We continue our series of articles about personalized cigar recommendations and what we consider to be the key factors for a personalized service. We also share with you some of the results of our latest customer satisfaction survey…
Science vs. Pragmatism
If you have read us before, you know that we operate as a lean-agile team. In alignment with our continuous improvement remit, we recently conducted a non-scientific survey among our members to understand where we’re doing well and where we can further improve our service. The areas we selected are the ones that we consider to be key success factors for our personalized service:
We thank our members for the terrific feedback we received on all these fronts.
Today, we’ll focus on the transparency aspect.
What is Transparency?
According to the study “Recommendations as personalized marketing”, the customer will be satisfied if there is a convincing connection between the recommendation and her/his true preferences. The algorithm is strictly related: if it is perceived as superior, it is one that adopts a customer-centered approach in learning and inferring customer preferences from the customer’s ratings, reviews and other data, to present a true picture of the customer. On the other hand, the customer will be dissatisfied if she/he perceives an algorithm as inferior, which usually occurs when a service adopts a company-centered approach, inferring preferences solely from the company’s internal data such as search and purchase history. This often yields a fragmented or incomplete picture of the customer, and leads to a poor quality of recommendations.
In short, the more a service is transparent to a user, the more she/he will perceive such service as trustworthy and will enjoy using it. Transparency allows users to meaningfully revise their input in order to improve recommendations, rather than making “shots in the dark” with unconvincing, hard to understand or even bizarre connections (see the Amazon backpack anecdote).
Feedback is critical to allow transparency. It can be viewed as a request to put more effort in the process, with the reward being improved quality of the recommendations.
How do we deal with Transparency at Cigar Sense?
Our users are asked for feedback according to the steps below:
The first 3 steps for free users are part of our on-boarding process. The on-boarding by itself gives some insight into the parameters used to formulate personalized recommendations.
Users can create (or update) their gustatory profiles based on the various aromas and tastes/mouthfeel that they know they liked or disliked. Or, if they don’t feel comfortable identifying detailed aromas and tastes manually, they can indicate specific cigars from our database which they know they like, and Cigar Sense will infer a gustatory profile for the user based on them. Our users’ personal profile data is available and modifiable at any time, no matter how it was originally derived. Also, after each change made by the member in her/his profile, the service provides a revised list of recommended cigars right away, in real time.
The image above illustrates the actions that our users take in a progressive way towards the goal of obtaining accurate recommendations. Needless to say, paid members who extensively use the service are extremely likely to obtain spot on recommendations.
How do our members rate us on Transparency?
We asked our members:
As I complete the Cigar Sense on-boarding or I update my profile, it is easy for me to understand based on what information Cigar Sense suggests cigars to me and can I see how my preferences drive the suggestions for new cigars?
We’re very pleased that our members rated the Cigar Sense service as 8.7 out of 10 for Transparency. In fact, until a few months ago, most free users could not manually adjust their gustatory preferences (aromas and tastes/mouthfeel), as they were strictly determined by inference only . We’ve given everyone that capability now and look forward to gaining their further feedback as we continuously improve our service. Stay tuned!
