Rank tracking for the Uzbekistan region without distorted results
Checking by typing a query into your own browser shows your personalised results, not real ones. Here is how to track correctly and which metrics are more honest than positions.
Why manual checks lie
Results are personalised by search history, previously visited sites, location down to the district, interface language and device. The browser where you open your own site daily will show it higher than anyone else's.
Incognito helps partially: it drops history and cookies but not IP-based geolocation and not the interface language. In Tashkent that still means Tashkent results, not Samarkand or the country as a whole.
The consequence is the endless 'I see position three, the client sees nine' argument. Both see genuine results — just different ones. The only escape is a unified tracking method.
Setting the region correctly
In Google, region is set through query parameters: gl=uz for the country and hl=ru or hl=uz for the interface language. Finer geolocation uses the uule parameter with an encoded location — this is how professional trackers work.
In Yandex the region is the lr parameter with a numeric region identifier, while lang sets the language. Uzbek region identifiers are listed in Yandex's reference.
Fix the methodology once and never change it. Comparing a city-parameter measurement against a country-level one compares different quantities and produces imaginary trends.
Mobile and desktop results differ
Mobile has a different block composition: maps occupy more space, organic results start lower, some rich results do not appear. Position five on desktop and position five on mobile deliver different traffic volumes.
Given the mobile share of traffic in Uzbekistan, tracking desktop only means measuring the segment that does not deliver most visits.
Track both and store them separately. The divergence is diagnostic in itself: a mobile lag usually points at speed or mobile layout problems.
What average position in Search Console really means
Average position in the Performance report averages across every impression where the page appeared. It blends regions, devices and result types, so it does not — and should not — match a tracker's number.
Average position can worsen while traffic grows: if a page starts appearing for new broader queries at low positions, the average falls while clicks rise. That is positive movement easily mistaken for decline.
The correct use is segmentation. Filter by country, device and query group, and average position becomes meaningful for comparing periods within one slice.
What to track instead of positions
The number of URLs earning at least one click in the period. This is the most stable growth metric: it reflects expanding coverage and is immune to personalisation.
Total clicks and impressions by query cluster rather than by individual keyword. A cluster represents the whole topic, and its trend is more informative than swings on one query.
The share of impressions in the top three and top ten across your query set — available in Search Console by filtering on position, and it moves far more smoothly than any single query's rank.
Tracking frequency and reporting
Daily rank tracking is excessive for most projects: it generates noise and provokes reactions to random fluctuation. A weekly snapshot suffices, with monthly trends as the horizon for conclusions.
In client reports show the chain: clicks and impressions from Search Console, the number of working URLs, and enquiries from organic. A list of positions for twenty keywords with no traffic data is the least useful SEO report format there is.
Annotate the chart with release dates, algorithm updates and major edits. Without those markers any trend becomes inexplicable six months later and correlations are lost.