Our Alexa skill’s retention rate is off the charts. Here’s why:
Avg. Retention Rate: 3% Ours: 21%
Is it just me or have we been playing a global game of Where’s Wal- I mean, Alexa, recently?
As we all nostalgically remember the game, the challenge was in finding Waldo. But thanks to Bezos’s masterful game keeping, the challenge today is in not finding Alexa.
Alexa ads are plastered everywhere I turn. Radio commercials sound plain wrong without an Alexa shout out. I go to my friends’ apartments to find that they may or may not have TV’s in their living rooms, utensils in their kitchens, or bed frames in their bedrooms…But that infamous black puck — that physical vessel housing Bezos’s greatest brainchild, Alexa — is a must.
In fact, as difficult as it is for a beyhiver to admit, Alexa may be poised to become a bigger household name than Beyoncé…
But there’s still one massive challenge Alexa will have to overcome to dethrone Beyoncé on that front — loyalty. While Beyoncé fans consume her music as we do air, Alexa fans may purchase a device and use it once or twice. However, its novelty more often than not wears off and the puck’s utility is reduced to a paperweight.
This is no secret amongst voice technologists banking on the dot voice boom. The big question that remains is whether or not [Alexa, Siri, Cortana, Google Assistant…] will stick. If voice technology is to truly become as ubiquitous as many hope, usage will need to climb. Usage is twofold:
1. Usage of the Voice Devices, themselves (ie: Echo Dot)
2. Usage of the Voice Applications on the Devices (ie: The Gwynnie Bee Alexa Skill)
Usage of voice devices
A 2018 PwC consumer intelligence survey confirmed that while adoption of voice technology continues to skyrocket, usage lags far behind.
41% of 18–24 year old voice device adopters use them just a few times a month to a few times a year. That percentage is 34% for 25-49 year olds and 43% for 50+ year olds.
The study also found that though the youngest consumers surveyed are adopting voice technology at a faster rate than their older counterparts, they are statistically more likely to regularly use their voice assistants less.
Usage of voice applications on the devices
Voice devices like the Amazon Echo have voice applications called “Alexa skills” which users can enable on their devices. Since launching 3 years ago, the Alexa skill count has surpassed 30,000 in the US alone. However, skill developers share the same lack of usage dilemma.
A 2017 Voice Report by Voicelabs.co found that when a voice application acquires a user, there is only a 3% chance that user will be active in the second week.
Most voice applications collect more dust than they do repeat users.
Even so, from a business perspective, not having an Alexa skill today is like not having a website. You can’t not. So, back in January, my team and I were tasked with building an Alexa skill for Gwynnie Bee, a clothing rental service wherein members continuously rent their everyday wardrobe instead of buying it.
After launching, we were pleased to find that our skill achieved a 21% Week 1 to Week 2 retention rate. What’s more, 20% of those users returned in Week 3 and 11% returned in Week 4.
Here’s how we did it:
Building Our Skill
First, we had to determine which features to offer in our skill. We took a “blue sky” approach, ideating every feature we could possibly think of.
To narrow down that list, we had to align on what our objective would be for the skill. Our objective was decidedly retention, repeat usage, since our research revealed that was the underlying struggle for most skills. Specifically, we aimed to beat the 3% week 1 to week 2 retention rate cited in the aforementioned research report.
We prioritized 5 things:
1. Trust
In a world where Alexa frequents headlines as a wiretapper, maintaining the highest level of trust with our user base was critical. One decision we made light of this was to implement a double account authentication. Unlike most Alexa skills, you can’t enable our skill without authenticating your account with us too.
2. Utility
In scouring countless reviews of Alexa skills to piece together what Alexa adopters wanted most in an Alexa skill, we found that “utility” was the buzzword. So from all those “blue sky” features we initially brainstormed, we eliminated the ones that didn’t drive utility. Then, we chose our top 2 features to go out with first.
Feature 1 allows a Gwynnie Bee member to ask Alexa for her shipment status. We chose that because it is a top customer service inquiry. Simple enough.
Feature 2 is what we refer to as “Return Notify” in our service. To explain what that means, Gwynnie Bee functions as a woman’s virtual closet. When a member builds a virtual closet with us, she is shipped clothes from it. She wears the clothes, and when she’s ready for new ones, she exchanges her clothes at home for others in her virtual GB closet. The member can trigger her next box even faster if she “Return Notifies,” or notifies us of the return shipment in our system. We then start prepping her next box. Knowing that a priority for our members is to get their clothes as fast as possible, this feature was a no brainer.
3. Ease of Use
Voice technology brings with it the added challenge of not being able to rely on a screen. In our testing, we found that user interfaces that intuitively work with a screen are not feasible without one. This is also why you cannot simply transcribe a visual experience into an audio one and expect to retain users.
Our team went through multiple iterations of the dialogue flow to ensure that it was short, crisp, and descriptive. For example, when a user return notifies a garment with our skill, we programmed Alexa to prompt the brand and color of the item so our users can visualize it without actually seeing it.
4. Reward Usage
Much to my surprise, I have yet to come across an Alexa skill that effectively uses a rewards program. Our users are rewarded with loyalty points every time they “return notify” an item with our Alexa skill. These points can be cashed in for a variety of things in the Gwynnie Bee ecosystem. We don’t have enough data yet to say that the loyalty points issuance, specifically, is driving more usage of our skill. However, rewards programs exist because they work so voice applications should use them.
5. Build an Intimate Feedback Loop
Experimental projects like Alexa skills are ripe for testing and testing and more testing. Our team is all ears — literally, in listening to user feedback. We instrumented a closed discussion forum from the very beginning in which we constantly run polls and ask for improvement suggestions. 70% of our users engage in the forum and needless to say, the conversation loop allows us to be more thoughtful and agile in our iterative process.
Kaeya | Let’s mindshare. Tweet me your thoughts: @kaeyainc
(On behalf of the people at Gwynnie Bee who contributed to this effort)
