Study : Château Trigant

Чистка базы данных

В чем заключалась проблема нашего клиента?

Château Trigant has a customer data entry software. This software can track the status of the Château's sales. The tool takes into account a large number of data and different indicators.

However, the Château has a problem : the software does not allow to finely visualize all the data. This lack affects the sales department in setting up relevant actions towards the customer. To solve this problem, the Château has provided us the customers database, issued by their software, that had to be cleaned, made exploitable and able to supply necessary information for the decision making process.

Какое решение?


Data Classification:

We made a choice of classifying relevant data, in collaboration with the customer according to certain variables : date of purchase, wine type, quantity, etc.


Data cleaning:

We have cleaned the data according to 4 guide criteria using the statistical software R : customers individualization, data correction, etc.


Definition of hypotheses:

We have determined a classification of the data thanks to assumptions of making the database intelligent and commercially exploitable.

Focalisation client

With the last step of our process, we have classified the vintages in relation to the amount of customers who have purchased them.
A customer has been considered « regular » when he has bought at least one vintage 2009 and at least another vintage since the 2009. If we consider « Château Trigant » and « Magnum » : we were able, for example, to detecte that the vintage which has attracted most of the customers was the 2009.
This analysis is made even faster by setting up an intuitive dashboard that calls easily the database. When a customer calls, with just a click the person in charge can find the records of the customer's purchases. In addition, they have an alert which allows them to contact again a customer if necessary.

Чем это было выгодно нашему клиенту?

Château Trigant can now consult their data according to different classifications they have chosen. The data are cleaned and made reliable to provide an usable interpretation. Château Trigant had the opportunity to implement a customers relaunch system according to their habitudes and can anticipate the seasonality of their sales in a more precise way.

И на десерт...

10h11 did not resist to work on a forecasting model for Château Trigant's sales. Indeed, the record and the veracity of the data allowed us to elaborate a statistical model in according with the quality of the vintage. After analysing weather data, we have identified a strong correlation which has allowed us to predict the 2015-2016 sales for the Château.