Customer experience management is becoming increasingly crucial for businesses as customers have more choices and higher expectations. Two central problems facing customer experience management are low response rates on customer surveys and difficulty calculating customer experience projects’ return on investment. Artificial intelligence (AI) can be a valuable tool in mitigating or solving customer experience management’s two problems. AI can predict if non-respondents are detractors or promoters. It can also assign the right communication channel for the close-the-loop process and select suitable campaigns. A strong link between financial figures and Customer Lifetime Value, in particular, allows us to assess and predict the Return On Investment correctly. The whole process and methodology have been created and delivered in the sandsiv+ platform to enable business people to roll out quickly, gaining a significant competitive advantage.

Customer experience management is becoming increasingly crucial for businesses as customers have more choices and higher expectations. However, managing customer experience has its challenges. Two central problems facing customer experience management are low response rates on customer surveys and difficulty calculating customer experience projects’ return on investment.

Low response rates on customer surveys can be a significant problem for businesses trying to gather data on customer experience. Surveys are one of the most common ways businesses gather feedback on customer experience, but response rates can be low, making it difficult to gather a representative sample. This can lead to inaccurate or incomplete data, making it difficult to identify problem areas and track the success of customer experience initiatives.

Another common challenge is calculating the return on investment (ROI) of customer experience projects. Customer experience can be a subjective measure, and it can be difficult to gather accurate and reliable data. Additionally, the benefits of improving customer experience may take time to become apparent, making it difficult to calculate the ROI of CX initiatives. This can make it difficult for businesses to justify their investments in customer experience projects and make informed decisions about where to allocate resources.

Despite these challenges, businesses must continue to invest in customer experience management to stay competitive and retain customers in today’s marketplace. It’s crucial to find ways to increase the response rate on surveys and to find accurate ways to measure and calculate the ROI of customer experience projects.

Artificial intelligence (AI) can be a valuable tool in mitigating or solving customer experience management’s two problems: low response rates on customer surveys and difficulty calculating customer experience project return on investment.

AI can improve the survey process for low response rates on customer surveys and make it more engaging for customers. For example, natural language generation (NLG) can make survey questions more conversational, making them easier for customers to understand and respond to. Additionally, AI-powered chatbots can be used to conduct surveys in real-time, such as through SMS or messaging apps, which can increase the response rate.

In terms of calculating the return on investment of customer experience projects, AI can be used to gather and analyze customer data in real time. Machine learning algorithms can identify patterns and trends in customer data, such as purchase history or customer interactions, which can provide insights into customer behavior and preferences. This can help businesses understand which customer experience initiatives have the most impact and adjust their strategy accordingly. Additionally, AI can be used to predict customer churn, allowing businesses to intervene before the customer leaves.

This last aspect, in theory, can also be used to better measure KPIs such as NPS, but in my view it is not necessary. In fact, statistical techniques discovered decades ago make it possible to calculate exactly these KPIs with precise information of the margins of error and inconsistencies in the sample, so it is not necessary to bother machine learning, or deep learning techniques for this purpose. It is more useful to identify potential targets for close-the-loop or marketing automation to create actions on non respondents, create value, and prove ROI.

What we have done in the sandsiv+ platform was precisely that:

  • Create predictive machine learning models to identify customers who do not respond to surveys and try to predict whether they are promoters or detractors.
  • At the same time, it automatically interfaces with CRM or Marketing Automation platforms such as Salesforce, MS Dynamics, Adobe, and many others to make the information piece immediately actionable. The result is then saved to calculate the return on investment in customer lifetime value (CLTV).

Creating predictive models

To create predictive models, we decided to integrate into the platform Automated Machine Learning (Auto ML) libraries to make our client’s lives easier and create models automatically with the highest accuracy.

For this task, we integrated two auto ml libraries, one for machine learning and one for deep learning: Auto-Sklearn and AutoKeras. I will not go into the technical specificity of these two libraries; however, they are Auto ML based on one on the famous scikit-learn library and the other on Keras, a simplified and less verbose version of Tensorflow.

SANDSIV predictive analytics

When we created the solution in sandsiv+ we thought, as always, about business people who need to learn the technical notions of creating artificial intelligence solutions. From an intuitive graphical user interface, the business user can connect, transform, and generate training datasets and train machine learning models efficiently. All this in just a few clicks.

The methodology

The key to making this process accessible and immediately usable for our clients is to change nothing in the actual procedure. In the example below, 10’000 customers are invited to respond to a survey. Only about 3K respond, so how can we tell if the other nonrespondent customers are detractors, passives, or promoters?

SANDSIV predictive analytics

  • The sandsiv+ platform records the results of survey respondents and defines who is genuinely a detractor and who is a promoter.
  • The platform creates two predictive models: one to identify detractors and the other to identify promoters.
  • The two models are created using the Auto ML methodology fully automatically and without input from the platform user.
  • The metadata uploaded to the platform when the surveys are launched used to create the models.

Creating predictive models and checking their quality is automatic and requires no particular intervention by the user. There is a section where more experienced users can control various model creation parameters.

SANDSIV blog article predictive analytics

  • Auto ML then autonomously chooses the model with the best performance (precision, recall, and F1).

SANDSIV predictive analytics

  • The model is used to predict the status of nonrespondents: is he a detractor? Is he a promoter?

SANDSIV predictive analytics

In this specific case, 10’000 clients have been invited to the survey, and 3’000 answered the survey. Out of the 3’000, 1’510 have been flagged as DETRACTORS while 964 as PROMOTERS. The two predictive models have then identified out of the 7’000 nonrespondents, 524 DETRACTORS, and 1’660 PROMOTERS.

  • Detractors identified by the model are included in the close-the-loop process.
  • Promoters identified by the model are included in up-selling and cross-selling campaigns.

SANDSIV predictive analytics

  • The predictive models also decide which channel (agent or marketing automation) and which campaign is more relevant to the customer and generates a higher propensity to accept the offer.
  • A second survey measures close-the-loop and campaign activities to understand the impact on the customer.

SANDSIV predictive analytics
SANDSIV predictive analytics

  • The platform calculates CLTV on the individual customer measured before and after the activity to understand any improvement and the Return on Investment.

SANDSIV predictive analytics

In this case, we can see that the close-the-loop actions and the campaign actions on promoters generate approximately an extra 1.5 million in CLTV. The positive financial result to pass to C-level asking about the impact of CX activities on the P&L of the Company. This is possible because every next-best action activity is recorded in sandsiv+ and can be tracked in financial terms.


Customer experience (CX) management manages all interactions between a company and its customers to improve customer satisfaction and loyalty. In recent years, the use of artificial intelligence (AI) has begun to play a significant role in this process, leading to the evolution of CX management to customer intelligence (CI).

One of the key benefits of using AI in CX management is the ability to analyze and interpret large amounts of data in real time. This data can include customer interactions on social media, website behavior, and customer feedback. By analyzing this data, companies can better understand their customers’ needs and preferences, which can be used to improve the overall customer experience.

Another benefit of using AI in CX management is automating specific tasks. For example, AI-powered chatbots can handle customer inquiries, allowing companies to respond to customers quickly and efficiently. Additionally, AI-powered recommendations can be used to suggest products or services that a customer may be interested in based on their browsing history or purchase history.

CI also allows companies to personalize customer interactions. Companies can create personalized experiences for each customer by using AI to analyze customer data. This can include tailored marketing messages, product recommendations, and even customized interactions with customer service representatives.

As we have seen in our practical example, AI also allows companies to predict customer behavior. This is done by analyzing customer data and identifying patterns. Companies can anticipate their needs by understanding the customer’s behavior and taking proactive measures to improve the customer experience. This can include offering discounts or promotions to customers likely to purchase or providing customer service representatives with information about a customer’s history to help them better assist them.

Overall, the integration of AI in CX management is leading to the evolution of CX management to Customer Intelligence. By analyzing large amounts of customer data in real time, automating specific tasks, personalizing customer interactions, predicting customer behavior, and identifying potential issues, companies can improve the overall customer experience and ultimately improve customer satisfaction and loyalty.

In conclusion, integrating AI in CX management is revolutionizing how companies interact with their customers. It allows companies to gain a deeper understanding of their customer’s needs and preferences, which can be used to improve the overall customer experience. Additionally, AI allows companies to automate specific tasks, personalize customer interactions, predict customer behavior, and identify potential issues before they become significant problems.


Stay caught up in the race for Customer Intelligence. sandsiv+ can help you apply AI to your Customer Experience Management and stay ahead of the competition. Contact us now for a demo:

Say goodbye to low response rates and hello to ROI
Federico Cesconi

Read the article on LinkedIn

Start growing with sandsiv+ today