It is widely accepted that superior customer experience (CX) drives revenue growth. Ever since Frederick Reichheld introduced the Net Promoter Score (NPS) in the 2003 Harvard Business Review, marketers’ focus has been seeing how customer experience can be captured. Often by managing the answer to the seminal question “On a scale of zero to 10, with 10 being highest, what’s the likelihood that you would recommend us (our company/brand) to a friend or colleague?”
For many, Net Promoter Score is the measure of consumer experience. 2003 was an ancient time before big data and AI. As machine learning advanced and advances, the question is: In a world where massive data about customer abound, is NPS defined in 2003 still relevant? Net Promoter Score is survey based, which is the root of its inherent weakness, a weakness that grows when comparing it to software-based measurements. Let’s explore those weaknesses a bit more:
NPS is easy to game, especially when primed by service staff
Most of us have been reminded that someone’s compensation or rating depends on your scores. From a service centre representative to a contractor, the subjective nature of NPS is obvious.
NPS does not scale and is biased
For a business with millions of customers and hundreds of touch points, sampling the entire set of customers and getting further customer information is not feasible. NPS is thus prone to the twin drawbacks of segmentation-based bias or self-reporting bias.
Segmentation bias comes from the very nature of selecting who gets a survey or who does not, to increase response rate. Unlike valkyries who select those who will live or die on the battlefield, most segmentation software is not divine in nature, even with an oracle behind it.
Self-selection bias arises when individuals select themselves, or where the characteristics of the people who select themselves create abnormal results. This is a reason why political phone surveys are getting more flawed by the cycle. Self-selection causes a non-response bias, when the groups of people responding have different responses than the group of people not responding. Customers who are indifferent or apathetic are less likely to respond. This means the middle range of the NPS is under-represented.
NPS does not value one customer versus another
Not all customers are equal. Some spend more, some influence more (at the office or on Facebook). The impact on a business of what I might buy, is very different from what you might buy. What you buy, your friend is likely to buy as well, but she should be treated uniquely
NPS is for relationships, not transactions
Asking people about recommending you is not about the most recent interaction. It is about the entire experience. So, you must be careful about when and where you ask the question. For example, if a regular customer loves your company, but his most recent experience is with a support rep was not stellar, the NPS score he will provide after the interaction will be lower because of that one employee and may not reflect the overall sentiment the customer has for the company. NPS is not about the most recent experience, it's about the overall experience, yet it is measured at only one or a few moments
NPS does not explain why the customer is satisfied
A business typically asks the obvious questions about perceived key events. If those questions are left open ended, the customer is likely to write about events she has strong memories of, while it might be the accumulation of perceived small events that drive a decision. It might very well be that the fact an oil change took longer to be done, drove a customer to rate the experience at a car dealer negatively, which translates into her shopping for a different brand of car at the time of her next purchase
NPS does not predict future customer behaviour, which limits the computation of ROI
NPS is non-committal. We must realise that correlation does not mean causality as other elements than just CX can be driving revenue growth
NPS is extemporal
Like any relationship, it goes through ups and downs. Even before making a purchase, a consumer has a sense of how he sees a brand. If they don’t, why do we have marketing? The decision to make a first purchase is part of an overall consumer experience.
Moving beyond NPS
Is there a better way to quantity customer experience? Yes. The latest AI driven technology systems can leverage the vast amount of customer data and transaction data, spanning sales, marketing and service feedback, to generate a more complete measure. 10,000 surveys are fine, but a billion interactions are much better.
At the core of my team’s thinking about consumer experience is the realisation that stating your opinion is good, but spending your opinion is better. You opine with your wallet. The latest CX technology can import data from both, corporate internal sources and external sources, to build and visualise comprehensive customer journeys for tens of millions of individual consumers. Flagging significant outcomes and monetary commitments. Utilising machine-learning modeling to quantify the impact of each event in the journey for every customer. Monetising every journey event and valuing them in local currency terms.
Dynamically valuing the impact of subsequent events in a customer’s journey; alongside valuing a customer’s commitment to a brand, and the products they use. This ‘value’ goes up and down with events. The higher, the higher the propensity to buy. Enabling a better understanding of what customers qualify for offers being made, before such offers are sent to the customer. With the impact of every event valued in local currency terms, the latest technology allows us to calculate the next-best-actions for individuals, giving a business the opportunity to optimise its marketing and sales campaign.
Interested in hearing leading global brands discuss subjects like this in person?
Find out more about Digital Marketing World Forum (#DMWF) Europe, London, North America, and Singapore.
