Commercial SEO audit for a Tashkent site: a step-by-step process
An audit without prioritisation is a list of 200 findings nobody ships. Here is the sequence of work and the report format that actually reaches production.
Preparation: access and a baseline snapshot
Four accesses before you start: Google Search Console, Yandex.Webmaster, the analytics platform and, ideally, server logs. Logs are the most underused source — they show where Googlebot actually spends crawl budget, not where you assume it does.
Record the baseline: indexed page count in each engine, organic traffic by month for twelve months, and the number of URLs earning at least one click. Without that snapshot you cannot prove impact six months later.
Separately export the commercially significant URLs — categories, key product pages, service pages. The audit must answer 'why do these forty pages not rank', not 'what errors exist on the site in general'.
Step 1: a full crawl
Run Screaming Frog or Netpeak Spider with JavaScript rendering enabled if the site is React or Vue. Without rendering the crawler sees an empty shell and you get a false 'the site has no content' picture.
What surfaces here: non-200 response codes (301, 302, 404, 5xx), redirect chains longer than one hop, duplicate titles and h1s, empty or short meta descriptions, and pages deeper than four clicks from the homepage.
Check separately that canonicals are self-referencing on indexable pages and point to the primary version on utility pages. Broken canonicals pointing at 404s are endemic in bespoke CMSs and drop pages out of the index entirely.
Step 2: reconcile the index with reality
Compare three numbers: URLs found by the crawler, URLs listed in sitemap.xml, and what the Page Indexing report in Search Console reports. The gaps are where the findings live.
A typical picture in Uzbek e-commerce: 12,000 URLs in the sitemap, 3,000 indexed, and the crawler finding 40,000 because of filters and sorts. Crawl budget is being burned on filter combinations while product pages queue for months.
The standard remedy: block parametric combinations via robots.txt and meta robots noindex, follow; add Clean-param for Yandex; keep only canonical indexable URLs with accurate lastmod values in the sitemap.
Step 3: structure and semantics
Check whether the site structure covers actual demand. A common Tashkent retail problem: categories follow the supplier's internal logic rather than how people search — there is 'Climate equipment' but no dedicated landing pages for 'inverter air conditioners' or 'split systems 12'.
Pull frequencies for your categories in Wordstat with a regional filter and in the Search Console Performance report. Demand with no dedicated landing page is a missed entry point, and such findings always outrank technical trivia in a report.
Check cannibalisation: filter Search Console by a specific query and see how many URLs appear for it. Two or more of your own URLs on one commercial query means consolidation or rework.
Step 4: commercial factors
Text is not everything for a commercial site. Verify prices in soum on product pages, payment and delivery options, legal entity details, clickable tel: phone numbers, office address, a map and a returns policy.
Each of those blocks answers a user question and simultaneously gives the engine a corroborating signal. Mark up the organisation and products in JSON-LD: schema.org/Organization with address and telephone, schema.org/Product with offers, price and priceCurrency in UZS.
Validate the markup in the Rich Results Test and the Enhancements reports in Search Console. An error in a required field such as price makes the markup useless, not partially useful.
Step 5: a report that gets implemented
The format that works: a table with problem, exact location (URL list or template), impact, effort estimate, and owner — developer, content or marketing. Sort by impact-to-effort ratio, not by audit chapter.
The top five items must be shippable in a single sprint. If item one is 'rework the catalogue structure', the report dies in review. Break large items into stages with interim deliverables.
Schedule a re-measurement in four to six weeks against the same baseline metrics. An audit without a follow-up measurement is a document, not a tool.