Semantic Core
Модуль 2Сбор и чистка4/5

Search suggestions: a free source of 1000+ long-tail ideas

4 min read

When you type a query into Yandex or Google, suggestions appear under the search field — a list of 5-10 options for what you "probably" want to search for. These suggestions are a gold mine for the long tail. They're queries that real people are really searching right now, pulled from the search engine's recent statistics.

Suggestions vs other long-tail sources
1000+
Suggestions
across 3 levels of expansion
400+
Wordstat (right column)
for the same seed phrase
50
AI keyword research
/free/keyword-research
~50
On-site search
from analytics
Suggestions are the most 'alive' source: whatever makes it into autocomplete is actively being searched now.

Where suggestions come from

These aren't random words. Suggestions are real recent queries from other users, aggregated by the search engine. When a thousand people a month type "buy cake to order moscow" and the phrase gains popularity, it lands in autocomplete.

Suggestions update often (Yandex every few days, Google almost in real time). That means trends show up there earlier than in Wordstat. If a niche suddenly comes alive, suggestions reveal it first.

Suggestions also account for:

  • The user's region — Moscow and St. Petersburg get different suggestions.
  • History — what you yourself searched recently.
  • Season — in December it's "New Year cake," in January it's already gone.

Method 1: manual expansion

The simplest method, works without any tools.

Step 1. Open Yandex/Google in incognito mode (so your personal history isn't counted). Set the right region in your search settings.

Step 2. Type in your seed phrase. You'll see 5-10 suggestions — copy them all.

Step 3. Take each suggestion and type it in again — now each one has its own suggestions. And so on, down to 3-4 levels.

Example with pediatric dentistry:

Level 0: pediatric dentistry moscow
  ↓
Level 1 (5-10 suggestions):
  · pediatric dentistry moscow 24/7
  · pediatric dentistry moscow free
  · pediatric dentistry moscow districts
  · pediatric dentistry moscow center
  · pediatric dentistry moscow zao
  ↓
Level 2 (for "pediatric dentistry moscow 24/7"):
  · pediatric dentistry moscow 24/7 on-call
  · pediatric dentistry moscow 24/7 free
  · pediatric dentistry moscow 24/7 metro
  · pediatric dentistry moscow 24/7 cheap

Across 3 levels of one seed phrase you'll get 50-200 queries. Across all 5 seed phrases — 500-1000+ unique suggestions. That's the entire working long tail.

Method 2: suggestions + alphabet expansion

A hack: after your seed phrase, add letters of the alphabet one at a time — a, b, c, d, and so on. Each letter opens up a new set of suggestions.

pediatric dentistry moscow a → ...arbat, ...akademicheskaya, ...avtozavodskaya
pediatric dentistry moscow b → ...belyaevo, ...babushkinskaya, ...b free
pediatric dentistry moscow v → ...vdnh, ...vykhino

This is how you pull out every metro station, district, competing doctors' names, and other narrow queries. Run the whole alphabet and you'll get 5-10 times more suggestions.

Method 3: question words

Suggestions answer questions too. Add question words before the phrase:

how → how to choose pediatric dentistry moscow
where → where pediatric dentistry moscow
how much → how much does pediatric dentistry moscow cost
why → why does a child's tooth hurt

This gives you the informational long tail — ideas for the blog. Not for commercial pages, but for content that attracts warm traffic.

Method 4: automatic parsers

Once the core is already large (1000+ queries) and you need thousands of suggestions, manual expansion becomes unmanageable. That's when suggestion parsers come in.

Popular options:

  • Bukvarix — a free parser for suggestions and Wordstat.
  • Key Collector — a desktop program, powerful, ~5,000 ₽ one-time.
  • A-Parser — for those who understand parsers and want scale.
  • Suggestion API — Yandex and Google have no official public API, but there are wrappers that mimic a browser.

In our service, we plan to add a suggestion parser as a paid feature (planned tools). For now — Bukvarix/Key Collector.

What NOT to do with suggestions

  • Don't parse thousands of them through your own IP. Search engines ban such IPs for a day to a week. Use parsers with proxies if you need large volumes.
  • Don't ignore the region. Suggestions are regional. Parse from region "all of Russia" and then you're surprised why St. Petersburg and Krasnodar show up in the results.
  • Don't fully trust the trends in suggestions. They're part of the data, not all of it. Strong trends are visible, small ones aren't.

Frequently asked questions

How many suggestions can I collect by manual expansion?

500-2000 in an hour of work across 5 seed phrases with 3-4 levels of depth. That's enough for a mid-sized site. Need more — use Bukvarix or Key Collector.

Do suggestions in Yandex and Google differ a lot?

Yes, by 30-50%. Yandex has more regional specifics, Google more international ones. Parsing both systems gives you different slices of the audience.

Can I get suggestions through an API?

Officially, no — Yandex and Google have no public suggest API. Parsers mimic a browser. They work at a limited rate; for large volumes you need proxies.

Next article: seasonality and trends — when to publish so it takes off.

Read next