Your first month of Google Ads in Uzbekistan, without illusions

The screenshot people send me most often on Telegram before a first call looks like this: the campaign has been running for ten days, over a thousand clicks, about a hundred dollars spent, two leads. Sometimes zero. And the question: "Does Google even work in Uzbekistan?"

It does. But the first month of Google Ads is not about sales. It is the month when the system learns and you find out which searches bring people who are ready to pay. If you expect it to pay for itself, the campaign usually gets switched off right when it starts to figure something out.

Below is what that month looks like in our studio. Week by week, including the mistakes we made ourselves.

Week zero: settings you can't redo later

A couple of things in Google Ads are set once. The account currency cannot be changed at all; you would have to open a new account. We pick US dollars: clients think about budgets in dollars anyway, and so'm with all its zeros just makes reports harder to read. Time zone goes to Tashkent (GMT+5). Otherwise the schedule drifts and your "business hours only" ads start running at night.

Payment goes through an international Visa or Mastercard. Google does not accept Uzcard or Humo. Sounds obvious, yet we have more than once seen a launch slip by a couple of days because accounting only had a corporate Uzcard.

One more rule before any ads go live: the account is registered to the client, and we are added as managers. If you ever move to another contractor, the campaign history and the data you paid for stay with you, not with us.

Location and languages

Location settings have an option called "Presence or interest". By default Google also shows your ads to people who are merely interested in Tashkent while sitting in another country. For a salon in Yunusabad that is wasted money. We set it to "Presence".

Languages are messier here. The same person searches in Russian, in Latin-script Uzbek and sometimes in Cyrillic Uzbek, while their phone interface is in English. We almost always leave "all languages" on and shape the audience through keywords. A few times we tried limiting a campaign to Russian: impressions dropped noticeably, and we saw no benefit.

A Telegram click is a conversion, and that's fine

Most customers in Uzbekistan don't call and don't fill in forms. They tap "Message us on Telegram". If that click isn't counted as a conversion, Google never learns that the ads work and optimises for something else. Clicks for the sake of clicks, for example.

The minimum we set up before the first impression:

The catch is that a Telegram click is not yet a lead. Someone could open the chat and close it. So once a week we compare the number of "conversions" in Google Ads with the number of conversations the manager actually received. A good share of clicks turn into a conversation, but nowhere near all of them, and that ratio is your real cost per lead. Mirodil covered how to tie a click on the site to a specific chat in his post on tracking leads that come through Telegram.

Search or Performance Max

Performance Max looks tempting: upload images, copy and video, set a budget, and Google decides where to show them. In large markets, where an advertiser gets thousands of conversions a month, that can make sense. A small Tashkent business has few conversions, while Google's inventory in Uzbekistan is huge and far from all of it is decent.

What we saw in PMax at launch: lots of cheap impressions in mobile apps and games, clicks followed by an instant bounce, "conversions" from accidental taps on a button. The report looks great; the manager on Telegram hears nothing.

That is why our first month is nearly always Search. The person typed "курсы английского юнусабад" or "mebel tashish toshkent" themselves; they are already looking, and we pay for exactly that intent. PMax and video come later, once there are at least 30–50 honest conversions and it's clear who your customer is.

The exception is online shops with a product feed. Shopping campaigns make sense earlier there, but that is a separate story.

What a click costs

Cost per click in Uzbekistan is still lower than in Russia or Kazakhstan and many times lower than in the US. But "cheap" quickly turns into "expensive" if they're the wrong clicks. Below is the order of magnitude we have seen in our search campaigns over the past year. It is not a price list or a promise, just a very rough guide. Your niche may differ.

Two caveats. First, a lot depends on the query language. Russian queries in Tashkent are often pricier than the same queries in Latin-script Uzbek, simply because fewer advertisers bother writing ads in Uzbek. Second, when one or two big players with large budgets enter the auction, as happens with developers and clinics, CPC can multiply within a week.

How much should the first month cost? Our answer: enough to get at least 300–500 clicks on relevant searches. For most niches that is $150–400 for clicks alone, not counting our work. Less than that, and a month later you won't have enough data to decide anything.

Negative keywords in three scripts

This is where the most money leaks in Uzbekistan. The negative list has to be kept in Russian, Latin Uzbek and Cyrillic Uzbek at once, and it has to cover how people actually type: "o'quv", "o‘quv", "oquv", "ўқув".

The base list we start almost every service campaign with:

Be careful with "ish" (work). We never add it as a single word: it shows up inside perfectly commercial queries like "ta’mirlash ishi narxi" (price of repair work).

The main tool of the first month is the search terms report. For the first two weeks I open it every day. It surfaces things you'd never think of in advance: competitor names, people looking for jobs at your own company, "how to do it myself" searches.

The month, week by week

A word on the learning period. When you change the bidding strategy, Google shows "Learning" and behaves erratically for about a week. Every big change (doubling the budget, a new strategy, reshuffling ad groups) restarts it. The most expensive beginner mistake, and ours in early projects too, is poking the campaign every two days because "nothing is happening". Now we change strategy no more than once every week or two and log every change. Without that log, a month later nobody remembers what affected what.

What I would do differently

In short: trust Google's reports less and the manager's chat history more.

For a long time we treated a falling cost per conversion as success. Then we noticed that in one campaign "conversions" were getting cheaper because the share of accidental taps on the Telegram button on mobile was growing. Conversations didn't increase at all. Since then our main number is how many real people wrote to us and how many of them got to payment. It's counted by hand, in a spreadsheet, once a week. Boring, but we haven't found anything more reliable.

Second, the landing page matters more than the ad. We have seen campaigns start bringing noticeably more enquiries after one screen of the site was reworked, without a single change inside Ads. Murod wrote about this separately: why sending ads to your homepage burns the budget.

Third, the first month has to be sold honestly. Not "you'll get a stream of leads", but "in a month we'll know what a lead costs in your niche and whether Google pays off at all". Less exciting, but after that conversation nobody switches the campaign off on day ten.

If you are choosing between Google and Yandex, Mirodil has a breakdown of Yandex Direct for Uzbekistan. And the way we run ads end to end is described on the Google & Yandex ads page.