Google Ads Performance Max vs Search campaigns in the UAE: budget split and when each wins

If your main conversion in Dubai is a WhatsApp tap, keep the majority of the budget in Search and give Performance Max a minority share, or none at all. If you run real e-commerce with tracked purchases and your monthly Google spend clears roughly USD 3,000 (around AED 11,000), the ratio flips and Performance Max can carry most of the money. That one difference, what you count as a conversion, decides more than any best-practice ratio you will read on an American blog.

The reason it matters here specifically: Performance Max is a bidding algorithm that learns from your conversion data. Feed it 300 chat taps a month where one turns into a paid booking, and it will optimise for cheap chat taps. We watch that happen in UAE accounts constantly. Google does not know the difference between a lead and a tap, and in this market the gap between them is enormous. If you have not yet decided which campaign types belong in the account at all, start with our overview of the five Google Ads campaign types and come back to this fork afterwards.

How to split the budget between Performance Max and Search

The industry-standard starting ratios are published widely, and they hold up reasonably well in the Gulf, with one correction for conversion quality.

Business model	Search share	Performance Max share	The UAE correction
E-commerce, tracked purchases with revenue values	30-40%	60-70%	Works as published. Needs roughly AED 11,000+ monthly spend before PMax leaves learning
Service lead generation (clinics, fit-out, studios, legal, beauty)	70-80%	20-30%	Only give PMax the 20-30% if you can pass back a payment or confirmed booking. Chat taps are not enough
B2B with long sales cycles	70%	30%	Keep PMax away from high-intent commercial keywords you already win in Search

Service lead generation covers most of the businesses we manage in Dubai, and that bottom-heavy Search split is not conservatism. Search gives you keyword-level control, match-type control, and a search terms report you can act on the same afternoon. In a market where a single emirate is your whole addressable geography, that control is worth more than reach.

Treat the table as a starting allocation for the first 60 days, then re-cut it using the checks further down this page.

Where Performance Max eats your Search traffic

Optmyzr studied query overlap across accounts running both campaign types, and the numbers are blunt. 91% of accounts had queries appearing in both Performance Max and Search. In 67% of accounts, Performance Max overlapped with Search on at least one query. When both campaigns competed for the same query, Search converted better 84% of the time. Accounts with that overlap paid 34% more per conversion than accounts without it.

Google's own documentation says exact match keywords should take priority over Performance Max when the query matches. In practice, Adalysis has recorded overlap even on exact match. So the rule exists, and it leaks.

What this looks like in a Dubai account: you have been bidding on your service keyword for two years at a known cost per lead. You launch Performance Max. Total conversions go up 20%, cost per conversion goes up 34%, and when you look closer, it is quietly serving your brand name and your highest-intent commercial queries, which Search was already winning at half the price. The new campaign reports those conversions as its own.

Three practical defences:

Use account-level brand exclusions to keep Performance Max off your own brand name. These exist and they work, which is different from campaign-level negative keywords: as of mid-2026 Google has still not shipped full negative keyword lists inside Performance Max in the way Search has them, and claims that it has are worth verifying in your own interface before you plan around them.

Tighten your Search structure so exact match carries your money keywords. If your account still runs mostly broad match, Performance Max will outbid you on your own terms. Our breakdown of Google Ads match types for Dubai accounts covers how to rebuild that layer.

Keep the two campaigns on different jobs. Search takes the commercial queries you can name. Performance Max takes discovery, Shopping inventory, and the long tail you cannot predict.

Pull your overlap report this week before you argue about ratios, because the argument is usually about money you are already losing.

What the 2026 reporting changes actually let you check

For three years the main complaint about Performance Max was that you could not see inside it. That changed in 2026.

Since January 2026, Google Ads API v23 reports Performance Max performance by channel: Search, YouTube, Display, and Maps. In June 2026, the Google Ads Developer Blog added search terms reporting for Performance Max. Asset-level metrics came with the same wave, so you can see which headline and which image are doing the work.

This matters for the budget decision in a concrete way. Before 2026 you had to assume how much of your Performance Max spend was going to Display and YouTube rather than Search. Now you can read it. In accounts where it turns out to be spending most of its budget on the Search channel, serving queries your Search campaigns already cover, the case for putting 60% of the budget there collapses.

One caveat to set expectations: channel reporting tells you where the spend went, it does not give you a switch to turn a channel off. Being able to see Display spend and being able to stop it are two different things, and the second one still does not exist inside Performance Max. Plan your split assuming you cannot exclude channels.

Export the channel report and the search terms report for the last 90 days, then decide the ratio with those two files open.

Minimum budget and how long Performance Max takes to learn

Smart Bidding needs volume. Google's documentation points at 30-50 conversions per month for a campaign to optimise, and 150 or more for reliable performance. Google also recommends a minimum daily budget of three times your target cost per acquisition. At a target CPA of AED 200, that is AED 600 a day before the campaign has room to work.

Learning time scales with budget. GROAS reports these ranges:

Daily budget	Typical exit from learning
USD 300+ (around AED 1,100)	1-3 weeks
USD 50-100 (around AED 185-370)	2-4 weeks
Under USD 50	5-6 weeks

Adnabu's position, which matches what we see: below roughly USD 3,000 a month, Performance Max rarely reaches the 30-50 conversions it needs to optimise properly. It will spend the money. It will just spend it learning.

There is a market-size factor on top of this that nobody writing from London or Austin has to think about. The UAE is one or two emirates of addressable audience. The same AED 20,000 a month buys far less signal here than it does across a country of 330 million people, because there is simply less inventory and fewer searches to learn from. So Google's conversion minimums take longer to accumulate in a UAE-only campaign than the published learning ranges imply. I am deliberately not putting a week count on that, because I have not seen a defensible dataset for it and neither has anyone else publishing on this. Treat the GROAS ranges as the optimistic end of your planning.

If your total Google budget is under AED 10,000 a month and you sell a service, put all of it in Search and revisit Performance Max when the budget doubles. Our PPC management service exists partly because this call gets made wrong so often in small accounts.

What Dubai and the UAE change about this decision

Four local factors change the arithmetic, and the first one is the big one.

The WhatsApp conversion problem. For most service businesses in Dubai, the primary conversion is a jump into WhatsApp. That counter inflates badly. A click on the chat button on the Google results page registers as a conversion, and across the accounts we manage, only one in several dozen of those taps reaches a payment. A Performance Max campaign optimising toward that signal is optimising toward noise, and it will find you more noise very efficiently. Either give it a deeper conversion action, a payment or a confirmed booking passed back through offline conversion import, or keep that campaign out of Performance Max entirely. I would rather run a smaller Search campaign on a clean signal than a large automated campaign trained on taps.

Attribution delay. WhatsApp conversions land in the report later than the click that caused them, sometimes days later. A seven-day read on a new Performance Max campaign in a UAE lead-gen account is close to worthless, because the conversions that justify the spend have not arrived yet. Judge on a 28-day window or longer. We have seen the same campaign look like a failure on Monday and a winner three weeks later with no changes made.

Two languages. Arabic and English traffic behave differently and convert differently. In Search you can separate them into campaigns, write proper Arabic ad copy, and read the two sets of numbers apart. Performance Max blends them into one asset-driven system where you cannot easily tell which language is carrying the result. If bilingual performance is something you need to manage rather than hope for, that is another point for Search. You will see confident claims that Arabic clicks are 15-25% cheaper; I have not found a source solid enough to repeat, so test it in your own account.

Small geography, fast saturation. With a tight location target, Performance Max runs out of genuinely new audience quickly and starts spending its extra reach on Display and YouTube impressions that do not convert. The channel report from January 2026 finally shows you when that is happening.

Before you change the split, confirm which conversion action each campaign is actually optimising toward. That single screen explains most bad Performance Max results in this market.

Six checks to make the call on your own data

Run these in order. They take an afternoon.

1. Count conversions at the money level. How many payments or confirmed bookings did Google produce last month, not how many chat taps? If the answer is under 30, Performance Max has nothing to learn from.
2. Measure query overlap. Pull the Performance Max search terms report and your Search campaign terms for the same 90 days. Quantify the shared queries and the cost on each side. Optmyzr's 34% penalty is what you are testing for.
3. Check brand leakage. How much Performance Max spend sits on your own company name? That traffic was already yours.
4. Read the channel split. If most Performance Max spend is on the Search channel, you are paying automation rates for traffic you could buy with keywords.
5. Test the budget floor. Is your daily Performance Max budget at least three times your target CPA? If not, raise the budget or close the campaign. Half-funded automation underperforms a well-built Search campaign every time.
6. Re-run every report on a 28-day window. Any seven-day verdict on a UAE lead-gen account is noise.

The decision that falls out is usually simple. Tracked purchases with revenue values and budget above AED 11,000 a month: Performance Max takes the larger share. Service leads, WhatsApp as the goal, no offline conversion import: Search takes 70-80% and Performance Max stays small or waits. B2B with named target accounts: Search carries it, Performance Max covers remarketing and the queries you never thought of.

FAQ

Does Performance Max steal traffic from Search campaigns, and how do I stop it? Yes, in most accounts. Optmyzr found query overlap in 91% of accounts studied, and accounts with overlap paid 34% more per conversion. Google says exact match keywords should take priority over Performance Max, but Adalysis has documented overlap even on exact match. Stop it with account-level brand exclusions, a Search structure built on exact match for your money keywords, and a clear division of jobs between the two campaigns.

What is the minimum budget for Performance Max to leave the learning phase? Google recommends a daily budget of at least three times your target CPA, and Smart Bidding wants 30-50 conversions a month to optimise, 150 or more to be reliable. Adnabu's benchmark is that under roughly USD 3,000 a month, Performance Max rarely hits those conversion volumes.

How long does Performance Max take to learn in 2026? GROAS reports 1-3 weeks above USD 300 a day, 2-4 weeks at USD 50-100 a day, and 5-6 weeks below USD 50. Expect the slower end in a UAE-only campaign, because a one or two emirate target produces less signal per dirham than a large market.

Can you see search terms in Performance Max now? Yes. The Google Ads Developer Blog shipped search terms reporting for Performance Max in June 2026, alongside channel-level reporting (Search, YouTube, Display, Maps) that arrived with API v23 in January 2026, plus asset-level metrics. You can see where the money went, but you still cannot switch a channel off.

Should I run Performance Max and Search on the same keywords at the same time? No. Where both compete for one query, Search converted better in 84% of Optmyzr's cases. Give Search the commercial queries you can name and Performance Max the discovery and long tail.

How do I split the budget for my specific goal? E-commerce with tracked revenue: 60-70% Performance Max. Service lead generation: 70-80% Search. B2B: 70% Search. Then correct for conversion quality, because a split built on WhatsApp taps rather than payments will be wrong regardless of the ratio. If you want that read done on your account, see how we report results for clients and send us the account for a free audit.

Written by Artur Gall, CEO and founder of SkyLight Marketing, Dubai.