Thank you for the post, Lina. ELSA is in a unique business – the language learning market is oversaturated (Rosetta Stone, Duolingo, flashcard websites, in-person classes, etc.) but there are few good, affordable options. I’ve never used any of these resources, but it seems like they each have a unique position in meeting a category of customer need – for example, ELSA students may be strong in writing and listening, but weaker in speaking.
The largest language app, DuoLingo, is free but supported with ads; I agree with Anand in that ELSA should experiment with a freemium model. I wonder if there’s an opportunity for ELSA a premium feature, where human tutors sign-in to the app to explain challenging concepts on-demand or help students who are working toward a particular goal (i.e. preparing for an interview in English).
Thanks for the article, Isha. ASOS is a terrific example of a company that leverages machine learning for custom curation; it’s done a terrific job keeping up with fast-paced trends in the industry. I look forward to seeing how ASOS leverages machine learning for retail price optimization, especially during promotional offers (common in fashion e-commerce) based on individual’s historical purchasing data and elasticity modeling. I’d also like to see how AI can address the industry’s return management issue – I agree with you – it’s a challenge that ASOS and the broader industry needs to address, but the retailers seem to think it’s just the cost of doing business. From organizational charts I’ve seen, few fashion companies devote a single divisional unit to returns management – creating further lack of ownership and transparency into its root causes.
Thanks for the post, Anand. Capital One is certainly a leader in FinTech. Eno sounds very interesting – one number per merchant definitely mitigates fraud and risk. I’m assuming it’s also great for free trial subscription services, especially those that may be hard to cancel as I’m guessing you could easily close the virtual card to stop a service from repeatedly charging you. P.S: Fun fact – “Eno” is One spelled backwards.
Thank you, Jiwon. Yes – the homogeneity in the women in striking (racial, gender, stance, hairstyle, dress). The only diversity I see is the direction of the stethoscope around their neck… ha! What also surprises me is that Craiyon doesn’t display “kindness” in their batch images beyond the nurse’s smile. Kindness should conjure images of understanding, helpfulness and care… these images just show women standing alone with their arms crossed. The only image demonstrating human interaction is the top-middle image of a figure in scrubs grabbing a nurse by the arm – that doesn’t seem particularly compassionate! Your Craiyon images further show how AI struggles to grasp the softer elements of the human experience.
This is so interesting, Katelyn. Yes, I agree with others – it’s fascinating to see 9 images of older men. I played with the Craiyon generator and when I type in “educators” it shows female school teachers and a mix of student genders, races, etc. I think the phrase that is triggering the images of older men is the phrase “leaders in the world.” I’m surprised there aren’t more references to the world in these images, such as flags behind the men, or other indications of traditional “world leadership”, such as images of people in military garb.
Thank you for the post, Amy! I think Craiyon’s images of the dogs (though slightly warped!) are absolutely adorable. It’s a stark contrast to the scary images it produces of humans – perhaps it will take Craiyon several more iterations to display humans properly. A part of me was hopeful that Craiyon may interpret your phrase with some nod to the Baha Men song!
I agree with the commenters that Craiyon has a difficult time with certain phrases – I tested several. Some phrases that are extremely common parts of our culture (“Star-spangled banner” , “The right to bear arms” , “Et tu, Brute?”) yielded very clear, expected results whereas others (“Who let the dogs out” or “There’s no place like home”) perhaps had too broad of an interpretation.
Thank you for the additional color, Irina! So glad to hear you’ve had stellar experiences with them! I’ll keep an eye out.
Thank you for the blog post, Kaitlyn. ToursbyLocals is a fascinating business; it’s interesting that it isn’t quite a tour company but rather a matching service for tourists and guides.
My top concern with ToursbyLocals’ business model is its scalability. First, it’s fundamentally difficult to grow a highly customized, private service. Each tour guide is offering a niche service to (presumably) a small group of customers per tour, resulting in limited reach and few opportunities to quickly build credibility in a new geographic region. Second, the more customized the service, the more challenging it is to identify (and compensate) high-quality talent. To recruit the best tourists, ToursbyLocals must undergo a long evaluation and onboarding period with competency and background checks, which makes it hard to source high volumes of talent at a fast pace to keep up with customer demand.
I am also curious about its operational sustainability. Consider this – what if a tour guide calls in sick… ToursbyLocals will likely have to cancel the trip and refund the customer. It’s impossible to find a stand-in or replacement for a tour guide if the tours are hyper-personalized. This can have a substantial impact on ToursbyLocal’s service continuity, customer retention, and profit margins.
Thank you for the blog post, Irina.
Too Good to Go sounds like a terrific initiative. I’d love to learn more about Too Good to Go’s profit model, as I believe generating additional revenue from the restaurant partners (and/or advertisers) is critical to scale (paying additional employees, building out complementary services, marketing to additional partner candidates like schools or farmers etc.). My top concern with Too Good to Go is its scalability. First, it seems this business is reliant upon a local cluster of grocer / restaurant partner participants to ensure appropriate supply to consumers. I’m sure there are inherent complications here – as lower income consumers (a potential target demographic) looking for a cheaper meal may live in a neighborhood with fewer grocers / restaurants available to participate. Second, I anticipate regulatory complexities that may impact food distribution laws and policies country-by-country, which may result in a customized launch approach that complicates scalability.
My secondary concern is in the quality of the meals Too Good to Go offers consumers. In my experience volunteering at a food bank, it was clear that most “partner” food establishments would only donate surplus (often stale) bread due to its low cost and very short shelf life (<1 week). How does Too Good to Go monitor restaurant partners and uphold the highest standards for food quality (no spoilage, mold, etc.) to maintain consumer trust in the platform?
Thank you for the blog post, Michelle.
It seems one of Goodreads’ significant value propositions is the ability to source a critical mass of book reviews from avid readers. My question is – how does Goodreads ensure the people who are rating books and posting reviews actually read the book? With such a large membership, Goodreads is likely the most influential book recommendation platform… it’s the perfect target for internet trolls. Although at first glance of the platform, most reviews seem legitimate, there have been complaints in the past that some members post reviews before a book has even been distributed. Similarly, some book reviews barely mention the book, but are instead responses to controversial social positions authors have taken (e.g. the reviews of the Harry Potter series). In order for Goodreads to remain a scalable and sustainable market leader, it needs to uphold the integrity of the platform in order to enhance trust in the quality of a stranger’s reviews. Otherwise, what’s preventing a member from leaving the service and joining a neighborhood book club?
Thank you for the blog post, Nitya. It’s true – publishers and book retailers are being disrupted by Amazon (as well as, I would argue, streaming services), as brick and mortar bookstores have historically bred connection with the consumer (author signings, author events, etc.) and discoverability. Metadata helps with discoverability and potentially predicting which books / titles will become best sellers, but beyond that, I’d be interested to see PRH and the other Big 4 publishers leverage iterative A / B testing to evaluate which titles or book cover designs would increase book sales. Additionally, it would be great to see PRH run experiments to refine which books they should publish in an ebook, hardcopy and/or audiobook medium. For example, I often buy longer books in ebook form (for convenience) and memoirs in audiobook (to humanize the author). I believe maintaining close relationships with authors to track sales, fan behavior/preferences, etc. will be the key to achieving cross segment sales and gaining market share; readers often remain loyal to an author and/or genre rather than a publisher (in my experience, readers rarely, if ever, set out to buy a book issued by a particular publisher).
I look forward to seeing how the publishing industry evolves to become more data-driven and targeted as the addressable market grows (with global literacy rates increasing) and barriers to entry drop.
Thank you for the blog post, Feifei.
I’ve seen first-hand all of the analysis you’re pointing to – from the Digital Flywheel program through to the tailored product expansion strategies by geography. Looking beyond how Starbucks collects customer data, it’s also been interesting to see how they leverage AI in Deep Brew to track inventory and recalculate replenishment orders, saving store employees hours of inventory management time. This technology is deployed at a granular machine level – collecting data on the beans used, beverage temperature, water quality etc.
As a Starbucks regular, I’ve benefited the convenience of the app that comes from ordering ahead and picking up, and have enjoyed seeing how technology has freed up staff operational time so that they can be deployed to the front of the store – interacting with customers.
Thank you for the blog post, Manuel.
I read your analysis through the lens of having just had a “false positive” with my American Express Platinum Card (i.e. my real purchase was flagged as potential fraud… perhaps because I was spending outside of my typical “pattern” after having made multiple transactions in a very short timeframe). I think machine learning is a great start to detecting fraud.
As an American Express client, I’ve noticed their ongoing efforts to improve their fraud monitoring, particularly with login attempts/account changes, keystroke patterns, and biometrics (i.e. fingerprints / face ID detection – which is normalized now but I’m sure initially sparked ethical questions amongst their Privacy and Risk teams). It’s also interesting to see how American Express has also framed many of these efforts to their clients as “time saving, convenient solutions” rather than fraud protection strategies, presumably to get clients who are not concerned with fraud to adopt these solutions. I’m happy to see American Express continue to invest in fraud detection and customer protection!