Free Keyword Research APIs: Which Give You Search Volume?

Choose the keyword output before choosing an API

A free keyword research API is useful only if it returns the data your application needs. A seed-to-suggestions service, a search-volume lookup, and a report of queries that already brought your site impressions solve different problems.

Start by choosing the output:

If you want keyword data inside an agent workflow, start with the DataForSEO endpoints available through Glasser. Its catalog separates keyword research from search-result operations, so you can inspect a relevant operation before building around it. This is a paid data-access path; native provider trials have their own terms.

The comparisons below use official documentation checked on October 10, 2026. They are access and workflow comparisons; we have not benchmarked vendor coverage or keyword accuracy.

Compare free access and production requirements

Google Ads API: check planning permissions first

Google's keyword-planning services support idea generation and historical metrics. Access depends on your approved API setup: the current access-level documentation restricts planning services at the Explorer level and directs developers needing them to Basic or Standard access. Standard access also has a permissible-use scope.

Confirm access to the specific planning service before designing your integration. A working Google login, a test environment, or access to campaign reporting does not establish permission to run keyword-planning requests against production accounts.

DataForSEO: a small live trial and a separate sandbox

DataForSEO documents a $1 signup balance for API testing. The trial balance has no time limit, but supports a finite number of live requests. Its sandbox is for integration testing with dummy data; use a live request to evaluate the relevance of actual suggestions.

For discovery, its Keyword Ideas endpoint returns terms associated with the product or service categories of your seeds, together with keyword metrics. Check this behavior against your intended workflow: semantic expansion and a lookup of an exact existing list have different acceptance criteria.

Search Console API: research your existing search visibility

The Search Console API is free to use, with usage limits. Its Search Analytics method returns your authorized property's search-performance data and supports query and page dimensions. It is useful for discovering terms already associated with your site; use a market-wide keyword source when you need demand estimates beyond that evidence.

Glasser: inspect a supported paid operation

Glasser provides one access key across supported data sources. Its documented workflow is to find an endpoint, inspect its contract, and run it. Use the inspected price and outcome-specific charging rules for your budget; a provider's direct-account trial does not set Glasser's price.

Run a small keyword research acceptance check

Use a fixed research brief before comparing results. For example: “Find English-language US queries about appointment scheduling for a small dental practice.” This is an illustrative brief, with no measured keyword volumes implied.

1. Select a small seed set. Include the product, a concrete user problem, and an adjacent term you would reject. This reveals whether suggestions stay within your actual business scope.
2. Fix the market. Record the country, language, search network, and any additional targeting the selected operation supports. Retain the actual request with the response.
3. Separate discovery from enrichment. If the discovery response already contains the required metrics, inspect their definitions before paying for a second lookup.
4. Review relevance before sorting by volume. Label each returned term as usable, adjacent, or outside scope. Keep that decision separate from the raw supplier data.
5. Check missing and merged records. Preserve returned spellings, close variants, missing metrics, and duplicate groups. A shorter response can reflect normalization or coverage limits.
6. Save the evidence. Keep the source, operation, retrieval date, metric period, market, and your inclusion decision beside every accepted term.

A suggested research table is:

seed | returned_term | market | source | metric_period
volume | volume_state | variant_group | relevance_decision

These are fields for your own application, not a shared schema guaranteed by the APIs in this guide. If you use Glasser, inspect the chosen keyword operation to map its provider response into your table.

For a pilot, manually inspect enough returned terms to understand the main failure patterns. A tool can return many valid words while producing few queries that your business can use.

Read search volume and competition correctly

Google's historical metrics documentation describes average monthly searches, month-level volumes, competition, and bid-related fields. It also explains that close variants can be consolidated into one result.

Three interpretation rules help keep your shortlist useful:

• Attach a period to the volume. An average over a year and the latest month's estimate answer different questions. A seasonal query can look modest on an annual average while having one relevant peak.
• Keep advertising competition separate from organic difficulty. Google's competition level concerns advertisers competing for placements. Evaluate organic ranking difficulty using its own evidence and definition.
• Preserve consolidated variants. If two submitted phrases belong to one returned variant group, count the shared metric once when calculating an illustrative opportunity total.

Suppose two requested spellings map to a single result with a fictional monthly value of 200. Copying 200 onto both input rows and summing them would produce 400 while representing one metric group. Store the relationship between inputs and the returned record instead.

A missing volume should also remain distinct from a reported zero. Save the provider's returned state and investigate market coverage or input normalization before excluding the topic.

Finally, demand is only one selection criterion. Review the actual search results and your ability to answer the query before assigning it to a page. A large number alone does not identify the right format or promise.

Budget for the research workflow you will repeat

Count the operations behind one finished research batch: candidate discovery, metric enrichment, pagination, and any later SERP inspection. Use the documented charging unit for each operation.

For example, 20 discovery requests that yield 300 accepted terms produce a different cost profile from 20 requests that yield 30 accepted terms. In your own pilot, divide the actual spend by accepted, deduplicated records with the required fields. Keep raw request cost alongside that derived measure so the relevance filter remains visible.

For direct DataForSEO access, the FAQ lists a $50 minimum payment after the trial. That is a funding commitment, separate from the cost of a single research run. Hosting, storage, and review time also belong in a production budget.

Before repeating the job, decide when metrics need refreshing and whether your chosen terms of use permit retention. Store dated snapshots so a later change in demand can be distinguished from a changed provider, market, or metric definition.

Questions about free keyword research APIs

Can I get keyword ideas without search-volume data?

Yes, idea generation can return candidate phrases without demand metrics. Add a documented metrics source when your decision requires evidence about search frequency.

Does a free keyword website include API access?

API access needs its own documentation and entitlement. Check authentication, supported operations, and limits before treating a browser tool as an integration option.

Is Search Console enough for a new website?

Search Console helps once your property has relevant search-performance data. For topics your site has not appeared for, evaluate a seed-based or market-wide keyword source and inspect the resulting search intent.