SEO10 min readSourced Data

Voice Search Statistics 2026: Smart Speaker and AI Data

Voice search statistics for 2026 from Pew, Edison Research, Alphabet, Amazon, Apple and OpenAI, with the widely quoted figures that do not hold up, checked.

Digital Applied Team
April 7, 2026• Updated November 5, 2026
10 min read
39%

Americans 12+ who own a smart speaker, Jan 2026 (Edison Research)

35%

US adults with a smart speaker, Feb 2026 (Pew)

150M+

People who talk to ChatGPT weekly using Voice and Dictation, Jul 2026 (OpenAI)

1 in 5

AI Mode searches using voice, images or documents, Oct 2026 (Alphabet)

Key Takeaways

We found no measure of voice as a share of all searches: Platforms disclose pieces: Alphabet says one in five AI Mode searches use voice, images or documents, and OpenAI says more than 150 million people use Voice and Dictation in ChatGPT each week.
Smart speaker ownership is a third to two fifths of Americans: Pew found 35% of US adults in February 2026, and Edison Research found 39% of Americans aged 12 and over in January 2026, after a change of survey method.
The assistants have been rebuilt on generative AI: Alexa+ is available to all customers in the US and Canada, Gemini for Home replaces Google Assistant on speakers, and Apple began rolling out Siri AI in beta on September 14, 2026.
Many voice SEO statistics come from 2018 studies: Backlinko’s 40.7% featured snippet figure and its 29-word answer length come from 10,000 Google Home results in 2018. The 29 words is the answer, not the query.
Voice recordings have led to large settlements: Amazon paid a $25 million civil penalty in 2023, Apple’s Lopez settlement created a $95 million fund, and Google agreed to a $68 million settlement fund over Google Assistant recordings.

How much do people use voice search in 2026? Nobody publishes a voice share of all searches, and many widely shared figures are forecasts, 2018 studies or numbers with no source.

This page collects voice figures whose publishers say whom they measured and when: Pew Research Center and Edison Research on smart speaker ownership, Eurostat, platform disclosures from Alphabet, Amazon, Apple and OpenAI, Backlinko’s and BrightLocal’s 2018 studies, a 2020 peer-reviewed speech recognition study, and court and regulatory records. Every row names its source and period, and figures were last checked on November 5, 2026. For help applying them to your site, see our agentic SEO work.

Voice Search in 2026: What the Data Shows

Voice search is harder to measure than most search channels. We found no search engine that publishes a share of queries that are spoken, and the surveys that ask about voice use differ in who they ask, what “voice search” means and when they asked. The most useful current signals come from platform disclosures about how people use their assistants, plus long-running surveys of device ownership.

Two recent disclosures show the direction. In October 2026 Alphabet’s CEO said one in five AI Mode searches globally use voice, images or documents, and in July 2026 OpenAI said more than 150 million people talk to ChatGPT each week using Voice and Dictation. Neither is a count of voice searches, and each comes from the company that sells the product.

For how these changes affect rankings and answers, see our guide to voice search optimisation and our data on zero-click search.

FigureWhat it measuresSource
1 in 5Share of AI Mode searches worldwide that use voice, images or documents, according to Alphabet’s CEO on the Q3 2026 call (October 28, 2026); the remarks do not give a voice-only shareAlphabet, Oct 28, 2026Official (company-reported)
Nearly 1 in 6Share of AI Mode queries that were non-text, using voice or images, according to Alphabet’s CEO on the Q4 2025 call (February 4, 2026)Alphabet, Feb 4, 2026Official (company-reported)
150 million+People who talk to ChatGPT each week using features like Voice and Dictation, according to OpenAI at the launch of GPT-Live (July 8, 2026); OpenAI does not split voice from dictationOpenAI, Jul 8, 2026Official (company-reported)
49%US adults who say they ever use AI chatbots like ChatGPT, Gemini or Copilot, Pew survey of 5,119 adults, February 17–23, 2026 (33% in 2024, when the question was worded differently and asked only of people who had heard of chatbots)Pew Research Center, Jun 17, 2026Survey
about 1 in 4US adults who use AI chatbots daily, Pew survey of 5,119 adults, February 17–23, 2026Pew Research Center, Jun 17, 2026Survey
93%Americans aged 12 and over who are familiar with at least one generative AI brand, Edison Research online survey of 2,050 people, January 2026Edison Research, Mar 2026Survey
62%Americans aged 18 and over who use a voice assistant on any device, including smart speakers, smartphones and in-car systems, NPR and Edison Research survey fielded February 28 to April 18, 2022 (the latest edition of the Smart Audio Report on the publisher’s page)NPR and Edison Research, Jun 2022Survey
57%Share of voice-command users in the NPR and Edison Research 2022 survey who use voice commands dailyNPR and Edison Research, Jun 2022Survey

Smart Speaker Ownership and Use

Smart speaker ownership has plateaued in the US rather than grown. Edison Research’s January surveys of Americans aged 12 and over show ownership between 33% and 36% from 2021 to 2025, then 39% in January 2026. Edison changed method between the two waves, from a mix of phone and online interviews in 2025 to an online-only survey in 2026, so part of that rise may reflect the change.

Pew Research Center asked a different population and found 35% of US adults say they have a smart speaker that uses AI. The two studies agree on roughly a third to two fifths of people, and neither measures households. Pew’s breakdown shows ownership rising with income, from 27% in lower-income families to 46% in upper-income families, and lowest among adults aged 65 and older at 24%.

Outside the US the picture differs. Edison’s UK study found 45% of people aged 16 and over owned a smart speaker in 2025, up from 25% in 2021, while Eurostat found 16.0% of people aged 16 to 74 in the EU used a smart speaker virtual assistant in 2024.

FigureWhat it measuresSource
35%US adults who say they have a smart speaker that uses AI, such as an Amazon Echo or Apple HomePod, Pew survey of 5,119 adults, February 17–23, 2026 (a person-level measure, not households)Pew Research Center, Jun 17, 2026Survey
33% / 41% / 36% / 24%US adults with a smart speaker by age group in Pew’s 2026 survey: ages 18–29, 30–49, 50–64 and 65 and older, in that orderPew Research Center, Jun 17, 2026Survey
27% / 37% / 46%US adults with a smart speaker by family income tier in Pew’s 2026 survey: lower, middle and upper, in that order (tiers based on adjusted 2024 earnings)Pew Research Center, Jun 17, 2026Survey
34% / 35%US adults with a smart speaker in Pew’s 2026 survey: men, then womenPew Research Center, Jun 17, 2026Survey
42%Asian American adults with a smart speaker in Pew’s 2026 survey, the highest of the racial and ethnic groups shown (White 33%, Hispanic 35%, Black 35%); the estimate represents English speakers onlyPew Research Center, Jun 17, 2026Survey
39%Americans aged 12 and over who own a smart speaker, Edison Research online survey of 2,050 people, January 2026 (35% in January 2025; the 2026 wave changed method, so part of the rise may reflect the change)Edison Research, Mar 2026Survey
112 millionEstimated Americans aged 12 and over who own a smart speaker, January 2026 (Edison’s projection of its 39% to the 12-and-over population)Edison Research, Mar 2026Survey
7% → 39%Smart speaker ownership among Americans 12 and over in Edison’s January waves: 7% in 2017, 18% in 2018, 23% in 2019, 27% in 2020, 33% in 2021, 35% in 2022, 36% in 2023, 34% in 2024, 35% in 2025 and 39% in 2026 (the 2026 wave changed method from phone and online interviews to online only, so the step from 2025 to 2026 may partly reflect the change)Edison Research, Mar 2026Survey
101 millionEstimated Americans aged 12 and over who owned a smart speaker, January 2025 (35%), Edison Research survey of 5,020 peopleEdison Research, Mar 2025Survey
22%Americans aged 12 and over who listened to audio through a smart speaker in the last month, January 2025 (21% in 2024)Edison Research, Mar 2025Survey
18%Americans aged 12 and over who listened to audio through a smart speaker in the last week, January 2025 (17% in 2024)Edison Research, Mar 2025Survey
45%UK population aged 16 and over who own a smart speaker, Edison Research’s Infinite Dial UK, 2025 (25% in 2021; the same publisher’s US figure for ages 16 and over rose from 33% in 2021 to 35% in 2025)Edison Research, read Nov 5, 2026 (2025 data)Survey
16.0%People aged 16 to 74 in the EU who reported using a virtual assistant in the form of a smart speaker in the three months before Eurostat’s 2024 surveyEurostat, Aug 28, 2025Official statistics
5,020Respondents in Edison Research’s Infinite Dial 2025: 1,006 telephone and 4,014 online surveys of Americans aged 12 and over, January 2025Edison Research, Mar 2025Survey
2,050Respondents in Edison Research’s Infinite Dial 2026: an online-only survey of Americans aged 12 and over, January 2026, half from the probability-based SSRS Opinion Panel and half from a non-probability sampleEdison Research, Mar 2026Survey
18% / 13% / 11%US adults who say they have a smart doorbell, a robot vacuum or a smart thermostat with AI features, in that order, Pew survey of 5,119 adults, February 17–23, 2026Pew Research Center, Jun 17, 2026Survey
37%US adults who say they have a smartwatch such as an Apple Watch or Google Pixel Watch, Pew survey of 5,119 adults, February 17–23, 2026Pew Research Center, Jun 17, 2026Survey
83%Share of US drivers and passengers aged 18 and over (in a car in the last month, with Apple CarPlay or Android Auto in their primary vehicle) who use it, Edison Research, January 2026Edison Research, Mar 2026Survey

How People Use Voice: Behaviour and Local Intent

The best-known local voice figures come from BrightLocal’s April 2018 survey of 1,012 US consumers, which is now more than eight years old. Its findings still circulate as if they were current. In that study 58% of consumers had used voice search to find local business information in the previous 12 months, and 27% visited a local business’s website afterwards.

Treat these as historical context for how local voice use began. For current local data see our local SEO statistics, and for the broader use of assistants see the NPR and Edison Research 2022 figures below.

FigureWhat it measuresSource
58%US consumers in BrightLocal’s April 2018 panel of 1,012 people who used voice search to find information about a local business in the previous 12 months (a use-ever measure, not a visit rate)BrightLocal, Apr 26, 2018Survey
46%US consumers in BrightLocal’s April 2018 panel of 1,012 people who said they look for a local business by voice on a daily basis (among voice search users)BrightLocal, Apr 26, 2018Survey
27%US consumers in BrightLocal’s April 2018 panel of 1,012 people who visit the website of a local business after making a voice searchBrightLocal, Apr 26, 2018Survey
76%Smart speaker users among US consumers in BrightLocal’s April 2018 panel of 1,012 people who perform local searches by voice at least weekly (53% do so every day)BrightLocal, Apr 26, 2018Survey
18%US consumers in BrightLocal’s April 2018 panel of 1,012 people who had used smart speakers for local voice searches, the least common of the devices asked about (smartphones were the most popular)BrightLocal, Apr 26, 2018Survey
3xHow much longer queries in Google’s AI Mode are than traditional searches on average, according to Alphabet’s CEO, October 28, 2026 (the comparison is with typed searches, not a voice-only measure)Alphabet, Oct 28, 2026Official (company-reported)
53%Smart speaker owners in the NPR and Edison Research 2022 survey who have heard an ad on their device and say they are likely to respond to ads on itNPR and Edison Research, Jun 2022Survey
50 billion+Visual searches per month on Google, according to Alphabet’s CEO, October 28, 2026 (a camera and image measure, listed because Google groups it with voice as a non-typed way to search)Alphabet, Oct 28, 2026Official (company-reported)

The Assistants Are Being Rebuilt Around Generative AI

Every large voice assistant has been or is being rebuilt on a generative AI model. Amazon’s Alexa+ is available to all customers in the US and Canada, and Google has said Gemini for Home replaces Google Assistant on its speakers and smart displays. Apple began rolling out Siri AI in beta on September 14, 2026, after announcing in January that its next foundation models would be based on Google’s Gemini models.

None of these companies publishes a comparable market share, accuracy or usage figure for voice. Amazon’s 600 million is a count of devices, and Amazon does not say how many are active. Cross-assistant “market share” tables that circulate online do not come from a publisher we could find.

For Google’s voice and camera search features, see our guide to Google Search Live.

FigureWhat it measuresSource
600 million+Alexa devices “out there today”, according to Amazon’s Alexa+ announcement (February 26, 2025; the figure was still on the page when Amazon updated it on July 21, 2026); Amazon does not say how many are activeAmazon, Feb 26, 2025 (updated Jul 21, 2026)Official (company-reported)
$19.99 a monthPrice of Alexa+, which Amazon says all Prime members get for freeAmazon, Feb 26, 2025 (updated Jul 21, 2026)Official (company-reported)
US and CanadaWhere Amazon says Alexa+ is available to all customers; it is in early access in Brazil, the UK, Mexico, Italy, Spain, Germany, Austria and FranceAmazon, Feb 26, 2025 (updated Jul 21, 2026)Official (company-reported)
Mar 14, 2025Date Google said it was upgrading Google Assistant users on mobile to Gemini, adding that classic Google Assistant would no longer be accessible on most mobile devices later that yearGoogle, Mar 14, 2025Official (company-reported)
Oct 1, 2025Date Google announced that the Gemini for Home voice assistant would upgrade and replace Google Assistant on its speakers and smart displays; advanced features need a Google Home Premium subscription, with plans from $10 a monthGoogle, Oct 1, 2025Official (company-reported)
$99.99Pre-order price of the Google Home Speaker, which Google calls the first device built for the Gemini for Home voice assistant (in stores June 25, 2026)Google, Jun 17, 2026Official (company-reported)
1 billionUsers of the Gemini app, according to Alphabet’s CEO on the Q3 2026 call (October 28, 2026); Google does not say how many use voiceAlphabet, Oct 28, 2026Official (company-reported)
Jan 12, 2026Date Apple and Google announced a multi-year collaboration under which the next generation of Apple Foundation Models will be based on Google’s Gemini models and cloud technology, to help power a more personalised SiriGoogle and Apple, Jan 12, 2026Official (company-reported)
Sep 14, 2026Date Apple began rolling out Siri AI in beta in English (French, Japanese, Korean, Portuguese and Spanish to follow the next month); it was not available initially in the EU on iOS, iPadOS and watchOS or in ChinaApple, Sep 14, 2026Official (company-reported)
Nov 2, 2026Date of Apple’s fiscal fourth-quarter results, in which Apple’s CEO referred to “the all-new Siri AI” as part of the holiday-season lineup; Apple gave no Siri usage figuresApple, Nov 2, 2026Official (company-reported)
Jul 8, 2026Date OpenAI launched GPT-Live, a full-duplex voice model that can listen and speak at the same time, as the new ChatGPT Voice (rolling out globally on iOS, Android and the web; nine remastered voices)OpenAI, Jul 8, 2026Official (company-reported)

Speech Recognition Accuracy

We found few independent accuracy comparisons between assistants that name their method. One peer-reviewed example is a 2020 PNAS study that tested five commercial speech-to-text systems and found large differences between speaker groups. Its systems are several generations old, so treat it as evidence that accuracy can vary by speaker, not as a ranking of today’s assistants.

FigureWhat it measuresSource
0.35 vs 0.19Average word error rate of five commercial speech-to-text systems (Amazon, Apple, Google, IBM and Microsoft) for Black speakers compared with white speakers, 2020 peer-reviewed study (lower is better)Koenecke et al., Apr 7, 2020Academic study
42 and 73White and Black speakers whose structured interviews the study’s authors had the five systems transcribe; the systems tested are those of 2019 and 2020, not the current generationKoenecke et al., Apr 7, 2020Academic study

Voice SEO: What Google Documents

Google’s documentation on voice-specific features is limited. Its speakable structured data is still labelled beta and applies to news content for US English users on Google Home devices, and Conversational Actions, the way developers built custom voice experiences for Google Assistant, were turned down in June 2023.

For content, the practical advice from the data above is the same as for other answer surfaces: write concise answers to the questions people ask, and be the page that answers them. Our voice search optimisation guide covers the tactics, and our agentic SEO service applies them to your content.

FigureWhat it measuresSource
BETAStatus of Google’s speakable structured data (Article and WebPage), which Google says identifies sections best suited to audio playback by text-to-speechGoogle Search Central, read Nov 5, 2026Platform rule
US, EnglishWhere Google says speakable works: for users in the US with Google Home devices set to English and publishers publishing in English; Google uses it to answer topical news queries on smart speakers, returning up to three articlesGoogle Search Central, read Nov 5, 2026Platform rule
Jun 13, 2023Date Google turned down Conversational Actions, the custom voice experiences developers built for Google Assistant; App Actions, smart home, Actions from web content and Media Actions were unaffectedGoogle for Developers, read Nov 5, 2026Platform rule

Privacy, Settlements and Regulation

Voice assistants record speech, and recent legal records show the cost of disputes over it. Amazon agreed to a $25 million civil penalty with the FTC and Department of Justice in 2023. Apple settled the Lopez case for $95 million, and Google agreed to pay $68 million into a settlement fund over Google Assistant recordings, according to the court-approved settlement notice.

In the EU, the AI Act requires providers to tell people when they interact with an AI system, and lists emotion recognition and biometric categorisation among its high-risk biometric uses. It does not create a separate category for continuous voice monitoring, whatever secondary sources say.

FigureWhat it measuresSource
$25 millionCivil penalty Amazon agreed to pay in the 2023 FTC and Department of Justice settlement over Alexa children’s voice recordings and deletion requests, announced May 31, 2023 (the order also required deleting inactive child accounts)FTC, May 31, 2023Law or court record
$95 millionSettlement fund in Lopez v. Apple over unintended Siri activations; class period September 17, 2014 to December 31, 2024, with payments capped at $20 per Siri device on up to five devicesLopez Voice Assistant Settlement, read Nov 5, 2026Law or court record
$68 millionAmount Google will pay into the settlement fund of the Google Assistant privacy class action over recordings from unintended activations, per the court-approved notice (agreed January 2026)Google Assistant Privacy Litigation, read Nov 5, 2026Law or court record
May 18, 2016 – Mar 19, 2026Period covered by the Google Assistant privacy settlement class, which covers people whose communications were recorded or obtained as a result of a “False Accept” or disclosed to a third-party review vendor; the claim deadline was August 27, 2026Google Assistant Privacy Litigation, read Nov 5, 2026Law or court record
Article 50(1)EU AI Act duty on providers to design AI systems that interact directly with people so that those people are told they are interacting with an AI system, unless that is obvious (Regulation (EU) 2024/1689)EU AI Act, Jul 12, 2024Law or court record
Annex III, point 1The biometric uses the EU AI Act lists as high-risk where permitted by law: remote biometric identification, biometric categorisation by sensitive attributes, and emotion recognition; the Act has no category called continuous voice monitoringEU AI Act, Jul 12, 2024Law or court record

Widely Quoted Voice Search Figures, Checked

Many voice search statistics circulate without a source, or with a source that says something different. Each one below is checked against the publisher’s own page.

Often quoted asWhat the source says
8.4 billion voice assistants are active worldwide in 2026, more than people on EarthJuniper Research’s April 2020 press release forecast that consumers would interact with voice assistants on over 8.4 billion devices by 2024, up from 4.2 billion devices expected by the end of 2020. It is a six-year-old forecast of devices, not a 2026 count of users.
4.2 billion people use voice search each monthThe only 4.2 billion in Juniper’s release is its estimate of voice assistant devices expected in use by the end of 2020. It is not a count of monthly voice searchers.
42% of US households have a smart speakerNo current publisher gives 42% of households. Pew’s February 2026 survey found 35% of US adults have a smart speaker, and Edison Research’s January 2026 survey found 39% of Americans aged 12 and over own one. Both count people, not households.
31% of all searches are voice searchesWe found no publisher that measures voice as a share of all searches. The nearest current disclosure is Alphabet’s: one in five AI Mode searches globally use voice, images or documents, which is a share of AI Mode searches and includes images and documents.
The average voice search query is 29 words longBacklinko’s 2018 study of 10,000 Google Home results found that the typical spoken answer is 29 words. It is the length of the answer Google Home read out, not of the question asked.
Voice results load in 2.68 seconds and come from pages of 1,890 wordsBacklinko’s 2018 figures are 4.6 seconds to load completely and 2,312 words on average. Its 36.4% schema and 9th-grade reading level figures match the page, but the load time and word count do not.
62% of US adults use voice search weeklyThe NPR and Edison Research Smart Audio Report (spring 2022) found 62% of Americans aged 18 and over use a voice assistant on any device, a use-ever measure rather than weekly voice search.
58% of voice searchers visited a local business within 24 hoursBrightLocal’s 2018 study found 58% of consumers used voice search to find local business information in the previous 12 months, and 27% visit the website of a local business after a voice search. It reports no 24-hour visit figure.
Voice commerce will reach $164 billion by 2028, growing 24% a yearWeb pages repeating this figure attribute it to Juniper Research without a link, and no Juniper page we could read contains it. It could not be traced to a publisher.
Google Assistant holds 36.2% of voice assistant share; accuracy is 93.7% for Google, 91.2% for Siri and 89.8% for AlexaNo publisher measures a cross-device voice assistant share in this form. We found no primary source for these shares or accuracy figures.
Amazon holds 53% of the US smart speaker installed base, Google 28% and Apple 11%No free primary source for brand shares of the US installed base could be located. Edison Research measures ownership of any smart speaker (39% of Americans 12 and over in January 2026), not brand shares.
48% of UK households have a smart speakerEdison Research’s Infinite Dial UK 2025 found 45% of the UK population aged 16 and over owns a smart speaker. It measures people, not households.
The EU AI Act classifies continuous voice monitoring as high-riskAnnex III of Regulation (EU) 2024/1689 lists remote biometric identification, biometric categorisation and emotion recognition under biometrics. It has no category called continuous voice monitoring. Article 50(1) requires people to be told when they interact with an AI system.

Quote the Source, the Sample and the Year

When you cite a voice search statistic, carry the source, the sample and the year with it. “39% of Americans aged 12 and over owned a smart speaker in January 2026 (Edison Research)” is checkable, while “42% of households” or “voice is 31% of searches” is not.

Use the data to plan content that answers spoken questions clearly, then measure results in your own analytics. For help turning this into an SEO plan, see our agentic SEO service.

Sources

Every figure on this page comes from one of the sources below.

Surveys and statistics from research organisations
Laws, regulators and platform documentation
Company disclosures and newsrooms
Vendor studies and court notice sites
  • BacklinkoWe Analyzed 10,000 Google Home Results. Here’s What We Learned About Voice Search SEOFeb 28, 2018
  • BrightLocalVoice Search for Local Business StudyApr 26, 2018
  • Settlement administrator (Lopez v. Apple)Lopez Voice Assistant Settlement: official notice siteread Nov 5, 2026
  • Settlement administrator (Google Assistant privacy class action)Google Assistant Privacy Litigation: official notice siteread Nov 5, 2026

Get Your Content Ready for Spoken and AI Answers

We audit how your pages answer the questions people ask by voice and in AI assistants, and fix what stops them being used.

Answer-ready content and structured data review
Local and conversational query research
Measurement against your own search data

Digital Applied newsletter

Deep dives on AI, marketing and development.

Practical guides and fresh insights by email. No recycled takes.

Related Articles

Continue exploring with these related guides