Running Yandex Direct and Google Ads campaigns for Uzbekistan
The two search engines split this market unevenly, and launching both the same way is the most common mistake. Here is how to build them separately.
Who searches where
Google holds the overwhelming share of search traffic in Uzbekistan, a direct consequence of Android phones shipping with it as the default. Yandex keeps a meaningful but smaller core: mostly Russian-speaking Tashkent users, plus traffic from Yandex Maps, Yandex Go and Yandex Browser.
The practical conclusion: Google Ads is the volume channel, Yandex Direct the supplementary one — often with cheaper clicks simply because its auction is thinner. In niches where competitors ignore Direct, cost per enquiry can be noticeably lower.
The mistake is treating Direct as a second Google and cloning campaigns wholesale. The audience differs by language, device and time of day, so ads and landing pages should differ too.
Building a bilingual keyword set
Search here is bilingual, and not in a fifty-fifty way. Commercial queries in Tashkent skew Russian or mixed. In the regions the Uzbek-language share is higher and phrasing is shorter and more colloquial.
Build the sets separately: Google Ads Keyword Planner with Uzbekistan as the region for one, Yandex Wordstat with the same region for the other. There will be overlap, but volumes and phrasing diverge sharply.
Account for both Latin and Cyrillic spellings of Uzbek words — people type both. If your account treats them as separate keywords, you need both or half the demand never sees your ad.
Campaign types and where to start
A starting set in Google Ads: a search campaign on high-intent commercial queries, a separate brand campaign, and display remarketing. Add Performance Max later, once conversion data exists — without it the algorithm has nothing to learn from and spends on cheap irrelevant impressions.
In Yandex Direct: search on the commercial core plus a separate display network campaign. The Yandex network delivers cheap reach in Uzbekistan but demands aggressive placement pruning, or a visible share of budget drains into mobile apps and junk sites.
Always split search and network into separate campaigns. A combined budget almost always drifts into the network, where clicks are cheaper and conversions worse.
Negative keywords and query hygiene
The first week of any campaign is search-terms report work, not bid work. A baseline negative list here: free, DIY, second-hand, vacancy, job, download, employee reviews, plus the names of cities and countries you do not serve.
A local specific: queries naming neighbouring countries and cities — Kazakhstan, Almaty, Bishkek, Dushanbe. Geo-targeting filters by location, but a Tashkent user searching for delivery to Almaty still triggers impressions unless the word is negated.
Maintain negative lists at account level, not per campaign, or every new campaign starts collecting the same junk from scratch.
Bid adjustments for the local audience
Devices: mobile share is very high in Uzbekistan, but mobile conversion depends on what you sell. Call-driven services usually perform better on mobile; long-form B2B enquiries worse. Set the adjustment from your data, not by default.
Time of day: activity peaks in the evening, and weekends deliver leads as well as weekdays in many niches. Turning weekends off only makes sense if nobody is there to answer.
Geography: separate Tashkent and the regions into different campaigns rather than using adjustments. Click prices, purchasing power and ad language all differ, and one budget averages away what should not be averaged.
Budgets and what counts as a result
Market reference points: a test launch for one service in Tashkent runs roughly 3M to 6M UZS per month in clicks. Competitive niches — real estate, medicine, legal, automotive — start higher, from 10M UZS upward.
Below a certain threshold the auction simply yields no statistics: 20–30 clicks a month supports no conclusion at all. With a limited budget, narrow the scope — one service, one city, a tight keyword set — rather than spreading it thin.
Measure cost per enquiry and the share of enquiries that became deals, not clicks and CTR. The typical picture in unmanaged accounts: the campaign with the best CTR delivers the worst-quality leads.