From transaction to relation

We’ve all heard of personalized content and ads. But how far have we really come, and what is the vision? Robin Hjelte, CRM Manager, shares his view on how to go from mere customer transactions to building user relations – without being annoying or creepy.

A light tap from your watch suggests that the store you are walking by offers a great bargain on the jacket you were looking at online a couple of days ago. When you enter, the store clerk has prepared the jacket in your size and favorite color for you to try on. It’s not cheap, but within your clothes budget. A quick glance at your phone confirms that this is the best deal on the jacket you can get in the area. You confirm the purchase with a touch to your watch.

As a little extra, on your way out the store offers you a complimentary snack, which is great, since you were beginning to feel a little peckish. It’s your favorite kind, and you have just the time to sit down for five minutes before you have to get to your next meeting. An excellent opportunity to check up on the latest news.

A staggering amount of data

This little scenario is pretty standard science fiction fare. Nothing too shocking, exciting or creepy. But of course the amount of personal data and computing power required to achieve an experience like this is staggering. Interests, positioning, blood sugar, income, calendar. Were this scene taken from a film it would likely move quickly into dystopian territory; intrusive ads, surveillance, hacking.

We’re clearly not there yet. But regardless of any personal opinions one might have on sharing personal data, the age old fact remains: the odds of providing a great experience to a prospective customer improve considerably the more a salesman knows the customer. He can anticipate needs, give advice, make suggestions. Furthermore a good salesman knows that investing in the relationship will pay back over time. This is also the basic logic behind loyalty programs of all scales – relationship building, turning the ability to build customer value over time into a competitive advantage.

Picking up on every signal

A salesman instead constantly following customers around, interpreting every glance as an invitation to push any random product laying around and jumping at every opportunity to close a sale would not be very successful. Yet, this is arguably what the online experience has turned into. Personalization and targeting in practice often means picking up on and acting immediately on every signal, no matter how weak, by pushing ads that sometimes follow the user around in endless efforts to retarget.

For ads this can be annoying, but the same logic applied to content recommendations can arguably lead to truly dystopian consequences, such as “filter bubbles”, echo chambers and confirmation bias.

Fortunately, personalization holds far greater potential. But in order to realize this potential, we need to move from treating every single contact, or touchpoint, with users in isolation, as an opportunity for a hard sell, to viewing them as a part of a whole, where we build relationship and user value over time. In short, applying common business sense to our digital user relations and becoming more like the good salesman. This is possible with better data, more advanced tools and sophisticated algorithms. And, increasingly, users expect us to deliver on this experience. A number of major trends from the last couple of years have influenced user expectations:

• The timeline – the endless stream of personalized content pioneered by the likes of Facebook and Twitter has become the primary interaction model for services with frequently updated content. No two users have the same exact experience, and user generated content, professional content and ads share the same space. Whatever finds its way into a user’s stream needs to be relevant in order to stand a chance
• The identified web – more and more services require login, which makes seamless experiences across devices possible. Native advertising and content marketing – ads and commercial messages share the same space as other content and need to be as engaging and relevant in order to compete for user attention.
• Wearables and the Internet of Things – means an increasing number of touchpoints with users: watches, small screen devices, health monitors, connected cars. These often interact with users in contexts that do not allow immediate conversions, but rather improve odds of conversion later. This all tells us that users expect constant updates, personalized to their liking, through mechanisms they may not fully understand but feel that they are in control of. They want a seamless experience and high quality.
  Clearly, the more touchpoints with users and the more data collected, the greater the opportunity to live up to these expectations and build valuable user relationships. Fortunately for Schibsted, as a large player with a strong local presence, we have the opportunity to collect this data because we can offer:

• Destinations and direct traffic – already established user habits.
• Highly valued, updated, original content – reasons for users to return frequently
• An ecosystem of services to capitalize on key events in the customer lifecycle.
• Trusted brands. Trust is the foundation of any relationship,
  and perhaps the hardest and most long term valuable asset
  any company can have. This is doubly important when
  handling large amounts of user data. Experience shows
  that convenience is more important than privacy concerns
  as long as there is trust. But as soon as trust is broken, this
  may be quickly reversed.

To make full use of these strengths we need to recognize that each user interaction has consequences, large or small, that either improve or worsen the commercial potential of the relationship. The trick is to subtly nudge the user over time toward long term maximization of customer value. In order to approach the challenge of building long term user relationships at Aftonbladet we have recently taken an initiative where we analyze and segment the user base based on behavior going back 12 months, tying together user profiles over time and across devices as far as possible. This means handling huge amounts of data, but it turns out to be very useful to understand usage patterns and changes in loyalty and engagement.

Monitoring small shifts

Aftonbladet’s most loyal 20 per cent of users make up a hugely disproportionate share of all page views, and an even greater proportion of revenues generated by ads and direct payments. Any changes in behavior from these users, such as them moving to a competitor for their daily news updates, have an enormous impact on revenue.

This means that even very small and slow shifts in behavior over time need to be closely monitored. In contrast, we have a considerable number of users that visit Aftonbladet less frequently than once a week. This highlights the danger of forming strategy and basing product development decisions based on averages. There is no such person as the average user, and trying to optimize for them is a certain way to make no one happy.

These are in and of themselves not extraordinary findings. In fact, it is so commonly observed in businesses that it is almost considered a natural law: the Pareto principle.

By segmenting the user base by degree of loyalty and personalizing the experience in accordance we can optimize every touchpoint for long term value, sometimes by sacrificing immediate revenue: a loyal user that is showing signs of decreasing return frequency might find a less distracting experience on their next visit. A previously infrequent user that is starting to return more often might be given some incentive to reinforce this behavior.

More specifically we are currently using this segmentation logic to identify which users we should try to convert to logged in users. By targeting users with an already loyal behavior we have the best chance of describing the benefits of logging in, as well as the greatest number of touchpoints to build the case over time. This allows us to prioritize development of logged in functionality to improve the user experience for this specific group.

“The trick is to subtly nudge the user over time toward long term maximization of customer value.”

The next step in this initiative is to further expand and improve our usage of data on our users to improve relationships. Improving loyalty over time requires that we become even more relevant. In order for long term conversion algorithms to be effective, they then need to factor in several other aspects that can be learned from behavior data:

• Context – how is the user coming to us?
• Interest – do we know what this user is interested in and expects to find?
• User mode, intent – the same user may have very different goals on two separate occasions. Sometimes the user wants to be in complete control, sometimes she wants to relax and be entertained.

Using these insights will make it possible to create strong relationships with our users and to provide excellent user experiences. This is where Schibsted needs to be going, and where our ecosystem of users and services will create the most value. Without being annoying or creepy.