to answer: Search, or Performance Max? One of them shows text ads to people typing a query. The other spends your money across Search, YouTube, Gmail, Display, Discover and Maps, and decides the split for you.
For a SaaS with a real but modest budget, this choice matters more than any bid setting you’ll fiddle with afterwards. Here’s how the two actually differ, and the honest rule for picking.
The one difference that explains all the others
A Search campaign buys queries. You supply keywords, Google matches them to what people type, and your text ad appears above the results. You can see the exact search terms that triggered you, add negative keywords to block the bad ones, and pay for a click from someone who was, seconds earlier, describing your problem in their own words.
Performance Max buys outcomes. You hand over a conversion goal, a budget, and a pile of assets — headlines, descriptions, images, logos, video — and Google’s automation assembles ads and places them wherever its model thinks the conversion is cheapest. Instead of keywords you give search themes: phrases that hint at the intent clusters you want, without match types or bid control.
Everything else follows from that. Search is narrow, legible, and slow to scale. PMax is wide, opaque by comparison, and hungry.
PMax got less black-box — that part is real
The old advice (“never run PMax, you can’t see anything”) is out of date. Google has steadily opened it up:
- Search terms reporting is now available for Performance Max campaigns, so you can see which queries triggered you instead of guessing (Google Ads Help).
- Campaign-level negative keywords apply to Search and Shopping inventory, so you can block your own brand, job seekers, “free”, competitor names and the rest (Google Ads Help).
- Channel-level reporting arrived in January 2026, showing where the budget actually went across channels (Google Ads Developer Blog).
That’s a genuine improvement. But notice what those three things have in common: they’re all reporting and exclusion. You can now see where the money went and forbid places you don’t want. You still can’t direct it. PMax remains a campaign type you steer with signals, not with instructions.
The thing that actually decides it: conversion volume
Automation is a learning system, and learning needs examples. A Search campaign can do useful work on a handful of conversions a week because you are supplying the intelligence — the keyword list is your judgement, encoded. PMax has to derive that judgement from your conversion data, across far more inventory. Starve it, and it spends your budget exploring.
This is where most small SaaS accounts go wrong. Not “PMax is bad” — PMax is underfed.
Here’s the arithmetic. Suppose your product converts trial signups at €40 CPA:
| Monthly budget | Conversions/month | What the automation has to work with |
|---|---|---|
| €400 | ~10 | Almost nothing. Every channel gets a couple of data points. |
| €1,200 | ~30 | Thin. Enough to bid, not enough to allocate across six surfaces. |
| €4,000 | ~100 | Now the model has something to learn from. |
The number that matters isn’t your budget, it’s conversions per month — and specifically conversions of the event you’re optimising for. If you’re optimising for paid subscriptions and you get 8 a month, no amount of asset quality fixes that. If you’re optimising for trial starts and get 90, you’re in different territory. That trade-off — signal volume versus signal quality — is worth thinking through before you pick a campaign type at all.
Where Search genuinely wins for SaaS
Search has a structural advantage for software that most other channels can’t touch: people search for solutions to work problems by name. Someone typing “invoice tool for freelancers” or “alternative to [competitor]” has already done the hard part — they’ve admitted they need something. You’re not creating demand, you’re intercepting it.
That makes Search the right default when:
- Your budget is under roughly €1,500/month. You’ll get more from 20 well-chosen keywords than from an automation with 20 data points.
- Your category has real search volume. Check it before assuming — genuinely novel products sometimes have no query to buy.
- You need to learn what language converts. Search terms reports are the cheapest market research you’ll ever run, and the winning phrases usually belong on your homepage.
- You’re defending your brand name, or testing whether competitor terms are worth it.
The catch: high-intent search is expensive per click, and in competitive B2B categories a single click can cost more than a coffee. Low volume, high cost, high quality. Budget accordingly and judge it on cost per customer, not cost per click — the difference between those two is where CPM, CPC and CPA quietly mislead you.
Where PMax earns its place
PMax isn’t a trap, it’s a tool with a prerequisite. It becomes the better bet when:
- You’re consistently past ~50–100 conversions/month on the event you optimise for.
- You’ve exhausted search volume. Your Search campaigns are capped by impression share on the terms that matter, and more budget just raises CPCs.
- You have strong visual and video assets. PMax spends heavily on YouTube and Display; without decent creative you’re buying impressions nobody registers.
- You feed it good signals: a clean conversion action, value-based bidding if you can, and customer lists to exclude people who already pay you.
A reasonable sequence for a growing SaaS: run Search until it’s genuinely capped, add PMax as a second campaign with its own budget, exclude your existing customers, add your brand as a campaign negative so it can’t cannibalise cheap branded clicks, and compare on real revenue rather than platform-reported conversions.
The comparison, condensed
| Search | Performance Max | |
|---|---|---|
| You control | Keywords, match types, bids, ad copy | Budget, goal, assets, search themes |
| Inventory | Google Search (+ partners) | Search, YouTube, Display, Gmail, Discover, Maps |
| Intent | High — they typed it | Mixed — much of it interruption |
| Minimum viable data | Low | High (~50+ conv./month) |
| Fails by | Running out of volume | Spending on the wrong surface |
| Best first use | Your first €500 | Your second €5,000 |
Don’t run both on a small budget
The tempting move is to split: half Search, half PMax, see which wins. On a small budget this is the worst option. You halve the data feeding each, PMax will happily bid on your brand terms and take credit for clicks Search would have won cheaply, and you end up unable to tell which campaign caused anything.
Concentration beats diversification until you have volume to spare. Pick one, give it enough budget to produce a readable result, and let it run long enough that you’re reading a trend rather than noise. When you do compare them later, compare on the same footing — platform reports and your revenue tool answer different questions, and PMax’s wider attribution surface makes it look better than Search on platform numbers almost by construction.
Where Flowjat fits
Flowjat connects your Google Ads account alongside Meta, Apple Search Ads and your revenue data, so a Search campaign and a PMax campaign are judged by the same measure: money that actually arrived. You can see both campaigns’ real ROAS against your break-even line rather than two dashboards each grading their own homework — which is exactly the comparison this decision needs.
The Flowjat team
Building the ad copilot for builders