A brief template for commissioning Uzbek-language SEO copy
Bad copy is almost always the result of a bad brief. Here is the brief structure point by point: fill it once, then copy it per page.
The header block
The first five fields: page URL (or 'new'), page type, language, length in characters excluding spaces, deadline. State the language explicitly as 'Uzbek, Latin script': without it a third of writers deliver Cyrillic and another third mix both scripts in one paragraph.
The sixth field is the page's target action, in one sentence: request a measurement visit, message on Telegram, call. Without it the writer produces an encyclopedia entry rather than copy that leads somewhere.
Give length as a range, not an exact figure: 3,500–4,500 characters. An exact figure pushes the writer to pad.
The keyword block
A three-column table: query, volume, whether inclusion is mandatory. Mark the primary query as 'exact match in H1 and the opening paragraph'; the rest as 'any inflected form, once each'.
List forbidden constructions separately: unnatural strings like 'custom furniture Tashkent price buy'. If a query will not fit a natural sentence, let the writer rephrase it rather than break the prose.
For Uzbek copy, supply queries in the form users actually type. Collect them from Yandex Wordstat filtered to the Uzbekistan region and from Google autocomplete, not by translating the Russian list.
Text structure
Specify the heading plan rather than leaving it to the writer. Write out the H1, four to six H2s, and one sentence describing each block. It is the only reliable way to get copy that answers the query instead of paraphrasing a competitor's homepage.
Mark which blocks should not be prose: a pricing table, a document list, a calculation form. The writer prepares the table content rather than describing it in sentences.
Add a rule for the opening paragraph: under 400 characters, answering the user's question directly with no preamble. That paragraph is what gets read in search results and in AI answers.
Facts and sources
Supply the facts or the writer will invent them. At minimum: real prices or a range in UZS, timelines, what the service includes, three to five common customer questions, and what you do not do.
State the rule that every number in the copy comes from this block and no new numbers may be introduced. Invented claims like 'over 500 happy clients' are the most common fix at acceptance.
Attach two or three competitor links, noting what to look at: structure and depth of coverage, but not phrasing.
Specifics of Uzbek copy
Set a terminology rule: some industry words are used in their Russian form in everyday Uzbek, and forcing a translation makes the text read as artificial. Provide a short glossary of what stays and what gets translated.
Ban machine translation of the Russian text. It is recognisable by calqued word order and constructions Uzbek does not use. If you need both languages, write the Uzbek version from the meaning, not from the Russian sentences.
Require a single script across the whole page, including table captions, buttons and breadcrumbs. Mixing Cyrillic and Latin within one page reads as carelessness and makes it harder for search engines to match the page to a query.
Acceptance criteria
Express them numerically so the discussion is not a matter of taste: uniqueness above 90 percent, no single word exceeding five percent density, all mandatory queries present, no invented figures, heading plan followed.
Add a revisions clause: one round of edits is included, a second as well, a third is billed separately. It disciplines both sides and ends open-ended rewriting.
The final item is a native-speaker proofread. For Uzbek copy this is not a luxury: even a strong bilingual writer makes suffix errors only a native speaker catches.