跳至正文
数据、工具与监测 · Google SEO

5 Keyword Research Tools with Free Access: Features, Limits and Workflow

// / / 光算科技

You can start useful keyword research without buying a large SEO suite. Use Google Keyword Planner for demand estimates, Ubersuggest for keyword and competitor-page leads, AnswerThePublic for question discovery, Keyword Tool Dominator for platform-specific autocomplete, and Google Trends for seasonality. The output should be a reasoned page plan, not an article for every exported phrase.

“Free” means different things across these five tools. This comparison uses their public documentation checked on October 21, 2026. It distinguishes free access, restricted results and trials; it is not a paid-account benchmark or a promise that free products replace a fixed percentage of a commercial suite.

Which tool fits the task?

ToolUseful starting taskAccess or data limitation
Google Keyword PlannerExpand seeds and examine search-demand estimatesGoogle Ads account setup, including billing information, is currently required by the official guide
UbersuggestExplore keyword ideas and candidate competitor pagesFree search and full-access trials differ; reports and exports have restrictions
AnswerThePublicOrganize questions and comparison languageFree exploration is limited; subscription features and export rights differ
Keyword Tool DominatorCollect suggestions for a particular search or shopping platformThe current FAQ lists two free searches per keyword source each day
Google TrendsCompare relative interest and seasonal patternsPublic exploration, without an exact monthly search-volume count

Google Keyword Planner: establish the market before reading the numbers

Google's instructions distinguish “Discover new keywords” from “Get search volume and forecasts.” Use the first to expand words or a website into ideas; use the second when you already have a list. Keyword Planner is an advertising-planning tool. Its information can support SEO research, but an ad forecast is not an organic-traffic forecast.

The current guide says you must complete account setup by entering billing information to access basic features. It should not be presented as an anonymous tool with no account requirements. Examining a plan and launching an advertising campaign are different actions; check which workflow you are in before proceeding.

  1. Start with products, services or customer problems you actually serve. Enter relevant words, or use a suitable website as the seed.
  2. Check the location, language, Search Network and date settings. Save those conditions with the export so a global estimate does not silently become a forecast for one country.
  3. Remove irrelevant terms before sorting by volume. A fixed minimum-volume filter can discard specific, useful buying questions.
  4. Group the remaining ideas by the job the searcher wants to complete. Identify the page that could answer each group and whether it already exists.

Google defines average monthly searches using a keyword and its close variants, the selected period, location and network. Search-volume statistics are rounded. They are not counts of unique people, and overlapping variants should not be added together as though each represented a separate audience.

The Competition column describes advertisers, not organic keyword difficulty. Top-of-page bid ranges describe historical advertising bids; they are not the price of earning an organic ranking or a measure of your profit margin. Use them as context alongside relevance, demand, product economics and the actual search results.

Ubersuggest: turn estimates into pages worth examining

Ubersuggest's current landing page offers free search alongside a separate full-access trial. Its feature guide describes keyword overviews, ideas, questions, comparison phrases, competing domains and traffic estimates. Some exports are explicitly associated with paid accounts. Check the current allowance, visible rows and report permissions instead of assuming an old “three searches per day” statement is permanent.

For a product topic, enter a seed and check the location and language before reviewing suggestions. For competitor research, use a relevant domain and examine candidate keywords and pages. Open the pages: are they product listings, tutorials, definitions or comparisons? Estimated traffic is a lead for investigation, not access to the competitor's analytics account.

Keyword difficulty can help organize a shortlist, but scores from different vendors are not interchangeable. “Below 40” is not a universal new-site qualification rule. Ask whether you can serve the searcher's task and supply useful specifications, evidence or instructions that are missing from the current answers.

Record the useful phrase, relevant result URL and reason for considering it. When an export or report is locked, document only the information actually available. Keep AI-generated prompt ideas separate from observed search queries and volume estimates; brainstorming output is not a log of private user conversations.

AnswerThePublic: organize questions before deciding what to write

The official starter guide describes keyword wheels, question and comparison modifiers, People Also Ask and sections for different platforms. Use those views to explore what an audience may need, then assess search intent and the right page type.

Select the language, location and source supported by the interface. Enter a product or service topic and organize the returned questions by task. For another language, inspect the actual settings and results rather than appending an undocumented URL parameter. Sparse suggestions are a reason to combine sources, including genuine customer questions and relevant local search results.

“How do I choose?”, “How do A and B differ?” and “Can I use this in a particular environment?” may call for different evidence. Decide which questions fit together on one page. Do not automatically turn every phrasing into a separate article or paste the entire list into an FAQ.

Read the source labels carefully. The guide describes the AI Models prompts as suggestions generated from keyword variations and adapted to language and location. “Get AI Response” produces a generated answer. These features do not establish how often real people asked that prompt, and an answer generated during research does not establish consistent brand visibility across users.

The current plans guide separates searches, AI responses, content-generation credits and export permissions. It also notes additional export restrictions for free accounts. Check usage in your account before building a workflow around unrestricted CSV or image downloads. Limited exploration, a trial and a subscription are distinct access arrangements.

Keyword Tool Dominator: match the source platform to the project

Keyword Tool Dominator's FAQ describes suggestions drawn from platform autocomplete. At the time checked, its website tools offered two free searches per keyword source per day without registration, a credit card or an extension. The account dashboard and fuller research jobs use a separate workflow with the Chrome or Edge extension.

Choose the relevant source—such as Google, YouTube or a marketplace where you sell—before entering a short seed. Check the location, marketplace and language options available for that tool. A YouTube suggestion and an Amazon product phrase come from different contexts; neither is automatically a measure of Google web-search demand.

Review the modifiers for use case, model, compatibility, comparison or review intent. Keep terms your business can answer, merge duplicates and flag unfamiliar brands or models for checking. Have local-language wording reviewed before using it on a product page; translating a list is not enough to create a useful localized catalog.

The FAQ defines Keyword Score as relative popularity from 0 to 100 within the suggestions for a seed and says these autocomplete results do not provide search volume. A score of 90 is not 90 searches or a 90% chance of ranking. Comparing scores from unrelated result sets does not establish which market is larger.

Google Trends: check the pattern behind a promising idea

Google explains that Trends uses sampled searches, normalized by time and location and scaled from 0 to 100. A value of 100 is relative peak interest in the selected comparison, not 100 searches. A low-volume term can appear as zero without proving that nobody searched for it.

For a seasonal product, examine a longer period in the target market before narrowing to recent activity. Keep category and search-type settings consistent. Be clear whether you are comparing search terms or topics. Investigate a one-off spike for news or a temporary event before planning permanent content around it.

The related-search documentation distinguishes Top and Rising queries. Rising describes growth relative to the preceding period; “Breakout” means growth above 5,000%, not necessarily a large absolute audience. Use this view to find questions worth investigating, not an excuse to insert unrelated trending words into an existing introduction.

Worked example: a page plan without invented keyword metrics

Suppose a supplier sells camping tents. The phrases below are a teaching example, not measured keywords or a report of searches performed for a client.

Candidate phraseReader's taskPossible pageEvidence still needed
four person tentBrowse and choose productsA collection containing real productsDimensions, weight, season suitability and availability
single wall vs double wall tentCompare constructionA comparison guideActual structures, use conditions and limitations
how to clean and store a tentPerform maintenanceCare instructionsManufacturer guidance for the materials and model

The collection can cover close variations of its main buying task. The comparison needs evidence about the products being compared; the maintenance article needs applicable instructions. Producing three interchangeable sales articles would miss those differences.

In the research sheet, record the phrase, tool, platform, country and language, date, original metric name, observed value, intent, candidate URL and next action. Mark unavailable data as unavailable, not zero. Keep a reason for combining synonyms or assigning them to different pages. See the keyword-to-page optimization workflow for the next step.

Use your own results to improve the plan

Once pages receive impressions, Search Console's Performance report helps you examine your site's queries, pages, clicks and impressions. Google documents privacy omissions and data truncation in the query table; it does not list every query or represent the entire market. A candidate term missing from your property report does not prove zero demand, and a term with no recorded impressions has no observed CTR to use as a filter.

Review comparable periods and filters, then ask whether the right page is appearing and whether its answer needs work. A low CTR can have several explanations, including position, query mix and result presentation; it is not automatically a command to rewrite the title.

Pay for a tool when a specific limitation—such as historical coverage, complete exports or research scale—blocks valuable work. Start by using the available information to create one accurate, useful page. A defensible explanation of which page should answer a query is a better research outcome than a large list with no decisions attached.