Google Ads is the only channel where someone tells you exactly what they want before you spend anything. That makes it the highest-intent traffic available, and the place where a badly structured account burns budget the fastest.
What I audit first
- Conversion actions: what is counted, what is double-counted, what is counted that should not be
- Whether “conversions” include form views, phone clicks or anything else that is not a real outcome
- Account and campaign structure against actual search intent
- Search terms report: where the money is going, which is rarely where the keywords suggest
- Negative keyword coverage, including Arabic and transliterated variants
- Match type strategy and how much broad match is being trusted
- Bid strategy versus available conversion volume
- Ad copy and landing page alignment
- Shopping feed quality and disapprovals
- Performance Max asset groups and whether brand traffic is being cannibalised
- Geographic and language targeting, including the presence-versus-interest setting
- Budget pacing and where it is being capped
The mistakes that cost the most
Conversions that are not conversions
Counting a WhatsApp click or a contact page view as a conversion teaches Smart Bidding to buy the wrong traffic. This one mistake invalidates everything downstream.
Untranslated negatives
Arabic search terms accumulate waste quietly because nobody reads the search terms report in Arabic. On regional accounts this is frequently the single largest leak.
Feed neglect
Shopping performance is mostly a feed problem. Titles, attributes and disapprovals move results more than bid adjustments do.
PMax eating brand
Performance Max will happily take credit for branded searches you were getting free. Excluding brand is usually the right call, and it usually makes the reported ROAS look worse and the business better.
Arabic and regional specifics
Google Ads in Egypt, Saudi Arabia and the UAE is not the same account you would build for a UK business:
- Users search in Arabic, in English, and in transliterated Arabic, often within the same session
- Language targeting does not work the way most people assume; it targets interface language, not query language
- “Presence” versus “presence or interest” location targeting matters enormously in expat-heavy markets
- Call extensions and WhatsApp routing convert differently than form fills across the three markets
- Ramadan shifts both auction prices and the hours when conversions happen
Frequently asked
Is Performance Max worth it?
For e-commerce with a clean feed and real conversion volume, often yes. For lead generation with weak conversion signals, usually not. It will optimise toward whatever you told it was a conversion, and if that definition is loose it will find you a great deal of worthless traffic very efficiently.
Should we use broad match?
Broad match with solid conversion data and tight negatives can work well. Broad match with a loose conversion definition is the fastest way to waste a budget. The conversion setup decides the answer, not the match type.
How much should we spend?
Work backwards from what a customer is worth. If you can afford $40 per acquisition and your conversion rate is 3%, you need roughly $1.20 per click to break even, and enough monthly volume to generate 30+ conversions, or the bidding algorithm never has enough to learn from.
Can you fix an account without rebuilding it?
Usually. Rebuilding loses learning history, which is a real cost. I rebuild when the structure makes correct measurement impossible; otherwise I fix conversions, negatives and structure in place.