Subscription services have seen tremendous adoption and growth, with more than 15% of online consumers saying they’ve tried or are currently using one or more subscriptions.
Valuations for startups are soaring into the billions for companies like FabFitFun, and, before their IPO, Stitch Fix. Consumers are welcoming these services and they’re scaling into millions of users, while competition in this crowded space is fierce.
In other words, they’re perfect for A.I. based personalization and optimization.
This is further evidenced in a McKinsey report stating that the most popular reasons users continue with subscriptions are: the desire to have a personalized experience (28%), the desire to be surprised and delighted (17%) and the desire to have a tailored experience (15%).
And each of those reasons can be driven by A.I., with initiatives like:
- Intelligent product recommenders
- Meaningful onboarding
- Powerful indicators of churn and customer LTV (lifetime value)
- Informed marketing
- Data driven sales
- Cart abandonment models
- Similar-product swaps
We’ve spent years developing expertise across each of these initiatives, scaling to millions of users and billions of impressions. If you’d like to learn more about what we learned and how it may benefit you, click hear to download our report: The top 7 ways A.I. is Transforming the Subscription Box Industry [Insider Report]
https://bennettdatascience.com/assets/BennettDataScience_AIforSubscriptionBusinesses.pdf
Of Interest
Fashion A.I.
Did you know that the fashion industry is estimated at nearly 3 trillion dollars in 2018, representing 2 percent of global GDP?! And that’s not lost on big retailers. Can you think of a major retailer that doesn’t sell apparel? Target, Walmart, Amazon all do.
Along with big brands and big money comes competition and thinner margins. Increasingly, fashion brands are using A.I. and machine learning to maximize users’ shopping experience, improve the efficiency of sales systems through intelligent automation, and enhance the sales processes using predictive analytics and guided sales processes. A.I. is even used in the design process. Read more here.
With our deep experience and expertise in this space, expect to see more Tech Tuesdays dive into fashion A.I. in the coming weeks.
More on GANs:
A GAN is an A.I. that takes small amounts of guidance and creates often dramatic results. From text generation that is nearly impossible to discern from human authors to faces that look like real people but aren’t. GauGAN, is a type of GAN that creates photorealistic images from segmentation maps. It turns rough sketches into photorealistic masterpieces and is being used by artists to prototype ideas and make rapid changes to synthetic scenes.
The results are quite dramatic. With very limited input, GauGAN is able to create powerful scenery that artists are using to speed the creative process. See the video here: https://youtu.be/NKFrg9HMYaY. Interesting in trying it. Go here: www.nvidia.com/ai-playground
Some Things get Worse over Time
When a data scientist deploys a shiny new model she expects it to accomplish some objective(s) right away. She may also want to finally take that trip to blue water and boat drinks, after months of model development. Not so fast! Data science models degrade, for lots of reasons. Deployment isn’t the final step, even if models are set to refresh (re-train) on new data nightly.
“Assuming that a Machine Learning solution will work perfectly without maintenance once in production is a faulty assumption and represents the most common mistake of companies taking their first artificial intelligence (A.I.) products to market.” Read more here.