Excited about customer analytics? Sort your single customer view out first

“I have already told you via email, do I have to repeat myself on the phone?”

“Why is this company so incompetent?”

“You should know this information!”

Angry Client (Ex-Customer) 2016

It is a fact that if you cannot keep your customers happy they will leave you for another company and getting them back will be hard. Studies by the American Society of Quality Control have shown as much as 68% of customers leave due to poorly managed customer relationships and bad customer data is a big part of this. Customer Analytics has become the talk of the town and many companies are rushing to try and understand their customers using heaps of data due to fierce competition and customers demanding a more personalised experience.

One big problem. If you cannot get a “Single Customer View”, across all channels how can you deliver a consistent personalised experience using analytics? 

What is the “Single Customer View” and why is this important?

The single customer view is a synchronized, consistent and accurate representation of the customer without duplication.

All records concerning a customer are linked back to a single and unique identifier. Data includes any personal details, interaction points, complaints, campaign responses, written communication etc. All of these data points enable analysis of past behaviour in order to predict future behaviour. The outcome is a business being better able to cater for customers hence improving the Customer Lifetime Value.

Putting the cart before the horse

Customer Analytics is dependent on a very basic concept, the customer. The identification of a single customer underpins any customer analytics from simple dashboarding of a customer’s touchpoints over time to complex micro-segmentation analyses. A single customer view must be robust and able to link to the incessant barrage of incoming data points prior to any analytics work providing meaningful insight. Problems start surfacing otherwise

Frustrated staff without a complete view of their customer

Business staff who have a direct interface with customers need an accurate and complete analyses of the person they are interacting with. There is nothing more frustrating to a customer than to have to repeatedly give information across multiple channels and different departments. Also consider the impact of trying to cross-sell or up-sell during a customer call without knowing that the last interaction with that client was relating to a complaint. I acknowledge that in some businesses, particularly finance, regulatory constraints prohibit the sharing of customer details between departments and this is a valid barrier but not in the majority of cases. 

Unconfident management scrutinising reports and not getting correct insights

When duplication makes it hard to count the number of customers in a certain process, reporting and reconciliation become an ongoing problem. This is exacerbated when consecutive reporting shows unexplained changes in numbers. Reconciliation demands very granular customer level data and when a single customer cannot be discerned, the comparison process becomes unclear on what the basis for comparison is. 

Unhappy and distrusting customers

A number of important logical business processes use customer statuses, details and interactions to determine an action or treatment. When these processes are built under the assumption that each customer is not duplicated, they could lead to a business treating a customer in a way in which it did not intend to, for example, sending a letter twice or sending entirely wrong communications. This can confuse a customer and lead them to be suspect of a company’s internal workings.

Why do companies have broken “Single Customer Views”?

A poor or non-existent “Single Customer View” can be caused by a variety of factors, some related to the applications though which data is entered or generated and some related to the way that the data is handled afterwards.

• Bad data capture – Reduced quality can be expected when systems are exposed to free-text and un-validated data entry. It is not surprising to see honest mistakes and sloppy attention with names, personal IDs, dates of birth etc.
• Bad customer identifiers – Some companies do not have customer identifiers and rely solely on the uniqueness of name, date of birth and address. This leads to confusion especially with customers with common names or customers with missing data
• System silos - The use of different systems in different departments for different processes. This often leads to chaos when it comes to combining customer data across the boundaries to create a harmonised view
• Age - many of the older companies have legacy data and systems which are difficult to integrate with newer sources of information
• System migrations - When an old system is being replaced with a new system, data is migrated from old to new. In some cases, both old and new systems continue to be used simultaneously during the transition without proper thought being put into how data is synchronised behind the scenes.
• Poorly designed matching routines - Some businesses use routines to try and match customers based on certain details, these routines are crucial to reducing duplication and the poor design of which can serve to match the wrong customers or to not match customers at all.

Getting closer to the “Single Customer View”

In my view, the “Single Customer View” should not be seen as a single one-time large scale project to gather any and every piece of information about each customer, but a progressive ongoing effort to pull in meaningful data only. A perfect system will always be a continuous struggle but a company can do a lot to reduce the probability of some simple failures. Here are a few;

Stop letting bad data in

1. Try to restrict the data type and content of any system that is exposed to human entry
2. Set up processes to continuously monitor any data that is entered and automatically alert the right people in your company to resolve
3. Monitor to a lighter degree machine generated information, incorrect coding can also lead to erroneous data being passed through

Link every interaction back to the customer

1. The creation of a single customer identifier is a fundamental accomplishment in any system. This identifier should be patronized by all departments and all systems. Where a single customer identifier is not achievable, multiple identifiers can be mapped to a single “Super” identifier. This would allow flexibility in catering to a customer with incomplete picture whilst continuing to develop a full view
2. Seek to identify any customer that comes into the sight of your company. Capturing at minimum an email address or mobile phone number to serve as a frame on which a customer’s details and interactions can be populated in future is better than capturing nothing at all
3. All points of interaction should have a connection back to the unique customer identifier

Improve your matching routines and seek to refresh customer data often

1. Create routine processes to synchronize data between systems. Make use of cloud based services and systems if these routines push your systems past maximum capacity
2. Not all matching is equal. All matching should be classified under levels of confidence and treated differently. As an example, low confidence matches should be manually checked and high confidence matches should be manually sampled at a lower rate.
3. Proactively seek to fill missing gaps by reaching out to customers or even using external agencies to reconstitute customer details

“Wow, I’m surprised you know that!”

“Yes, I do love buying red watches!”

“You know me!”

Happy Customer 2016

Happy customers are less likely to leave than angry ones. If you want to retain more customers, you need to personalise their experiences. This can be done once you have a connected view of who they are and how they have behaved in the past. Getting this view is a crucial step to uncovering meaningful predictive analyses and surprising your customers with an unparalleled level of satisfaction.