top of page
  • Writer's pictureGiulia Di Lorenzo

Strategic analysis of CRM data: a business case

In management studies it is widely accepted that information represent a critical success factor for companies that invest, therefore, in technical and organizational structures to improve their ability to collect, store, process and share data and information.

From a company point of view, knowledge obtained from data can be useful in order to have a more objective view of itself, to know customers and competitors, to analyze operations, to identify strengths and possible improvements, to maintain in-house knowledge, regardless of people.

Among the points listed above, a key aspect for a business is the knowledge and relationship with the customer.

These aspects are relevant especially in Business-to-Business companies where each relationship is unique and important (the number of customers is relatively limited compared to Business-to-Consumer companies) and therefore it requires a specific evaluation.

The value of the relationship with the customer goes far beyond the transaction: the long-term performance of a company depends on its ability to properly manage the relationship with customers, as well as with suppliers and other counterparties.

The tool that allows to manage the relationship with the customer is CRM: Customer Relationship Management.

The adoption of a CRM system has positive effects both outside the organization on customer satisfaction and inside the organization on the convergence of many company functions (in particular marketing, sales, customer service, IT) on a single customer management strategy.

Criticisms of CRM are often based on a rigid and hierarchical corporate culture that implies the perception that CRM represents a limit to the individual freedom on the job (CRM as a control system) and that it is an excessively time-consuming tool for its outputs.

CRM data could provide interesting information for a business. They can be used, for example, to identify the characteristics of products/projects that are the most successful on the market.

This article treats the analysis of data of the CRM of a company operating in B2B market. The analysis has been carried out by Giulia Di Lorenzo.

For confidentiality reasons, neither the company nor the specific results obtained will be mentioned.

The research question was "what are the characteristics of successful projects?"

Answering this question means to understand where the company could intervene to increase the chances of success and when it should change its strategy allocating its resources differently.

In the dataset, each statistical unit corresponds to a row of an Excel file and it is referred to a project as an opportunity of business.

Each column reports a specific feature of the project. 12,245 projects were considered in total. In order to respond to the research question the most important project’s attribute concerns the final status (project won, lost or still in progress). Other relevant columns relate to the name, the headquarters’sites and the industry in which the customer operates, the type of product offered, the channel that led the opportunity.

The work on the dataset began with a preparatory activity of dataset cleaning.

For each variable (column), all the modalities were identified and then their respective frequencies were calculated. The threshold of 90% of filled fields was chosen: the variables presenting more than 10% of empty fields were discarded because they could not be statistically evaluated.

The preparatory activity of the dataset led also to the reduction of the number of the projects in order to exclude those once with anomalous data (following management’s guidelines).

In order to answer the research question, the work was divided into three parts: univariate, bivariate and multivariate analysis, carried out using Excel and SPSS.

The univariate analysis was the preliminary tool used to study each variable without seeking any relationship with the others. Descriptive statistics was applied (indicators and graphs).

The bivariate analysis studies pairs of variables looking for associations. The variable that was included in all pairs was the one that expresses how the project was concluded (won or lost). In fact, for the management it is interesting to understand how to increase the chances of success and how to achieve challenging business objectives.

The bivariate analyses were carried out using the contingency tables, the chi-squared statistic and the Pearson contingency index.

Considering the object of the analysis as a sample of the universe represented by the Company's past, present and future projects (hence, not as a population), a statistical test was performed.

The two hypotheses were:

H0: absence of association between the two variables,

H1: presence of association between the two variables.

To answer the test, p-value was compared with the usual values ​​of α (0.05). For p-value less than α, the null hypothesis was rejected.

Furthermore, since the chi-square statistic increases as the size of the sample increases, the Cramer’s V was calculated to provide information on the degree of association.

In the present business case, an association between the variables resulted. Consequently, clear indications for management were derived: some areas, industries, channels, products resulted to be more associated with the closing of the deal.

The multivariate analysis, on the other hand, studies several variables simultaneously.

Being interested to the determinants of a successful project, a logistic regression analysis was performed. This tool allows to identify the determinants of the occurrence of the mode of a dichotomous variable and, in this business case, what are the characteristics of a project that seemed to affect its outcome the most. If the p-value is greater than 0.05, the mode is not significant in the model. Odds ratios were also evaluated.

In the business case, there were few modalities that were not significant for the model.

To evaluate the goodness of fit of the model, the Receiver Operating Characteristic (ROC) curve relates the sensitivity (the rate of true positives, corresponding to projects that were actually won and that were classified as won by the model) and specificity (the rate of false positive, corresponding to projects that were lost, but that were recognized as won by the model).

In this business case, the ROC curve showed that the model was able to make moderately accurate predictions, in fact the Area Under the Curve (AUC) was greater than 0.7.

Figura 1 Output R: ROC e AUC

The discussion with the management confirmed the associations found by the study but new perspectives and promising ways for the business were also outlined. In addition, ideas for subsequent analysis emerged.

To enhance the potential of CRM, a similar analysis should be carried out in every division of the company. It would be extremely interesting for the company at the corporate level to collect the various analysis and to examine them both in a comparative way and as part of a whole. In fact, strengths and weaknesses of the overall business could emerge, while appearing scarcely relevant at the micro level.

It would also be interesting to compare the results across divisions to identify the best practices of each of them and so to extend them to the others.

Furthermore, being aware of the numerous functions of a CRM system (data storage, forecasting, alignment between company functions,…) leads to the effort of making the most of it.

Therefore, it appeared useful to highlight the strengths and weaknesses of the company's CRM system.

Undoubtedly, having developed its own system internally makes it more flexible and thus it allows for easier improvements. It follows the sales process step by step, although it should be updated more frequently by the operators for a greater effectiveness.

In addition, in order to avoid compilation errors, it would be useful to provide a closed list of fields to be selected and a range of values ​​in order to include the data to be entered.

The characteristics of the projects (columns) included in the study were a small group compared to those contained in the CRM. The company’s CRM aims to collect very detailed data, but many fields usually remain empty. Therefore, it seems useful to delete some columns in order to make the users more inclined to enter the data and to update them often.

It could also be interesting to collect data regarding the project and the relationship with the customer after the deal is closed. In fact, it would be useful to investigate customer satisfaction and the reasons for the recourse of customer service. And also, new functions aggregating data by industry, area or customer could allow a global view of the company positioning from which to start to develop future projects.

The limitation of the analysis derives from the incompleteness of data contained in CRM. Some attributes were not taken into consideration in the analysis because the corresponding fields weren’t filled in in many cases. In addition, the sales process may disclose other information that didn’t fall within the present CRM data scheme such as data on any history with a particular customer. The analysis could have been more complete having more information.


bottom of page