Summary: In this article we consider the potential advantages and challenges of an organisation having access to a unified digital profile for each of their customers. We use an example of booking a vacation online to illustrate how the user experience differs greatly depending on the amount of information available during the booking process. We conclude by exploring solutions to this problem that meet both the privacy needs of the individual while providing an alternative revenue model for those organisations that have detailed, but incomplete digital profiles on their customers.
In this article we consider the potential advantages and challenges of an organisation having access to a unified digital profile for each of their customers. We use an example of booking a vacation online to illustrate how the user experience differs greatly depending on the amount of information available during the booking process. We conclude by exploring solutions to this problem that meet both the privacy needs of the individual while providing an alternative revenue model for those organisations that have detailed, but incomplete digital profiles on their customers.
Each day, our interactions with commercial and public sector organisations allow them to create a more accurate and complete digital profile of us. Whether we are sharing photos of a recent vacation with friends on Facebook or Google+, updating our professional skills on LinkedIn, tweeting an interesting article on Twitter or buying a book from Amazon, we allow the service provider to learn a little more about us. As each organisation gains a more accurate and complete profile for each of us, it should lead to them being able to serve us better. Most would agree this is generally a good thing for both consumer and provider. However, using machine learning over our individual profiles to provide a better service, isn't without challenges or concerns. The first challenge clearly relates to how accurate and complete a profile is and how the profile relates to the services being provided. The second challenge, as an organisation develops a more accurate picture of who we are as individuals, the greater our concerns around privacy and data ownership. Finally, as new organisations are formed, each has to go through the costly and often frustrating process of learning the digital profile of each customer from scratch.
To illustrate the potential advantages of sharing our digital profile with an organisation, let us consider how a travel website recommends a vacation for us today and contrast that to how amazing it could be. I will make the assumption that you choose to explore options and book on-line. The process typically consists of you providing information on where you will depart from, the date of departure, how many nights, who is going and the country and possibly city you want to visit. In addition to this information, some travel websites may present you with wide categorisations such as "Beach Holidays" or "Adventure Holidays". You then spend the next thirty minutes exploring the website looking for a location that you think would result in a great vacation. You are still not sure so decide to postpone the decision until you can spend a little longer looking at possible destinations. This often occurs a number of times and often across a number of different travel websites until eventually you decide to settle on a package.
Now let us consider what could be possible if a travel website had access to your unified digital profile. Firstly, when accessing the site, you would be asked if the website could temporarily access your profile for the purpose of presenting you with a vacation that best meets your needs. You are told that they will not have access to your raw profile data and cannot make a copy for future use; this is verified by a trusted internet security provider. They also state that all recommendations presented to you will not be recorded for future use unless you explicitly allow them to. You agree to provide them with temporary access and immediately are presented with three suggestions for a vacation. Your full profile allowed the website to learn that you follow a number of photography related blogs, frequently share photos with friends and followers on Facebook and Pinterest and you recently bought a new camera lens from a local store. Your profile shows that you recently bought a travel guide on South America and one about Machu Picchu from Amazon.com. Your profile also shows that you have a business trip booked for North America during the same month that you want to go on vacation. Combining each of these facts, it suggests a photography vacation in Peru and furthermore, given that you are a member of a local trail walking group, it suggests the more physically challenging variation of the tour that requires a greater level of physical fitness. After getting over the seemingly erie accuracy of the recommendation, you proceed with the booking and complete the entire process within a matter of minutes.
So why are no travel websites currently providing this level of sophistication when recommending a vacation? Apart from the challenges of building sophisticated machine learning models, the travel websites do not have access to your unified profile. Actually, unless you had previously booked a vacation through them, there is a good chance that they know almost nothing about you. Even if you sign-in to the website using your Facebook credentials, Facebook's application privacy statement states that only a small amount of your profile is shared; this includes your name, user id, username, profile picture, gender, network, age range, language and country plus any other data that is public. Assuming that you minimise your public digital footprint, this information reduces the number of possible recommendations by only a small amount. Even if Facebook changed their privacy policy and allowed the third party to access your full Facebook profile, then while it would dramatically improve the accuracy of recommendations, using the prior example, it wouldn't have visibility into your business travel, recent purchase history from Amazon and a local store.
So what would a potential solution to this partial profile problem look like? I see two possibilities: the first possibility and one that resonates with privacy and data ownership advocates is for the individual to own and maintain their own profile. This profile represents your true self, your likes and dislikes, past experiences, skills and knowledge, significant others, purchase history etc. A world in which you are the owner and gatekeeper to your profile has significant benefits and could turn a number of business models on their head. However, there are a number of challenges in owning your own profile, first and foremost, how do you keep it up to date automatically based on how you interact with the world around you and secondly how can you provide third parties such as a travel website with access to it in such a way that they cannot simply copy its content in the process of providing you with a service. The latter is something that we can address today and is the foundation of a technology I am actively collaborating on bringing to the market. The former however is far more challenging as today organisation see the information they gather on individuals as highly valuable and not something to simply give away unless required to do so by law or in exchange for a new revenue stream.
Another possibility, and one that I believe is a stepping stone to individuals ultimately owning their unified digital profile, is to provide a mechanism to connect together profiles collected by different organisations and allow third-party organisations to perform machine learning across combined data sets for the purpose of a clearly defined task, such as recommending a vacation. We can leverage privacy preserving techniques to protect each distinct digital profile and ultimately allow a third-party to access a unified profile to better serve you. This not only provides great value to you, the consumer, but also provides an new revenue stream to any organisation that possesses a valuable, accurate and up to date segment of your profile.
The value to both consumers and providers in working with a unified digital profile is simply too great to continue along the current path of working with partial, incomplete data. There are of course a number of challenges and concerns that will need to be overcome across a number of areas, including privacy (within the field of security), machine learning, governance and advertising. However, if we demonstrate value to both consumers and providers at each step in the evolutionary process, I see little reason why booking a vacation in the future cannot be as pleasant and efficient an experience as highlighted in this article.