Voice shopping has the potential to disrupt the CPG space unlike any other technology before it. While most brands are aware of the importance of voice as a sales and marketing channel, many still don’t understand exactly how tools like Amazon Alexa and Google Assistant work. Even if they do, they lack visibility into how their products are performing on this new channel.

To better understand how Alexa recommends products and utilizes Amazon product page content, we recently used the Echo Dot, Amazon’s best-selling Echo device, to test a variety of high-volume queries in consumer goods categories such as Household Supplies, Beauty & Health, Pet Supplies, and Food & Grocery.

Our test confirmed what many CPG firms have been fearful of all along: that Alexa gives Amazon more control over brands than ever before.

Below, we’ll outline exactly how Alexa recommends products in response to shoppers’ search queries, along with how it utilizes the content on detail pages to provide additional product information.

How Alexa recommends products

Alexa voice shopping is currently only available for Amazon Prime members, and it can only be used to order Prime-eligible products that are sold or fulfilled by Amazon.

Prime members can use a variety of voice commands to order products via the Amazon Echo and other Alexa-enabled devices. These commands include:

• Alexa, order [product].
• Alexa, reorder [product].
• Alexa, add [product] to cart.
• Alexa, search for [product].

When prompted with one of the above commands, Alexa first searches through Prime-eligible items in the customer’s order history.

Shoppers can respond to Alexa’s first product recommendation with “Yes,” “No” or “Tell me more.”

If they respond with “No,” Alexa asks if they would like additional results delivered to their phone. If they decline, Alexa then provides a second recommendation.

Alexa uses the above hierarchy to determine which product it recommends next. For example, if it suggests an Amazon’s Choice product first, it will suggest the top search result second.

Throughout the duration of our test, Alexa did not suggest more than two products per query. Every time we answered “No” to Alexa’s second product recommendation, we received the following response:

 

What is Amazon’s Choice?

When the Echo was first introduced in 2014, customers could only use it to re-order products they had previously purchased off Amazon.

In May of 2015, however, Amazon released an update that allows customers to use Alexa to purchase products they haven’t bought before.

At the heart of this update is a feature called Amazon’s Choice. This feature makes Alexa voice shopping easier, as it algorithmically identifies the “best fit” product for a given query. As mentioned above, Amazon’s Choice is often the first product Alexa recommends.

While Amazon isn’t completely transparent about the factors this algorithm takes into consideration, the company has said that “Amazon’s Choice recommends highly rated, well-priced products available to ship immediately.”

Since Alexa can only be used to order Prime-eligible products that are sold or fulfilled by Amazon, it’s safe to assume that Amazon’s Choice products must meet this criteria as well.

While previously only surfaced via Alexa, the Amazon’s Choice distinction is now also shown on Amazon’s website and app. It is not considered an ad space, and brands can neither apply nor pay to receive the distinction.

How Alexa uses product content

Our test revealed some interesting insights about how Alexa uses Amazon product detail page content when recommending items.

In response to a shopping query, Alexa always states the product title and price first:

When prompted to “Tell me more,” Alexa then shares the average rating, number of customer reviews, and bullet points, in that order:

Product Titles

During our test, Alexa rarely read product titles exactly as they appeared on Amazon product detail pages. Of the 27 products we reviewed, it used the exact on-page title only five times; in all other instances, it used a modified version.

The major difference we noticed between the on-page titles and the Alexa titles is that the latter tended to be significantly shorter. The average character count of the titles read to us by Alexa was 55, while the average of the on-page titles for the same products was 73.

Compared to the on-page titles, Alexa’s titles included only the most important details, leaving out superfluous terms and keywords used for search engine optimization.

Most of the time, these omissions were relatively inconsequential. On a few occasions, however, Alexa actually left out several important words, making it difficult to understand exactly what product was being recommended.

In addition to being shorter than the on-page titles, titles read by Alexa also seemed to follow the same general template:

While the brand name, product and scent/formula (if applicable) were present in every corresponding on-page title, Alexa frequently rearranged these elements to correspond with the above template. It also added the unit size and quantity whenever they were missing.

Bullet Points

When asked for more information about a recommendation, Alexa rarely recited all of the bullet points listed on the Amazon product page. In fact, of the 27 products we tested, Alexa read complete bullet points for only three.

In half of the remaining instances, Alexa provided some, but not all, of the on-page bullet points (anywhere from one to four). There was no discernible pattern or order in which these bullet points were chosen. However, the total character count did not exceed 300, with the average length being 221 characters. For the on-page bullet points, those numbers were 654 and 348, respectively.

Most interestingly, perhaps, is that for the remaining 13 products we tested, Alexa didn’t recite any of the on-page bullet points verbatim. Instead, it used a clearer, more natural-sounding description.

These descriptions typically reiterated the brand name, product, unit size, and quantity in the first sentence, and highlighted features and benefits from the bullet points in the second sentence:

A few of these descriptions were less straightforward than the above and used quirky remarks to showcase Alexa’s “personality”:

Next steps

Amazon clearly maintains a significant amount of control over both the products that Alexa recommends to shoppers and the content it uses to describe those products. This should be concerning for brands, especially those that are investing considerable time and resources into perfecting their Amazon page rank and product page content.

So what can brands do to get ahead of the curve and ensure a commanding presence on voice search? Here are a few steps to get started:

1. Compile a list of your highest-margin, highest-revenue product pages on Amazon.
2. Identify the keywords that are driving the most traffic to those pages from Amazon site search.
3. Determine where your product pages rank for those keywords, making note of any Amazon’s Choice distinctions.
4. Using an Alexa-enabled device, query the keywords for which you rank first or are Amazon’s Choice.
5. Audit the product titles and descriptions Alexa uses.
6. Contact your Amazon rep to discuss what can be done to fix any problem areas.

Looking for a scalable way to audit your brand’s presence on Alexa voice search? OneSpace is here to help.

By combining the industry’s largest database of consumer search insights with proprietary performance monitoring tools and on-demand content optimization services, we enable brands to respond to market changes and execute product page updates with unrivaled speed and scale. Learn more.