From MediaPost:
Millions of consumers use Amazon’s Alexa for wide variety of things.
The voice assistant’s thousands of skills, basically voice front ends to standing apps, let a consumer talk to Alexa through an Amazon Echo device to order a coffee from Starbucks, tell the device to get an Uber or have it ask Fitbit how you’re doing today.
However, one thing many are not doing is making purchases through Alexa.
The majority (85%) of Amazon shoppers buy at least monthly on Amazon, based on a new study by Branding Brand.
. . . .
More than a quarter (28%) of consumers buy at least weekly from Amazon and 3% make purchases every day.
. . . .
Here’s how people buy from Amazon:
- 37% — Desktop, by website
- 25% — Smartphone, by app
- 23% — Smartphone, by website
- 7% — Table, from website
- 5% — Tablet, from app
- 1% — Alexa
- 0% — Amazon Dash buttons
Link to the rest at MediaPost
Looks like the same rigged study they use to prove more readers read paper than ebooks.
How can I tell you ask? Simple, they don’t add up to over 100% as they would if they counted those that use more than one way in each of the ways they are used.
I’m guessing they asked their questions pretty much in the order you see them up there — and stopped asking as soon as they got a ‘yes’. If they’d kept asking and recording ‘yeses’, we’d see the over 100% we didn’t see.
The pbook/ebook went the same way.
“Did you pick up and read anything in paper form this last year?”
“Well, yeah, I picked up some paperbacks at Half-priced Books — those that were too overpriced for my kindle.”
“Good, I can mark that as ‘paper reader’, thank you.”
“But I do more reading on my kindle.”
“Thank you for helping us with our study. Next!”
There were only 1,000 people in the study. Plus the Echo and the Dash buttons are relatively new. I’m skeptical of these numbers.
1,000 members in the sample is pretty good (n=1,000). Margin of error is inversely related to the sample size. 1,500 is the magic number (n=1,500 => 3% MOE). The problem with larger margins of error is that they are derived from samples too small to be robust (n=150 => 4% MOE); that is, they are not trustworthy. (That dropped the sogenannte news media into deep doo-doo in the last election. They bought their stats fast and cheap — Hey! 4% is eight times cheaper and faster than 3%. What’s the difference in 1%?) If the sample size is not robust, it may not support the conclusions reached.
Me? Yeah, I would like the sample size to be half again as large, but I will not lose sleep over n=1,000.
Let’s see if this is a one-off or if they keep at it.
Looks like Dash buttons are ignored, despite all the promotion Amazon gives them. I threw mine out because they wouldn’t connect to my wifi.
I don’t think people are generally looking for easier ways to buy stuff. It’s already very easy to buy things. People second-guessing their purchases is a problem for Amazon, but not for anyone else.