When someone types a search, they usually write in shorthand: "aircon repair Hanoi". When the same person speaks to their phone, the question becomes: "where can I find a reliable aircon repair service near me?" Both express the same need, but they produce very different keywords. That is why voice search deserves attention of its own, rather than being treated as just another slice of ordinary traffic.
How voice search differs from typed search
The first difference is length and naturalness. Spoken questions are usually longer, grammatically complete, and start with words like "where", "how", or "how much". Typed searches, by contrast, are stripped down to a few core words. A page optimised only for short keywords can therefore miss the whole group of long queries that people actually speak.
The second difference is local intent and immediacy. Many spoken questions include "near me", "open now", or "right now", because the speaker is often on the move or needs something solved immediately. For a local business, these queries are especially valuable, because the person asking is often close to calling or visiting.
Why voice makes rankings harder to read
The biggest difficulty is that you cannot separate voice traffic in ordinary reports. When someone asks by voice and then taps a result, that visit still shows up as a normal organic visit. When an assistant answers aloud without any tap, it is even harder to measure, because no visit is recorded at all, even though your brand has just been mentioned.
Another difficulty is the "single answer" nature of voice. With typed search, users see a whole results page and may click position two or three. With voice, the assistant usually reads out one answer, and that answer often comes from the page holding the featured snippet or ranking very highly. Competition for question queries therefore becomes winner-takes-all: either you are the answer that gets read out, or you are nearly invisible.
Indirect signs that voice is affecting you
Since you cannot measure voice directly, watch indirect signs. The first is a rise in question-style queries in your search data: phrases starting with "how", "where", "how much", or "should I". The second is mobile traffic to pages that answer specific questions, since most voice searches happen on phones.
The third sign is a change in landing-page behaviour. If a page answering a specific question suddenly gets more impressions without a matching rise in clicks, its answer may be getting read aloud or shown directly on the results page. That is not necessarily bad news: your brand is still being seen, even if click-based measurement does not capture the full value.
How to track rankings for voice-style queries
The first step is to carve out a separate group of question keywords in your tracking project. Choose questions real customers actually ask: about price, about how something is done, about location, about comparisons. Do not just track the short root keyword and guess; track the long questions themselves, because those are what people actually say.
On Keyword.help, you track rankings by country and device down to the Top 30, so pay special attention to mobile rankings for this group. A question keyword might sit at position 3 on desktop but position 8 on mobile, and for voice, the mobile position is the one that matters. Record the question group's rankings weekly, and compare them with your short keywords to see how differently the two groups move.
Next, cross-check rankings against featured snippets. For each important question, open the results page from time to time and see who holds the answer box: your page or a competitor's. If your ranking is still good but the answer box has changed hands, review how your answer is presented: put the question as a subheading and answer it briefly right underneath, before the detailed explanation.
Preparing content for voice queries without harming current SEO
A common worry is that writing for voice will dilute your content. In fact, the two complement each other if you get the order right: the page stays a comprehensive page about the topic, and the short answer to the question is simply one section within it. You do not need a separate page for every short question; you need each important question to have a clear, direct answer that is easy to find on the page.
Do not stuff spoken language everywhere, either. People still read with their eyes, and a page full of long spoken sentences is hard to read. The balance is to keep a natural tone, use question-style subheadings where there really is a question, and let the rest of the page do the deep explaining. If you do not yet have a question group in your project, you can start at the sign-up page and add the ten questions customers ask you most often.
Misconceptions to avoid
The first misconception is that voice optimisation is a separate project, apart from SEO. In reality, most of the work — answering questions clearly, loading fast on mobile, keeping local information accurate — is good for SEO in general. Treating voice as a lens for prioritising, rather than a separate channel, avoids duplicated work.
The second misconception is expecting an exact count of voice visits. No report gives that number reliably, and chasing it wastes time. Measure what you can measure: question-group rankings, featured-snippet ownership, and mobile traffic to answer pages, and use them as proxy indicators. The third misconception is that only large businesses need to care. For a local business, one well-answered "near me" question can produce a phone call immediately, and that value is very real even if it does not show clearly in a report.
Conclusion
Voice search does not replace typed search, but it changes the shape of keywords: longer, more natural, more local, and contested at the single-answer position. Since you cannot measure it directly, track a separate question-keyword group, watch mobile rankings closely, and check featured snippets weekly. For more rank-tracking methods, see the Keyword.help blog and start with the ten questions your customers ask you most.