How do you find voice search keywords?

To find voice search keywords, start with seed terms related to your business, use keyword tools to find specific, naturally phrased questions, and turn to Google Autocomplete for additional ideas. But these are candidate keywords, not confirmed records of voice queries: Semrush explicitly notes that the tool data is not designed specifically for voice search, so its search volumes cannot be treated as voice search volumes.

Start with specific needs, not just longer keywords

A 2021 Semrush article notes that voice queries tend to be longer than typed queries and more closely resemble complete questions. Semrush’s 2023 keyword research recommendations focus on three categories:

• Long-tail keywords: These reflect specific needs and usually have lower search volumes; word count alone does not define them.
• Question keywords: These use question words such as “how,” “where,” and “when.”
• Conversational keywords: These read like natural conversation rather than an awkward string of nouns.

The three categories can overlap. As an editorial example, “air conditioner cleaning” is a seed term, while “Where can I book air conditioner cleaning nearby?” adds location and service needs. However, rewriting a phrase as a conversational question only gives you a candidate to investigate; it does not prove that users have actually searched for it.

Use a question filter to find keywords, then check their relevance

Semrush’s 2023 article describes a workflow for Keyword Magic Tool. The steps below reflect the process at that time and may not match the current interface:

1. Enter a seed term related to your business or topic and select the target country.
2. Use the “Questions” filter to find questions related to your services.
3. If needed, narrow the results using the “Easy” or “Very Easy” difficulty filters, but do not choose a keyword solely because its difficulty is low.
4. Keep queries that sound natural, express specific needs, and can be answered by your website’s content. Organize them into a list, then review the search results.

A difficulty score is not the probability that a voice assistant will use your answer. The KD metric described in the article measures the difficulty of reaching the top ten traditional search results; long-tail keywords are not necessarily low-competition, either. Google Autocomplete can provide additional keyword ideas, but it cannot tell you whether a query came from voice search.

Match “nearby” and “Are you open today?” to actual services

Local needs are also worth considering when looking for keywords. A 2025 Semrush article illustrates the connection between voice search and local SEO with an example of using a voice assistant to find a nearby sushi restaurant. You can therefore use “nearby services” as a starting point, but specify the actual service in your keywords rather than inserting the same phrase into every page.

For “Are you open today?”, first check whether your website can answer that question. A better-supported approach is to identify relevant questions first, then create content that answers them. Semrush’s 2023 article also recommends incorporating relevant keywords naturally into content and prioritizing readers’ needs. However, the available material does not directly assess whether adding these terms has negative effects, so there is no guarantee that doing so is risk-free. Nor can you conclude that adding them will improve rankings.

At Procrustes, our content services turn the questions that genuinely matter to clients into articles, with ongoing planning, writing, and updates.