Keyword Planner Said These Arabic Keywords Had Zero Searches. One of Them Had Five Million.

Same account. Same country. Same language setting. Ninety seconds apart.

I want to explain what happened, because if you work Arabic markets there is a decent chance you have already killed a good idea because of it.

What I was doing

Researching Saudization. It's the set of rules that require private companies in Saudi Arabia to employ a certain percentage of Saudi nationals. Every private employer in the Kingdom has to deal with it, hundreds of professions have their own quota, the rules change every few months, and the fines start around 20,000 riyals. Real problem, lots of people affected, and nobody maintaining a decent resource for it.

So I did the obvious thing and threw eighteen Arabic seed keywords at Keyword Planner. Saudi Arabia, Arabic, last twelve months. Things like نسبة السعودة and حاسبة نطاقات and المهن الممنوعة على الأجانب في السعودية.

Nothing came back. Not small numbers, actually empty.

Normally that's where I'd stop. Zero volume, move on, don't get attached. But two things bothered me. I could see websites ranking for these terms, and more to the point the Saudi government had built an official Nitaqat calculator into Qiwa. Governments don't build calculators nobody uses.

So before closing the tab I tried the same terms as single words.

Five million.

The pairs are the part that still annoys me:

نسبة السعودة returned nothing. السعودة returned 50,000.

حاسبة نطاقات returned nothing. نطاقات returned 5,000.

One extra word on a term with proven demand, and the tool reports zero.

If you want to check I'm not making this up, it takes about two minutes. Keyword Planner, Saudi Arabia, Arabic, run نسبة السعودة and السعودة in the same session.

Use short seeds

Keyword Planner handles Arabic phrases badly. An English three-word phrase will usually give you something back, even if it's a vague range. The Arabic version quite often gives you nothing.

I don't know why. My guess is morphology, since Arabic writes the same concept a dozen ways depending on prefixes, the definite article, and which spelling of a letter someone happened to use, and matching all that across a multi-word string is obviously harder than matching one word. But that's a guess. Nobody outside Google can see how the matching actually works.

What I do know is the practical bit. A zero on an Arabic phrase usually means your seed was wrong, not that the demand is missing.

Seed short and let Google expand. Use Discover new keywords rather than Get search volume and forecasts. This matters more than it sounds like it should: volume lookup only reports on the exact strings you typed, so you can only ever find keywords you'd already thought of. Discovery grows outward and hands you the long tail you were trying to guess at.

Even in English the gap is big. On a different project, seeding phrases got me a handful of head terms. Same topic through Discover: 155 keywords, 60,800 monthly searches, including about a hundred device-specific queries I would never have guessed.

Three other things the export lies about

The Arabic thing is the one that caught me out. These three catch everyone.

Count the different numbers before you trust any of them.

Open the CSV and look at how many unique values are actually in the volume column. Here's one of mine:

distinct volumes: [0, 50, 500, 5000]

Four numbers. Across 155 keywords. That's bucketing. Google gives coarse ranges to accounts that aren't spending real money, so 500 means anywhere from 101 to 1,000, and 5,000 means anywhere from 1,001 to 10,000. Your market could be a tenth of what you think.

Two things fall out of that. Every "0% change" is meaningless, because real keywords don't sit at exactly zero for a year, it just means the number never crossed a bucket line. And every "+900%" or "−90%" is one bucket hop. Could be a keyword genuinely taking off. Could be 99 becoming 101. You can't tell.

Check the monthly columns too. In my account all twenty-four were blank, which means Google withheld the whole series and I had no trend data at all, whatever the percentages claimed.

Rows with identical stats are the same keyword.

From a 684-keyword export:

All three columns match exactly. That's Google telling you it's one keyword listed three times. There was a second cluster in the same file doing it with four rows.

I checked properly: of the 193 rows with complete data, 56 were sitting inside a duplicate cluster. My headline total was roughly 30% too big.

Group by the exact combination of volume, competition and bid. Anything that shows up more than once, count once. Arabic makes this worse because spelling variants multiply the rows.

The competition column is about ads.

This one costs people actual money and I still see experienced people get it wrong.

Competition and Competition (indexed value) measure how many advertisers are bidding. That's it. They say nothing about how hard something is to rank for.

Same dataset as above. Device-specific keywords: 105 terms, 40,800 searches a month, median competition index of 19. Looks wide open.

It wasn't. Nobody was bidding on those terms, but well-funded companies were sitting at position one organically on most of them. The export had no way of telling me that. Nothing in Keyword Planner does. You have to go look at the SERP, which takes twenty minutes and would have saved me a day.

What I do now

Single-word seeds, Discover mode, one country at a time, three-month range instead of the twelve-month default. Then before I read a single number I count the distinct values, dedupe on the volume/competition/bid combination, and ignore the competition column entirely.

Then I open the SERP, because that's the only thing that tells you what you're actually up against.

One thing I should be honest about

Five million for قوى is real, but most of it is people typing the name to reach the login page. Qiwa is where every Saudi employer goes to do their paperwork. You're not outranking a government portal for its own name, and at a two dollar bid nobody's making much money on that click either.

It's still a signal. Five million people a month needing to do something on a platform throws off a huge "how do I" tail, and that's where the actual opportunity sits. But a big single-word number tells you a topic is alive. It doesn't tell you there's a market you can reach. Confusing those two is a different mistake and I've made that one as well.

Anyway. If you've ever looked at a zero in Arabic and moved on, it might be worth going back and trying it with one word.