IDENTITY – Is it always “On”? and Who should control the switch?

Is Identity Always On

Thursday 3G

Convener: Vidya Jayaraman

Notes-taker(s): Vidya Jayaraman

Tags for the session - technology discussed/ideas considered:

Discussion notes, key understandings, outstanding questions, observations, and, if appropriate to this discussion: action items, next steps:

In today’s digital world, customers are seeking brand experiences and solutions that are simple, proactive, personal and efficient. In order to meet these customer needs, the key to is identify customers in that specific context, and use machine learning techniques to progressively build on that customer knowledge and needs. It can be viewed as five steps as shown in the attached illustration:

• Identify – this may be a customer-specified identifier like – name, address, email, phone, social or device-rendered such as  cookie, IP address, device ID, etc.
• Context Filter – this step drives most of the personalization as this records all context signals with the identifier at every instance. Signals can be as far as we want to go. Examples of signals include, location, time of day, duration of visit, length of scroll, frequency of use, etc.
• Profile – this is where we use machine learning to compare signals across a broader group of identities and develop profiles at an aggregate basis for segmentation and targeting purposes
• Map – this is picking the solutions and communications based on the aggregate profile
• Serve – this is rendering the solutions and communications using contextual signals to personalize the delivery of the experience

Key discussion points/Take away:

At least four different states of customer identification exists:

• Unknown (no identifier)
• Known (generic customer or device identifier, e.g. cookie, IP address, device ID)
• Known Logged-in (customer identified, e.g. account number, username)
• Known Logged-out (customer identified but has signaled a lower engagement state, e.g. reverts back to cookie/device ID but additional info is available)

  Some points to keep in mind during implementation:

• Identification is fluid and contextual (e.g. work vs. personal identity)
• Signals are implicit and gathered; hence requires careful use of signals to avoid unintended consequences/awkward customer experiences
• Balancing the use of contextual signals (guess work) and 1st party inputs to be proactive; use this to differentiate the journey for different states of identification
• Logged out states still carry the customer identity/identifiers. So ‘log out’ by customer is a signal to the company that the customer is changing states. So the experience should flex based on that signal.

It was a very informative session for me and thought you might also find it interesting.

I’ll be happy if anyone wants to discuss this any further.   Thank you,

Vidya