I’m A Barbie Girl in a Digital World

Mattel turning around their business using digital innovation, focusing on machine learning for their unique product development


Do you have a brother?” – asked the HELLO Barbie doll as I sat next to my cousin’s 5-year old daughter who was playing with her toys. Shocked to see this interactive version of Barbie, I couldn’t help but reflect on how toys have changed over time.

Mattel, a billion-dollar US toy manufacturing company, over the past several years, has invested heavily in machine learning to enhance their product development. In an era powered by digital innovation, Mattel has faced significant financial challenges with four consecutive years of revenue decline1. Developing innovative products that resonate with the digitally-forward parents of today’s generation is Mattel’s strategy to stay relevant in the future.

Interactive Products that use Machine Learning

In the short-term, Mattel has developed products which use artificial intelligence to simulate a live-conversation between kids and toys. For example, HELLO Barbie doll, launched in 2015 uses sophisticated algorithms and machine learning to have a 1-on-1 conversation with children as they play with the doll. The doll is programmed with 8,000 possible dialogues that can be used during the interaction with the child based on things the child says to the doll2. The doll remembers responses so that it can customize the interaction with the child through machine learning.

In the medium-term, Mattel is focusing on fueling digitization of products all the way from the top. In February 2017, Mattel hired Margo Georgiardis, an ex-Google executive to be its CEO, showing its commitment to digital innovation. While Georgiardis has recently been replaced by Ynon Kreiz3, who also has deep expertise in digital, Mattel continues to focus on gathering data to drive long-term future innovation. For example, HELLO Barbie collects and records data that the child says to the doll4. This data can then be used along with powerful artificial-intelligence capabilities to figure out preferences of children.

Caution for future product development

While Mattel has created this next-generation of toys, in the short-term they need to deeply understand how their products affect children’s behaviors and development. Particularly, it is critical to ensure biases are removed from their products so that children are not negatively influenced to believe a certain thing or pursue a certain career. For instance, Mattel created a book titled “Barbie: I Can Be An Engineer” to promote more girls to pursue STEM careers5. However, the storyline with Barbie relying on boys’ help for the more technical skills made the book an unpopular choice for parents to buy for their daughters. Similarly, with HELLO Barbie, I’m sure there are several instances where a conversation with the child might influence them negatively – whether its about notions of life, health, happiness etc. Even with the most-sophisticated machine learning algorithms, it is difficult to ensure neutrality on topics that might impact a child’s personal growth and development.

In the long-run, Mattel also needs to truly understand consumer’s needs and not invest capital in products that might be crossing ethical lines. In 2017, Mattel called off production of their newest product Aristotle, a baby monitor that used machine learning to respond to a baby and act as a virtual nanny6. The product was designed with latest technology partnering with Microsoft Cognitive Services and Bing Search Engine, such that it could do things such as turning on lights, changing color of lights or singing a lullaby based on the child’s voice. But since this data on children’s moods and behaviors was being tracked by Mattel centrally and could be shared with other organizations, it ran into privacy and data security issues and ultimately Mattel was persuaded by media backlash and consumer advocacy groups to call off the release of the product. Such instances prove that it is not sufficient to just invest in technology itself without fully comprehending how consumers will react to the product. Along with innovative technology, Mattel must conduct deep thorough research on parent’s and children’s preferences, reactions to new product ideas and partner with psychologists who can help determine impact of products on children’s behaviors. For example, there might be certain products that require less top-notch technology, but are liked more by consumers. As I remember my own experience of playing with toys, there are certain traditional aspects of toys that are still relevant and could provide joy to kids – such as the ability to do simple things with them like throwing a ball through the hoop without having to worry about internet connectivity, sync issues etc. Small changes in these products rather than quantum leaps in innovation could potentially make them very viable for the market.

Open Questions

Analyzing Mattel’s past product development and plans for future products, I can’t help but wonder, will digital technology and deep consumer research be sufficient to revive the company? Overall, what other levers can the toy industry pull to achieve sustained long-term financial results? (800 words)


1. Carol, M. (2018). The Long, Slow Decline of American Girl and Yoda. Bloomberg.com. https://www.bloomberg.com/news/articles/2018-03-29/the-long-slow-demise-of-my-little-pony-and-yoda Accessed 13 Nov. 2018.

2. Vlahos, James. “Barbie Wants To Get To Know Your Child”. Nytimes.Com, 2015, https://www.nytimes.com/2015/09/20/magazine/barbie-wants-to-get-to-know-your-child.html. Accessed 10 Nov 2018.

3. Whitten, Sarah. “Mattel’s CEO Georgiadis Heads To Ancestry. Toymaker Taps Ynon Kreiz To Replace Her”. CNBC, 2018, https://www.cnbc.com/2018/04/19/mattel-ceo-margo-georgiadis-reportedly-in-talks-to-leave-the-company.html. Accessed 10 Nov 2018.

4. Marr, Bernard. “Barbie Wants To Chat With Your Child — But Is Big Data Listening In?”. Forbes, 2015, http://www.forbes.com/sites/bernardmarr/2015/12/17/barbie-wants-to-chat-with-your-child-but-is-big-data-listening-in/#2f8661c2ec08. Accessed 10 Nov 2018.

5. Polgar, David Ryan. “How Mattel Continues To Move Barbie Into The Future – Dell Technologies”. Dell Technologies, 2018, https://www.delltechnologies.com/en-us/perspectives/how-mattel-continues-to-move-barbie-into-the-future/. Accessed 10 Nov 2018.

6. Schoonmaker, S. (2018). Free Software, the Internet, and Global Communities of Resistance: Hacking the Global. Routledge.


Can Facebook’s Artificial Intelligence Kill Fake News?


Overcoming Information Asymmetries in Credit Markets: Machine Learning at LendingClub

Student comments on I’m A Barbie Girl in a Digital World

  1. Thank you for the thought provoking post! It seems toys have come a long way since my Furby!

    My apprehension with this toy comes from the data security and privacy concerns that you highlighted. In particular, I worry that the data on these children may be mismanaged. As such, it would be very important for Mattel to develop strong security and standard processes for how this data is measured, saved, and protected.

  2. Such an interesting application of machine learning! While I appreciate Mattel’s effort to be forward-thinking, I’m not sure that it’s totally necessary for a toy company to implement that technology. As you mentioned, AI and data collection present a lot of ethical questions, and it’s not clear to me that that technology is actually necessary in toys. It smells a bit like desperation: a company in decline using buzzy tech to seem forward-looking. What’s next, Bitcoin games?

  3. This is a really interesting article and a trend in an industry that I (like the other commenters) find a bit surprising!

    You mentioned that some of the toys and Aristotle use machine learning and essentially collect data about the preferences of the child. In addition to privacy concerns for a vulnerable group, I wonder if the preferences of a young child are entirely relevant for Mattel. If his/her parents don’t approve, the child’s preferences are essentially entirely ignored. I think this is a case where the overall mission of Mattel is a bit confusing. Is it trying to use machine learning to positively influence early childhood somehow? Is it simply to scrape tons of data? It isn’t clear what role machine learning plays aside from the novelty aspect.

    I do think though that the toy industry can still benefit from technology and machine learning. One thought that comes to mind is creating educational and adaptive toys with AI/ML since it would be engaging for the child but also would generally elicit a positive reaction from parents.

  4. Thank you for your great post! It was really inspiring and I enjoyed it a lot.
    The biggest takeaway from your analysis on “AI vs Toy” is that technology itself won’t be always valuable; it will become valuable only when it is put into proper application. The current talks around AI are more lauding the potential benefit of AI than gauging the potential risk of AI. This particular example of Barbie is exactly the case where we are reminded that we should have “ownership” on the technology. At the same time, I also believe that unless we try it, we never learn or notice the risk. So it is good to consider actively what AI can do not only in typical AI-friendly industry but also some AI-new industry such as toy; The important thing is we humans keep having the “ownership” of the technology.

  5. Interesting read, RJ!
    What came to mind when reading this is some of the toys I used to play with as a kid. I do believe even back then there was some use of machine learning (we had the ability to speak to Furbies and command them to do certain things). Yet, still the toy industry is very fickle, preferences change so quickly and competition is stiff. Trends come and go.
    The question that comes to mind for me is whether or not significantly improved algorithms or substantially more powerful computer hardware (from then until now) can give this company a true competitive advantage. Can machine learning really lead to a competitive advantage given industry characteristics I mentioned above, now that machine learning is much better and more powerful than it used to be?

  6. Thank you for sharing! I was unaware of these initiatives that Mattel is working on and am surprised by the directions they are going in considering the privacy and child development concerns that you outlined. Your mentioning of the “traditional aspects” of toys made me think about the value proposition that Mattel provides with its Barbie products. The way I see it, these toys provide children with an opportunity for prompted make-believe and imagination. With a Barbie doll, a child can design the world that the doll lives in and use their imagination to create dialogue. In terms of value or product competitiveness, I see the dolls themselves, their clothes, and all props and accessories as sort of prompts or loose structures that the child can then use as a springboard for imaginative thinking and play. With these highly calculated and tailored interactive aspects being introduced to this mix with machine learning, does the value proposition of provoking imaginative play go away?

  7. Thank you very much for the interesting report! I have never come up with the idea that Barbie can be transformed with machine learning! This report reminds me of the case of GAP where the company started utilizing machine learning to decide the design and the type of cloth they produce. To your question, will digital technology and deep consumer research be sufficient to revive the company?, I would think they can capture a certain preference and market trend of the type of Barbie. But at the same time, I feel the reason customers(kids) buy Barbie is that the experience provided by having the Barbie doll, from such perspective, I would think it is a good idea to organize events for kids or expand Barbie shop where kids can play together with their Barbie would be more effective than chasing the trend of the style of Barbie!

Leave a comment