Somebody will tell you that half of all searches will be voice by next year. That prediction was originally made about 2020, it did not happen, and it has since been quietly reissued for 2027.
Which is a reasonable place to start, because the published statistics on voice search are the least reliable body of numbers in local marketing, and most articles on the subject are recycling claims from 2019.
The numbers genuinely contradict each other
Look at what gets published about the single most basic question, how much of voice search is local.
One widely cited figure says 76% of voice searches carry local intent. Another dataset puts location-specific modifiers at 22% of voice queries. A third says local business lookups account for 16% of query categories by volume. A fourth says voice accounts for roughly 27% of local "near me" searches, which is a different measurement again.
These are not small discrepancies and they cannot all be describing the same thing. Some measure share of voice queries that are local, others measure share of local queries that are voice, and most articles quote them interchangeably.
So treat any confident voice search percentage with real suspicion, including the ones below.
What does appear to be true
Strip out the contested figures and a consistent picture remains.
Smart speakers stalled; voice on other devices did not. The dedicated speaker category has flattened, while voice queries through smartphones, cars, and inside apps continue rising. In-car voice has grown substantially with CarPlay and Android Auto, which makes sense: hands occupied, eyes needed elsewhere.
Voice queries are longer and shaped as questions. Reported average length runs far above typed queries, with a large majority phrased as complete questions and a high share containing question words. "What time does the hardware store close on Sunday" rather than "hardware store sunday hours."
Voice assumes local intent even without a local modifier. If a query could plausibly be local, the assistant treats it as local.
Voice returns one answer, not ten. This is the most consequential structural fact and the least discussed. There is no page two. A spoken result is position one or nothing.
The genuinely actionable finding
Different assistants pull from different places, and one of them is routinely ignored.
Google Assistant draws on Google Business Profile data and the local pack. Alexa leans on Bing Places. Siri pulls from Apple Maps, informed by Apple Business Connect.
Siri holds a substantial share of smartphone voice queries, and Apple Business Connect is free, takes about twenty minutes, and the overwhelming majority of local service businesses have never claimed their listing.
If you do one thing after reading this, that is the one. Not because voice search is transformative, but because it is a free listing on a platform a large number of iPhone users are querying, and almost nobody in your trade has bothered.
Bing Places is the same argument with less traffic behind it. Worth an afternoon if you have one spare.
There is no separate voice SEO project
Here is the part that saves you money.
Vendors sell voice search optimization as a distinct service. It is not one. Everything that makes you findable by voice is the same work that makes you findable in the map pack.
A complete Google Business Profile with services filled in. Consistent name, address and phone across the places assistants read. Recent reviews. Pages that answer real questions in plain language, because question-shaped content matches question-shaped queries naturally. A site that loads fast on a phone.
That is local SEO. If you are doing it, you are already optimising for voice, and anyone selling you a separate voice package is selling you the same work twice.
The single genuine addition is claiming Apple Business Connect and Bing Places, which is a listings task rather than an optimisation programme.
What voice actually changes about content
One thing, and it is worth adopting regardless of whether voice matters.
Write the way people ask. A page headed "HVAC Maintenance Services" competes poorly against one that answers "how often should a furnace be serviced" in the first sentence, because the second matches how the question was actually spoken.
An FAQ section with real customer questions, answered in one or two plain sentences each, is the highest-value voice work available and it costs nothing but the time to write down the questions you already get asked every week.
The honest read
Voice search is real, growing, and considerably less dramatic than the statistics industry around it suggests. It is a way people reach local businesses, particularly from cars and kitchens, and the queries carry high intent because somebody speaking a question usually wants something now.
It is not a channel you can invest in separately, and it is not a reason to change your marketing plan.
Claim Apple Business Connect. Add an FAQ that answers your actual common questions in plain language. Keep your profile complete and your details consistent everywhere.
Then ignore anybody quoting you a percentage about voice search, including this article, and get back to the work that produces jobs whether the customer typed the question or said it out loud.