My friend Bob Glushko and Lindsay Tabas wrote an interesting paper last year that Bob pointed out to me earlier this month – Bridging the “Front Stage” and “Back Stage” in Service System Design. I liked the paper as it raised some interesting issues about customer interactions, in particular whether interaction design or outcomes is the most important factor in determining a customer’s satisfaction after the interaction. This matters a great deal when focusing on the customer experience as one must know what matters to customers if one is to improve the overall experience.
I won’t repeat too much of the paper here but one of it’s key points is that service quality, or the perception of service quality, can be the result of many things – how personalized the service is, how “intense” the service offering (how much goes on during the interaction) and how variable it is (this tends to pull against personalization but unintended variability is bad). However it does not matter how focused an interaction is on quality if customers don’t get the result they want. Results matter – after all if the guest can’t check in or the product is not available then the customer will be upset no matter how good the interaction felt. Additionally, simple and predictable transactions may well be just want a customer wants (think ATM or other self-service environment). Customers may prefer a simple, predictable transaction to a highly interactive “intense” one that has more variable outcomes. This is complicated by the multi-channel reality of most organizations today. After all, people have different expectations of machines and will make trade-offs between automated and manual channels depending on the relative experience and success rates of using those channels. The paper then goes on to discuss the way in which the “front stage” interaction and the “back stage” information systems inter-operate to deliver on quality interactions. They make the point that predictable results matter and that “variability is the enemy” because variability, especially negative variability, is hard to recover from. Variability caused by problems “back-stage” – things like missing reservations or out of stock products – is particularly hard to recover from no matter how careful the interaction design – it’s hard to make a customer feel good about a transaction that failed. They also make the point that the traditional “moment of truth” in a customer interaction is when something becomes apparent to a customer. In reality this may not be when it becomes inevitable – some prior process or failure, some “back stage” event may mean that the interaction is going to “fail” no matter what is done in the interaction. Customer experience is thus best considered a process quality issue.
In the terms this paper uses, Enterprise Decision Management or EDM is about building “back stage” automation of decisions that support successful, high-quality “front stage” interactions. In particular EDM helps address the key question of “if every experience has to be different, how can our implementation be roust and scalable?”. Personalization in this sense does not have to mean a completely different interaction for each person, merely one that feels different and that is appropriate. It is true that people expect less of a machine so automated interactions need not be so personalized – after all, people typically choose automated/self-service channels for their predictability. There is, however, no reason why this kind of interaction cannot be improved also.
If we want to provide an intensely personalized experience, what should we do:
- Always apply customer preferences that you have
- Don’t wait for a customer to tell you something is a preference, infer it as soon as you can and let them correct you if you are wrong
- As soon as you know someone is in a segment, apply the preferences you believe that segment has and allow them to override those easily
- Capture preferences from users but don’t make customers hand them all over at the start – allow them to express them as they go.
- Think about information that would make the customer experience better and see if you can infer it from their prior behavior or that of customers like them. Think about your staff and see if they could capture it as they interact with customers.
- Consider if data is worth enough to you to justify purchasing it from outside sources
Making this work involves data capture, data management, analytics to infer preferences from prior behavior and rules specified by users as well as by business experts.
The paper talks about the power of multi-disciplinary teams in solving these kinds of problems and I should reiterate the value of business rules as a declarative specification of logic that can be shared between technical and non-technical folks. This is, at least in part, what the authors mean by model-based interaction design.
Lastly, check out this series of posts on customer experience.