I’m a strong believer in conversational interfaces — especially voice. Conversation is the natural way humans communicate, and it’s the future of human-computer interaction. If you remember the videos of two-year-olds swiping on iPhones and iPads, something similar is happening with devices like Alexa and Google Home: Kids already know how to interact with them.

Last year, my team conducted a survey of Alexa and Google Home users to better understand their behavior and satisfaction with the devices. It showed that interest in voice apps was beginning to really take off, with all types of enterprises and brands entering the space — media, CPG, retail, food delivery, banking, and a wide variety more.

This year, we re-ran the survey to see if, or how, user behaviors and feelings towards the devices may have changed. We also dove deeper into some of the interests based on demographics. The survey, conducted by Dashbot using Survata, covered 1,019 Alexa and Google Home owners across the U.S.

The key takeaways this year:

• Voice assistant devices are behavior-changing
• The core features tend to be the most frequently used
• Discovery of third-party voice apps is still an issue
• Users are quite likely to use the devices to make purchases
• Owners are satisfied with their devices and highly recommend them.

Voice assistants continue to be behavior-changing

As we saw last year, voice assistant devices are changing behavior. People use them throughout the day for a variety of use cases.

Nearly 75 percent of respondents use their voice devices at least once a day, with 57 percent using their device multiple times a day. These numbers are very similar to the results last year.

If we look closer at male versus female usage, approximately 64 percent of men and 53 percent of women use their devices multiple times a day. Among people who use their devices the least (less than once a month), women tend to predominate at 7 percent compared to just 1.4 percent of men.

More than 65 percent of respondents indicated the devices have changed their behaviors or daily routines. About a quarter felt the device has changed their behavior a lot, whereas 40.5 percent thought it has at least a little bit. Only 19 percent said the device has not changed their behavior.

A number of respondents described in their own words how much they rely on the device, how integrated it is in their life, and how surprised they are by how much they use it.

As voice assistants become more ubiquitous and the technology is embedded into even more types of devices, I expect to see more significant changes in behavior. If you are a heavy Alexa or Google Home user, how often have you caught yourself about to talk to the device when away from home — at work or in a hotel room while traveling? Amazon and Google are working on this though through their business initiatives to provide devices in hotels and other locations.

Men tend to report more behavior changes than women. Nearly 33 percent of men answered “yes, it has a lot” compared to 20 percent of women. As we saw with the frequency of usage, with women skewing more to infrequent usage, we also see a higher percent of women finding the device has not been behavior changing: 23.3 percent of women answered “no” compared to 13.7 percent of men.

Interestingly, even the 19 percent of respondents who indicated the device has not changed their behavior still use the device fairly regularly. Of those indicating “no,” roughly 33 percent still use the device multiple times a day, and another 17 percent use the device at least once a day.

Core features are the most frequently used

We asked respondents what features they use most frequently.

It turns out, listening to music, checking weather, and asking for information, are the most common use cases. They’re also core functionality of the devices. Using specific third-party skills is less common (more on that in a moment).

Approximately 75 percent of respondents use the device to listen to music, 66 percent check the weather, and 63 percent ask for information.

Approximately 58 percent of the those who listen to music do so multiple times a day, whereas only 34 percent of those checking the weather do so multiple times a day.

On the lower end of usage, only 23 percent of respondents use their devices for controlling home automation. However, those who do, do so quite frequently. Nearly 63 percent of respondents who use the device for home automation do so multiple times a day, and another 22 percent do so at least once a day.

If we look at the usage based on gender, interesting differences emerge.

While the top three use cases are the same for both male and female, women tend to have slightly higher usage for each — roughly 5-6 percent higher. For example, nearly 77 percent of women listen to music while 71 percent of men do.

There are some features that men are significantly more likely to use than women. For example, nearly 42 percent of male respondents use the devices for sports scores compared to 18 percent of women. Other features include getting news (49 percent of men vs. 40 percent of women), shopping (36 percent of men to 26 percent of women), playing games (33 percent of men to 22 percent of women), and home automation (29 percent of men to 18 percent of women).

Speaking of shopping, let’s take a closer look at this use case.

Users are willing to make purchases through their devices

Both Alexa and Google let users make purchases through their own e-commerce services and — with the addition of account linking — other retailers and services. Developers and brands can also monetize their voice apps through subscriptions and “in-app” purchases.

We asked respondents whether they have ever made a purchase through their voice assistant. It turns out 43 percent of respondents have, including 58 percent of men and 32 percent of women.

In regards to what respondents are purchasing, products from the providers own e-commerce service (Amazon or Google Shopping) are the most common at nearly 83 percent.

Interestingly, food delivery is also fairly common at 53 percent. The “reorder” case, i.e. the ability to reorder the same items as the previous order, works quite well through these interfaces, as it can be done in shorter, more concise statements than complex menu ordering. We’ve also heard from many food delivery services that reordering is quite common — consumers tend to order the same thing each time.

We also asked respondents how likely they are to make a purchase in the future. Approximately 41 percent said they are “very likely” to make a purchase in the future, with an additional 20 percent saying they are “likely” to do so.

Interestingly, one of the biggest indicators of whether someone has made a purchase in the past, or is more likely to make a purchase in the future, is whether they have both an Alexa and a Google Home. Over 56 percent of respondents who own both devices have made a purchase in the past, compared to 43 percent who only have an Alexa and 39 percent who only have a Google Home. In terms of future purchases, similarly 57 percent of respondents who own both are “very likely” to make a purchase in the future, compared to 41 percent of those who only have an Alexa and 35 percent who only have a Google Home. It may be that consumers who have both devices tend to be early adopters and more likely to try making a purchase through the device.

Discovery of third-party voice apps is still an issue

Voice interfaces are still a relatively new space. Between Alexa and Google Home, there are approximately 50 million devices in the US. Approximately 40,000 third-party skills exist for Alexa. We found in our last survey, that many respondents did not even know the term for a third-party voice app is a “Skill” on Alexa and an “Action” on Google Home.

The good news is, consumers are using third-party skills, they’re just not using very many of them. Based on the survey, 48 percent of respondents use between one and three voice apps, and an additional 26 percent use between four and six. Only about 15 percent of respondents said they do not use any.

We asked respondents what their favorite voice apps are. The more common responses were the native features — listening to music, checking weather, and getting info. The more common third party-apps named include Pandora, Spotify, Uber, and Jeopardy.

For third-party app makers, both discovery and user acquisition are challenges.

The most common ways users find out about Skills and Actions are through social media, friends, and the device app stores.

We often hear from brands and developers that social media, either paid or organic, is one of the best channels for user acquisition for voice apps. According to the survey, over 43 percent of respondents found skills through social media. Viral video influencer campaigns are also recommended as they serve two purposes — reach through the influencer, and instruction on how to interact with the voice app. Since it’s a new space and a new user interface, users may not know what they can say to, or do with, the particular voice app.

With Alexa, users can ask the device for the latest Skills or recommendations, even within categories. The device will walk through a set of Skills, listing each by name and asking if the user wants to install or continue.

In addition, Alexa supports a “can fulfill intent” that developers and brands can implement to help users discover their voice apps. For example, if an Alexa Skill can support ordering a pizza, the developer can list that as a “can fulfill” intent and potentially be recommended by the device when a user asks to order a pizza.

Google Home does not yet appear to have a searchable directory via voice. Asking the device for the latest Actions, or recommended Actions, results in either the fallback “I don’t understand” type of response or attempts to provide some form of definition depending on the request — e.g. describing a “sports action” when asking for the latest “sports Actions.”

User satisfaction is high

Users tend to be quite satisfied with their voice devices and recommend them highly.

We asked respondents how satisfied they are with the device’s ability to understand, the device response, and the overall experience. The results are quite positive.

In regards to the device’s ability to understand, nearly 44 percent of respondents were very satisfied, and an additional 34 percent were somewhat satisfied. Only about 13 percent were either somewhat, or very, unsatisfied.

Similarly, in regards to the device responses, 44 percent of respondents were very satisfied, and an additional 35 percent were somewhat satisfied. Only about 12 percent were either somewhat, or very, unsatisfied.

Based on the overall experience, 53 percent of respondents were very satisfied, and an additional 29 percent were at least somewhat satisfied. Only 10 percent were either somewhat, or very, unsatisfied.

In addition, we asked respondents if there was anything about the device that surprised them, and the results also indicate a high level of satisfaction. Owners were surprised by how much the devices can do and how knowledgeable the devices are. A fairly common comment was how quickly the device updates itself — “every day something new” and “like Christmas everyday.”

Owners are quite happy with their devices and would happily recommend them. When asked how they would rate the device overall on a one to five star scale, the respondents’ average rating was 4.4 stars.

When asked to rate how likely they are to recommend the device to others on a one to five scale, the respondents rated 4.4 as well.

If we look closer at the ratings based on the impact the device has had on behavior, we see overall positive results. Respondents who said the device has changed their behavior a lot rated the devices 4.9 stars and are very likely to recommend the device, with a 4.9 rating as well. Even users who said the device has not changed their behaviors rated their device nearly 4 stars and are still likely to recommend the devices with a 3.8.

We asked respondents if anything surprised them about the devices, and the more common responses were:

• How much the device can do
• How smart the device is and the ability to answer a variety of questions
• The ease of use
• The ability to understand the user’s request
• The user’s dependence on the device and how life-changing the device is
• The speed of responses
• The quality of responses

While most of the comments were generally positive, there was a small number of complaints. The biggest complaint (still occurring rarely compared to all the positive responses) was the frustration with the device’s ability to understand a user’s request.

Conclusions

Overall, owners of Alexa and Google Home devices are very happy with their devices. They are pleasantly surprised by all the things the devices can do, how smart the devices are, and how reliant they have become on the devices.

While the voice assistant space is still relatively new, there is an opportunity for brands to monetize as there is a strong indication of willingness to make purchases through the devices. As more brands develop voice apps, it will be interesting to see what use cases they support — how they take advantage of the voice interface and whether they implement monetization opportunities.

As many of the respondents mentioned, the devices are continuously getting better — not just in terms of improved comprehension, but in all the functionality provided.

I continue to be bullish on this space and look forward to seeing what the future holds.

Arte Merritt is the CEO and co-founder of Dashbot, a chatbot analytics platform for Alexa, Google Home, Facebook, Slack, Twitter, SMS, and other conversational interfaces.