Turkish Keyword Research: Suffixes, Inflections, Letters

Turkish keyword research breaks the assumptions of an English-trained workflow. A tool may show "ayakkabı" (shoe), "ayakkabıyı" (the shoe, as an object), "ayakkabılar" (shoes) and "ayakkabi" (the same word typed without the dotless i) as four separate rows, even though one buyer intent sits behind them. If you are an agency or brand preparing to publish in Turkey, this guide shows what suffixes do to a keyword list, why the letters ı, İ, ö, ü, ç, ş and ğ cause trouble in spreadsheets and URLs, and how to map the many forms of one query to the right page without splitting your content.

The examples, tables and numbers below are hypothetical and contain no real volume, ranking or traffic data. Where the behaviour of a search engine or a keyword tool is not documented, the guide says so and suggests a small test you can run on your own data instead of asserting a rule. You do not need to speak Turkish to follow the method, but you will need a native-level editor for the final checks.

Why Turkish Keyword Research Works Differently

One root, many surface forms

Turkish is an agglutinative language: words are built by stacking suffixes onto a root. The root "kitap" (book) becomes "kitabı" (the book, as an object), "kitapta" (in the book), "kitaplarımız" (our books) and "kitapçı" (bookseller). English expresses the same ideas with separate words such as "in", "the" and "our", which a keyword tool treats as a phrase. In Turkish the grammar sits inside the word, so each combination is a different string that people can type into a search box.

For a foreign team this has a practical consequence. A seed list of ten English-style terms can easily turn into several dozen Turkish strings, and the tool you use may or may not group them. Treat the list as a map of one idea with many spellings, not as a set of independent targets.

Tools do not all treat suffixes the same way

Some keyword tools keep every inflected form as its own row, some merge forms, and some apply stemming (reducing a word to a root) only in certain languages. This guide does not claim how any specific tool counts Turkish forms, because that is tool-specific and can change. Run a simple test: type three inflected forms of the same root into your tool one at a time and compare what comes back. If the results differ, you will be doing the grouping yourself; if they are identical, read the tool's documentation to learn which form it reports.

Third-party volume, difficulty and traffic figures are tool-specific estimates, not Google values. Two tools can disagree for the same query, and the gap is often wider for inflected Turkish forms.

Where the English habit misleads you

In English research, singular and plural are usually enough. In Turkish, plural, possessive, case and question suffixes can all appear on one word, so the number of plausible forms is much larger. A second problem is software: functions written for English, such as lowercase conversion, sorting and matching, can mishandle the Turkish letters i and ı. Knowing both problems before you start saves a clean-up pass later.

Suffixes and Case Endings: What Each Form Signals

Plural, possessive and the bare form

The bare form of a noun ("backlink", "guest post") usually matches a searcher looking at the topic in general. A plural ("backlinkler", "tanıtım yazıları") often signals comparison or browsing, because the person expects a set of options. A possessive suffix ("sitemin hızı", meaning "my site's speed") shows the searcher describing their own situation, which is closer to a problem-solving intent. Treat these readings as a first hypothesis only. Confirm the real intent by opening the results page and by checking your own Search Console data.

Case endings: object, direction, location, origin

Turkish marks grammatical roles with case endings: -i (object), -e (direction), -de (location) and -den (origin). "Backlinki nereden alırım" ("where do I get the backlink") and "backlink satın alınır mı" ("can a backlink be bought") share a topic but may sit at different stages of a decision. Record the stage in a separate column next to the keyword so that the intent is not lost when the list is sorted.

Vowel harmony and consonant softening

Two rules decide which suffix form is correct. Vowel harmony chooses between forms such as -lar and -ler or -ı, -i, -u and -ü depending on the last vowel of the root. Consonant softening changes the final consonant of some roots when a vowel-initial suffix follows: kitap becomes kitabı, ağaç becomes ağacı, renk becomes rengi, ayak becomes ayağı. If someone generates variants by simply gluing a suffix to a root, the output will contain non-words such as "kitapı". They add rows without adding searches. Have a native editor review a sample of any generated list before you rely on it.

Question Words, Compound Terms and Brand Names

The question particle and common question patterns

The Turkish question particle is written as a separate word (mi, mı, mu, mü), for example "sitemap gerekli mi". Searchers also write it attached ("gerekli mi" becomes "gereklimi") or leave it out entirely. For informational pages, collect the patterns "nedir" (what is), "nasıl yapılır" (how is it done), "ne işe yarar" (what is it for) separately, and check whether each pattern is satisfied by the same answer. A definition and a step-by-step answer are different jobs, even when one page can host both.

Spacing and compound spelling

Variants such as "web sitesi", "website", "internet sitesi" and "web sayfası" share meaning but not exact wording, and "e-ticaret" (with a hyphen) appears next to "eticaret". For published copy, follow the official spelling guide of the Turkish Language Association (TDK). For research, record the incorrect spellings people use as variants. That way the page uses correct Turkish and the research still counts the way users type.

Suffixes on brand and proper names

This is the point foreign brands most often get wrong. Turkish separates a suffix from a proper name with an apostrophe. The TDK rule page states: "Özel adlara getirilen iyelik, durum ve bildirme ekleri kesme işaretiyle ayrılır" (possessive, case and predicate suffixes added to proper names are separated by an apostrophe), with the example "Yunus Emre'yi". For brands this means "Google'da" (on Google), "LinkedIn'den" (from LinkedIn) and "Search Console'da" (in Search Console). Users often type "googleda" or "google da" instead. Keep the correct form in titles and copy, and keep what users actually type in the research sheet. The rule is on the TDK apostrophe rule page (Opens in a new tab).

Characters: the Dotted and Dotless i, Diacritics and ASCII Spellings

The dotted and dotless i problem

Turkish has four separate letters here: lowercase i, uppercase İ, lowercase ı and uppercase I. The Unicode special-casing data lists dedicated Turkish and Azeri mappings: uppercase I lowercases to dotless ı, uppercase dotted İ lowercases to i, and lowercase i uppercases to İ. A language-neutral lowercase function therefore turns "ISPARTA" into "isparta" instead of "ısparta". In a keyword sheet this creates two strings that look alike but do not match, which breaks deduplication and lookups. When you normalise a list in a spreadsheet or script, check whether the function you use applies Turkish locale rules. The mapping is in the Unicode SpecialCasing file under Turkish and Azeri (Opens in a new tab).

ASCII spellings: do they change intent?

Many people type Turkish without its special letters: "cagri" for "çağrı", "guvenlik" for "güvenlik", "ilac" for "ilaç". Keyboard, device and speed all play a role. The question for a researcher is whether the ASCII spelling expresses a different intent. In most cases it does not, and one page serves both. A few pairs are ambiguous, though: "saç" (hair) and "sac" (sheet metal), or "kuş" (bird) and "kus" (vomit, or the verb "to throw up"), change meaning once the diacritic disappears. Flag such pairs in your sheet and open the results page before assigning a page.

In published copy, do not deliberately mix ASCII spellings into the text. Write correct Turkish. ASCII appears mainly in the URL slug, covered below, and in the research sheet.

Keyboards and devices

On phones, autocorrect and keyboard layout change what people type. The same person might search "şekerli" on one device and "sekerli" on another. If your tool only reports the correct spelling, you may miss the other form. The Search Console Performance report describes its query dimension as the search query users typed, so you can look there for the spellings that already bring impressions to your Turkish pages. Use it as a second signal next to the tool data.

Building the Keyword Pool Step by Step

From seed terms to a variant matrix

The general process is described in our guide on how to do keyword research step by step. The Turkish layer adds the following steps:

1. Write five to ten seed terms in the bare form, with correct Turkish spelling.
2. For each seed, generate plural, possessive, object, direction, location and origin forms, checking roots with consonant softening by hand.
3. Add question patterns: nedir, nasıl yapılır, ne işe yarar, and the mi/mı/mu/mü forms.
4. Create the ASCII counterpart of each row and flag ambiguous pairs.
5. Join tool data to the matrix. Leave missing values empty rather than writing zero.

The aim is not a bigger list. It is a list in which no form is forgotten and every row has an intent attached.

Normalising: canonical form and variant column

Pick one canonical form for every concept: the correct Turkish spelling that will appear in your copy. Every other form lives in a variant column. A colleague who opens the sheet two months later can then see which row belongs to which page. Use a Turkish-aware lowercase function and test it on a small sample before running it on the whole sheet.

Checking against real queries

Many generated variants may never be searched; treat that as a rule of thumb and test it with your own data. To see which ones are searched, open the Search Console Performance report, which gives clicks, impressions, click-through rate and average position per query. Filter by page and look for your extra forms one by one. A form with only a few impressions does not justify its own content; it belongs naturally on the main page.

One Query, Many Forms: Page Mapping and Cannibalization

Suffix difference or intent difference?

Cannibalization risk comes from intent overlap, not from suffix differences. "Backlink satın al" and "backlink satın alma" are probably served by the same page, while "backlink nedir" (what is a backlink) and "backlink satın al" (buy a backlink) are different intents. To decide, compare the results pages: if the same URLs appear for both forms, they probably share one intent. For the detection method, see our article on how to detect keyword cannibalization.

Titles, headings and URL slugs

Use the grammatically correct, expected form in the title and H1. In subheadings, the inflection can vary naturally: one section can use the "nedir" pattern, another "nasıl yapılır". Repeating the same suffix form in every heading looks like keyword stuffing and tires the reader.

For URLs, Google's URL structure guidance recommends using words in the audience's language, transliterated where needed, says non-ASCII characters should be percent-encoded, and recommends hyphens instead of underscores. In practice, writing Turkish characters as their ASCII counterparts and separating words with hyphens produces a readable slug that survives copy and paste. See Google's URL structure guidance (Opens in a new tab).

Which page owns which form

Choose one primary page per intent and let inflected and ASCII variants appear naturally inside it. Open a second page only when the intent really differs. A mapping table showing which page owns which query cluster makes overlaps visible before you commission new content. The step from clusters to a content plan is covered in building topical authority with content clusters.

From Grouping to a Content Plan

Reading intent from patterns, and checking it

Suffixes and patterns hint at intent without proving it. Words equivalent to "prices" suggest commercial research, "how" suggests learning, "reviews" suggests evaluation and "buy" suggests a transaction. Keep that as a first assumption in a column and compare it with the format of the results page: lists, guides or product pages. Our article on matching search intent in articles explains how to confirm it.

Clusters and the main page

The output of grouping has three parts: a main query, its inflected and ASCII variants, and side questions. The main query drives the title and H1; side questions become H2s, H3s or FAQ items; variants appear in running text, not as a separate section.

Handing the list to a writer

When you pass the sheet to an author or agency, include not only the terms but also the canonical spelling, the spellings to avoid in the body and a note on intent. A writer who reads "never use the ASCII variant in copy, transliterate only in the URL" will stay consistent. For the writing stage itself, see the SEO article writing guide.

Worked Example: A Hypothetical Keyword Sheet

The table below is hypothetical. It contains no volume, difficulty or ranking values and only shows the column layout. The topic is an imaginary olive oil soap shop.

Two rules make a sheet like this work when adapted:

• The canonical column holds correct Turkish; the variant column holds everything a user might type.
• Each intent has one target page. Two rows pointing to the same page is fine; two rows with the same intent pointing to two pages is a cannibalization problem.

Common Mistakes and a Checklist

Data and tool mistakes

The most common error is adding up the volumes of inflected forms to find a "true" volume. Tools use different methods, and the same searches may be counted under several forms, so the sum may not mean anything. Use form counts and intent instead, and treat figures only as a relative priority. The second error is lowercasing a sheet without Turkish locale rules. The third is treating one tool as the only source without comparing it with Search Console.

Page and copy mistakes

Repeating one suffix form across all headings, writing incorrect spelling as if it were right, and placing raw Turkish characters in a URL that later breaks when shared are the usual faults. Correct Turkish in titles and body, readable transliteration in the URL and both spellings in the research sheet prevent most of them.

Pre-publication checklist

• Are canonical forms and variants in the sheet?
• Does each query cluster have one main page?
• Was i and ı normalised with Turkish locale rules?
• Were ASCII and inflected forms compared with Search Console data?
• Do brand names carry the apostrophe correctly?
• Is the URL transliterated, hyphenated and short?

Measurement, Review Cycle and Limits

What to watch

Open Search Console with the query and page dimensions together and look at clicks, impressions and average position for each variant. Average position is the average position of the topmost result from your site, so it is not the exact rank of one query. If your site appears twice for the same query, the report counts that as one impression, so do not rely on the report alone for cannibalization; look at the results page too.

When to review the sheet

There is no fixed schedule. Review the canonical table when a product launches, when competing pages change or when an unexpected variant appears in Search Console. Add a date and a reason on each change, so the decision is not questioned later for lack of context.

What this method does not do

The method promises no ranking. Correct mapping only makes sure your content can be considered for the right queries; rankings and traffic depend on many other conditions. Also, the issues described here are heaviest in everyday consumer topics. In sectors where many terms stay in English, the suffix problem appears less often.

Working With a Turkish Editor

Which claim is checked against which source

Three kinds of correctness are involved. Spelling and punctuation follow the Turkish Language Association guidance, character conversion follows Unicode documentation, and search behaviour follows your own Search Console data. Do not substitute one for another. A grammar rule does not explain search behaviour, and search data does not prove that a spelling is correct. Using an incorrect spelling in copy only because users type it weakens reader trust.

When a native-level editor is needed

A variant list produced by a foreign team or a script can include non-words such as "kitapı". Before using the list, ask a Turkish editor to review a random sample. If many rows are wrong, fix the generation rule and run it again; if few are wrong, accept the list. This test measures generation quality only; it does not show which rows are searched.

Colloquial and spoken Turkish

Users sometimes type informal speech such as "nasıl yapıcam" instead of "nasıl yapacağım". These spellings can appear as queries but do not belong in copy. Keep them in the sheet with a "spoken" tag, which helps you read intent, and write the page in standard Turkish.

Frequently Asked Questions

Does every Turkish suffix form need its own page?

No. Inflected forms mostly carry the same intent and one main page can serve them. A separate page is only justified when the intent really differs, for example a searcher who wants a definition versus one close to buying. Before deciding, open the results pages for both queries and check whether the same URLs appear.

Should I create content for Turkish queries typed in ASCII?

Usually not. An ASCII spelling mostly expresses the same intent and the same page serves it. Write the copy in correct Turkish and keep the ASCII form as a variant in your research sheet. Only for pairs where the diacritic changes meaning, such as saç and sac, inspect the results page and decide case by case.

Do keyword tool volumes merge Turkish suffix forms?

That is tool-specific and can change. Some tools list inflected forms as separate rows and others merge them. Query three forms of one root separately in your own tool and compare the output. Do not add the volumes together to find a true volume; use figures as relative priority only and back them with Search Console data.

Should Turkish characters appear in the URL?

Google's URL structure guidance says non-ASCII characters should be percent-encoded, suggests words in the audience's language (transliterated where needed) and prefers hyphens over underscores. In practice, writing Turkish letters as their ASCII counterparts and separating words with hyphens gives a readable slug that survives sharing.

How do I write Turkish suffixes on a brand name?

Under the Turkish Language Association rule, possessive, case and predicate suffixes on proper names are separated by an apostrophe, so write Google'da or LinkedIn'den. Users may type googleda instead; keep that spelling as a variant in the research sheet and use the correct form in titles and copy to protect reader trust.

How can I tell whether two Turkish forms cannibalize each other?

There is a risk if two different pages from your site appear for queries with the same intent. Open the query and page dimensions together in Search Console and then inspect the results page by hand. The report counts two results for the same query as one impression, so it is not sufficient evidence on its own.

Conclusion and Next Step

Three habits keep Turkish keyword research sound: separate suffix differences from intent differences, solve the i and ı problem at the data stage, and assign one main page to each intent. A sheet built this way also keeps the brief you give a writer consistent. The method promises no ranking or traffic; it only makes sure your content can be considered for the right cluster of queries.

For next steps, build the variant matrix from five seed terms, mark which forms appear in Search Console, and finish with a sample review by a Turkish editor. If you need a writer to turn the finished list into Turkish copy, you can browse the content writing listings on Akua Adforce. The scope and terms of each listing are written on the product and policy pages. Akua does not provide keyword research, so you would run that part yourself or with a third party.