Five common mistakes in Yandex Direct campaigns targeting Tashkent
The same five mistakes turn up in almost every account that arrives for audit. All are visible within half an hour and all cost real money.
Mistake 1: wrong geotargeting and expanded regional reach
The most expensive error is leaving expanded geographic targeting on by default. It shows ads to people outside the chosen region whenever their query contains your city name, so a Tashkent campaign collects clicks from other countries.
The reverse also happens: selecting all of Uzbekistan for a business that operates only in Tashkent. Budget diffuses across regions with different purchasing power and a different query language, and metrics degrade evenly and invisibly.
The correct approach is Tashkent and the regions in separate campaigns with separate budgets and their own ads. Then you can see which geography pays back and scale it deliberately.
Mistake 2: autotargeting left unsupervised
Autotargeting picks queries on its own, and at launch, without accumulated data, it drifts into broad adjacent themes. In a narrow niche that means a visible share of budget spent on queries you would never have added manually.
Check this in the search terms report, not by group name. Weekly pruning during the first months is the only way to keep autotargeting inside useful boundaries.
The practical rule: at launch either disable autotargeting or run it as a separate campaign with a capped budget. Mixing it with manual keywords in one campaign destroys your ability to tell what worked.
Mistake 3: running the ad network without pruning placements
The network delivers cheap reach, which is exactly why it gets launched first. The problem is that without regular work on the placements report a significant share of impressions goes to mobile apps and sites full of accidental clicks.
The symptom is immediate: high CTR with zero conversions and very short time on site. That is almost always junk placements rather than a landing page problem.
The working cadence is reviewing placements weekly for the first month and fortnightly after, blocking anything that spends without producing a single qualified session. It is routine that cannot be fully delegated to automation.
Mistake 4: single-language ads and mismatched audiences
A Tashkent campaign built only in Russian loses part of the demand; a campaign mechanically translated into Uzbek without rebuilding the keyword set loses quality. Both appear equally often.
Uzbek-language queries must be researched separately rather than translated from a Russian list: phrasing, length and volumes all differ, and Latin and Cyrillic produce different spellings of the same word.
One more detail is the landing page. An Uzbek-language ad pointing at a Russian-only page wastes the whole effort: the visitor lands somewhere they did not expect and leaves, after you have already paid for the click.
Mistake 5: conversions unconfigured or counted wrongly
The most common situation in un-audited accounts: analytics goals are either missing or pointed at a thank-you page that no longer exists. The campaign optimises on clicks because there is nothing else to optimise on.
The second layer is local: a meaningful share of enquiries arrive by phone, Telegram or Instagram direct. If those channels are not counted, the campaign looks unprofitable and gets switched off despite delivering leads past the counter.
The minimum working set: correct goals for form submissions and clicks on contact buttons, UTM-tagged links to messengers, and a 'how did you hear about us' question in the sales script.
Auditing your own account in half an hour
The checking order: region and expanded geotargeting settings, the search terms report for the last month, the network placements report, alignment between ad language and landing page, and the list of active goals with the date of the last conversion for each.
If even two of those are failing, optimising bids is pointless: you will simply be steering a budget that goes to the wrong place with greater precision.
The fixing order is equally fixed: geography and negatives first, then placements, then language and landing pages, and only then strategies and bids. Doing it in reverse is the most common reason people conclude that optimisation achieved nothing.