We all remember those moments when we stumbled upon another interesting article or video, just when we had finally decided to stop scrolling through our Facebook feed. Several years ago, Pocket built a solution to this problem of vacillating between postponing your important tasks for 5 more minutes (again) and risking missing out on something interesting. And recently, utilizing the huge amounts of data they gathered, the company launched a new service: Pocket Recommendations. In this post, I will point out how Pocket has managed to co-create the value provided through this new service with millions of people.
For the ones who do not know, Pocket is a service that enables you to save articles and videos to read/watch them at a more suitable moment. The service can be accessed through its own website, smartphone app and browser extensions, but also through an increasing number of external website and app plugins, allowing you to save content from almost every device, everywhere on the web.
Being someone who is often on the lookout to be more productive, I use Pocket on a regular basis. It turns out I am not the only one. Since its inception in 2007, the San Francisco-based company has built a user base of more than 17 million users that together saved over 750 million items to their Pockets in 2015 only (Kait, 2016; Pocket, 2016).
It goes without saying, especially for readers of this blog, that there is great potential in the user behavior data Pocket has collected over the years. Pocket now seems to be ready to take the next step in exploiting this potential, shown by the launch of its new service Recommendations several months ago (Chavez, 2015).
Pocket Recommendations provides you with personalized recommendations on content you are expected to find interesting, based on your and other users’ data. The launch of this new service indicates a shift for Pocket (Newton, 2015), from enabling users to save the content they found themselves, to providing them with interesting content right away, alleviating them from the need to actively look for and filter relevant content from the abundance of information available on the web.
Pocket’s Recommendations service, including proof of the fact that Pocket knows I like to learn how to be productive.
How does it work?
To get a better understanding of the mechanisms behind Pocket’s Recommendation service, it can be interesting to use the three building blocks of Personalization: Learning, Matching, Evaluating (Tsekouras, 2016). Firstly, users have been implicitly filtering and curating content for more than 8 years now, so there is an abundance of input to learn from. Pocket does not only know which articles/videos you saved for later, but also what you found interesting enough to read/watch when you, eventually, had the time. Secondly, by matching like-minded people and engaging in collaborative filtering, Pocket could be able to predict which content will be interesting to whom. Thirdly, Pocket Recommendations evaluates the effectiveness of the learning and matching mechanisms, not only by measuring on which items each user does and does not click, but also by allowing users to hide recommended items, the latter making them self-evaluating; active ‘Co-Creators’.
Evaluation: the Recommendations feed includes buttons with which you can hide an item.
There could be more to the evaluation phase than only assessing the effectiveness of learning and matching mechanisms, though. A ground-breaking success of Pocket Recommendations could theoretically result in a shift in user behavior, from implicitly participating in the filtering process to passively ‘consuming’ the content that is being presented in the Recommendations feed. The fact that a certain part of the user base does not engage in searching and filtering anymore, could lead to the problem that it becomes difficult for Pocket to stay up to date with changing or previously hidden preferences of each user. A solution to this could be to supplement content based on what is already known about a customer with content directed at learning about a customer, as suggested in a recent paper (Johar, Mookerjee and Sarkar, 2014). Looking at some of the suggestions that were shown in my Recommendations feed today (How to “Snapchat Like The Teens”), I am inclined to think they have already started to do so…
The example of Pocket and its recently launched Recommendations service shows how firms can leverage behavioral data to personalize their (content) offerings. It shows how firms can involve consumers in their value creation, even if these consumers are not fully aware of this. And it shows how one firm gathered data through one application for years, and now finally starts to exploit these through another.
Pocket has managed to involve millions of consumers in the curation of content, co-creating their own value. Assuming the company has managed to build a solid and well worked out new service – and based on my experience they have – the next question would be: how should Pocket monetize this value?
Chavez, R. (2015). Save-for-later app Pocket adds recommendations. [online] Mashable. Available at: http://mashable.com/2015/08/27/pocket-adds-recommendations [Accessed 10 Feb. 2016].
Johar, M., Mookerjee, V. and Sarkar, S. (2014). Selling vs. Profiling: Optimizing the Offer Set in Web-Based Personalization. Information Systems Research, 25(2), pp.285-306.
Kait, (2016). A Year in Review: Where We’ve Been and Where We’re Going. [Blog] Pocket Blog. Available at: https://getpocket.com/blog/2016/02/a-year-in-review-where-weve-been-and-where-were-going/ [Accessed 10 Feb. 2016].
Newton, C. (2015). Pocket starts recommending articles and videos in a new public beta. [online] The Verge. Available at: http://www.theverge.com/2015/7/30/9075497/pocket-recommendations-beta [Accessed 10 Feb. 2016].
Pocket, (2016). Pocket: About. [online] Getpocket.com. Available at: https://getpocket.com/about [Accessed 10 Feb. 2016].
Tsekouras, D. (2016). Customer Centric Digital Commerce – 02. RSM Erasmus University – Business Information Management.