Case Study: How We Analyzed Data from a Leading Multi-National Retail Brand & Ionio's Strategies to Reduce Churn

Attracting new customers is essential, but retaining existing ones is what truly drives long-term business growth. So, how can you tell when customers are about to churn? More importantly, how can you stop them before they leave?

Recently, we analysed 2.1 million customer records from a million-dollar retail enterprise struggling with declining customer retention. Using AI and behavioural psychology, we identified opportunities to optimise their customer loyalty program.

In this blog, we’ll share:

• How Loyalty Programs work and its types
• How we approached the problem of customer churn for this business.
• The strategies we used to evaluate and optimise their LP.
• Practical takeaways you can apply to improve your own loyalty strategies.

Let’s dive in!

Why You Should Care

If you’re a business owner, decision-maker, or analyst, this blog is for you. If you:

• Struggle with choosing or optimising loyalty programs
• Are focused on improving customer retention strategies
• Want to learn how AI can streamline decision-making processes

🔗 Ready to make smarter, data-driven decisions? Contact Us

Here’s how you can apply these insights to your business, from boosting customer engagement to designing more effective loyalty strategies.

What Are Loyalty Programs, Really? And Why Do They Matter for Your Business?

Customer loyalty programs are structured strategies designed to reward repeat business and foster long-term relationships. But why are they so critical?

Every business operates within a life cycle: attract new clients → deliver products or services → provide aftercare → retain clients → adapt to market trends → repeat. Out of all these steps, retaining your customers is the most cost-effective way to build a stable business.

Retention vs. Acquisition

Here’s the simple truth:

It costs a lot less to keep your current customers than to find new ones.

According to eConsultancy, 82% of business leaders agree that retaining customers is easier and cheaper. Forbes backs this up by showing that the chance of selling to an existing customer is between 60-70%, while the likelihood of selling to a new customer sits at just 5-20%.

Why is this important?

1. Your loyal customers trust your brand.
2. They’ve experienced your service and know what to expect. This trust often translates into higher spending. Think about it: your loyal customers have already invested their trust in you. They’ve had positive experiences with your brand, and that builds a bond.
3. That emotional connection can do wonders, it makes them forgiving if something goes wrong and more likely to act as your advocates.

When you prioritise retention by improving customer experiences and showing customers you value them, the benefits are huge:

• Higher profitability because loyal customers keep coming back.
• Word-of-mouth growth, which is far cheaper than traditional advertising.

Tip💡 : Engage with your customers (either via feedbacks or personally) to understand their needs better and adjust your offerings accordingly. Make sure every interaction reflects your brand's promise and quality standards.

The Risks of Poor Design

Designing an effective customer loyalty program requires aligning business goals with customer needs. When this balance is disrupted, businesses risk disengagement, increased churn, and lost revenue. Poor design decisions can lead to wasted resources and missed opportunities. Understanding these risks allows organisations to act strategically and avoid common pitfalls.

Misaligned Incentives

Offering rewards that don’t align with customer preferences is one of the most frequent mistakes in loyalty programs. Rewards that are irrelevant, too delayed, or perceived as unachievable can demotivate customers and cause them to lose interest. According to Loyalty Reward Co., customers quickly disengage when they see little to no personal value in a program.

How to Fix This: Businesses should invest in customer research and personalisation. Tailoring rewards based on customer preferences and ensuring they’re achievable within a reasonable timeframe can drastically improve engagement.

Ignoring Early Churn Signals

Failing to recognise signs that a customer may be at risk of leaving can lead to costly churn rates. Early signals, like declining engagement, frequent drops in purchase behaviour, or reduced interaction with offers, are opportunities for proactive intervention. Research shows that ignoring these signals can increase churn rates by as much as 30%.

How to Fix This: Strategies backed with AI like predictive modeling and survival analysis can help companies identify these signals early. Taking action through timely communication or personalised offers can re-engage customers before they churn.

But here’s the good news: With the right analysis and insights, businesses can identify these pitfalls and design optimised programs.

This is where we step in.

Let’s first understand how these programs work in the first place.

How Loyalty Programs Work

At their core, loyalty programs incentivise customers to stick with your brand. Here’s how they deliver value:

• 🌟Rewarding Repeat Behaviour: By rewarding repeat behaviour, you create a cycle of positive reinforcement that encourages customers to continue engaging with your business. Customers earn points, discounts, or perks for repeat purchases or actions like referrals and reviews.
  • Example: H&M lets customers earn points for every buck spent, unlocking discounts as they accumulate making them feel appreciated and valued.
• This taps into the psychological principle of reciprocity, the idea that when someone does something nice for us, we feel compelled to return the favour. By rewarding repeat behaviour, loyalty programs create a sense of goodwill that strengthens the customer-brand relationship.
  How you could utilise this, you may ask? Remember to :
  • Offer small but meaningful rewards early in the customer journey to establish trust and goodwill.
  • Ensure that rewards are attainable and clearly communicated to avoid frustration or disengagement.

• 💖Building Emotional Connections: By creating a sense of exclusivity and value, loyalty programs foster emotional bonds.
  • Example: Sephora’s Beauty Insider program offers personalised product suggestions and birthday gifts, strengthening brand loyalty.
  • When customers feel emotionally connected to a brand, they are more likely to stay loyal—even when faced with competitive offers. Emotional connections make customers feel like they are part of something bigger than just a transactional relationship.
  • Make sure you use customer data to personalise rewards and communications (e.g., sending birthday discounts or recommending products based on past purchases).
• ‍📈Encouraging Higher Spending: Tiered systems motivate customers to spend more by unlocking better rewards as they move up. Customers progress through different levels—such as Bronze, Silver, Gold—based on their spending or engagement.‍
  • Example: Amazon Prime members spend more than double what non-members do annually, thanks to exclusive benefits.
  • The goal gradient effect suggests that people work harder as they get closer to achieving a goal. In loyalty programs, this means that as customers approach the next tier or reward threshold, they are more likely to increase their spending or engagement.You could utilise this too by :
    • Ensure that higher tiers offer meaningful rewards that justify the effort required to achieve them.
    • Introduce progress trackers (e.g., “You’re 20 points away from Gold status!”) to keep customers motivated.‍
• 📊Data Collection and Personalisation: Every interaction, transaction, or behaviour leaves a traceable footprint. Businesses can use this data to anticipate customer preferences, personalise experiences, and foster stronger emotional connections. This reveals why customers act the way they do—what motivates their purchases, what they value, and how they interact with your brand.
  • Loyalty programs gather valuable insights into customer behaviour, enabling brands to tailor experiences.‍
  • Example: Nike’s Plus Program uses data to offer personalised rewards and exclusive product access.

How Your Business Can Apply This: 💡

Here’s how your business can put these insights into practice:

1. Leverage Rewarding Repeat Behaviour (Reciprocity):
   • Reward customers with small, meaningful perks like discounts or free items early in their journey. This creates a trust-building foundation.
   • Examples: Discounts for repeat purchases, referral bonuses, or free products after 3-5 purchases.
2. Build Emotional Connections with Personalisation:
   • Personalise your loyalty rewards by using customer data to align promotions, birthday gifts, or product suggestions with individual behaviours.
   • Example: Sending customers personalised offers on their birthdays or based on their past purchase history.
3. Implement Tiered Reward Programs to Motivate Higher Spending:
   • Establish reward tiers with exclusive benefits at higher levels. Offer progress tracking tools to motivate customers as they reach these tiers.
   • Example: A rewards program that offers increased discounts or exclusive access as customers advance from Bronze to Silver to Gold.
4. Use Data Collection to Optimise Customer Insights and Personalisation:
   • Collect data from customer engagement to track behaviour patterns. Leverage AI solutions or CRM tools to analyze metrics and tailor rewards and communication strategies.
   • Example: Reward personalised behaviour insights, like offering exclusive early access to popular product launches.

By implementing these strategies thoughtfully, your business can create loyalty programs that are more than just transactional. They can become psychological motivators, personalised customer experiences, and emotional touchpoints that foster long-lasting relationships.

Types of Loyalty Programs: Which One Fits Your Business?

Selecting the right loyalty program depends on your business goals and customer base. Let’s break down the main types:

1. Points-Based Loyalty Programs⭐

Customers earn points for every purchase or specific action (e.g., referrals, social media engagement). These points can later be redeemed for rewards such as discounts, free products, or exclusive offers.

• Example: Starbucks Rewards ☕
  Allows customers to earn "Stars" for every dollar spent, which can be redeemed for free drinks or food. Starbucks Rewards members account for 53% 📊 of in-store sales in the U.S.
• Ideal For: Businesses with frequent transactions, such as cafes, retail stores, and e-commerce platforms. If your business thrives on repeat purchases, a points-based system can effectively incentivise continued engagement.

🛠️ Tip: Ensure the points are easy to earn and redeem to maintain engagement and simplicity.

2. Tiered Loyalty Programs🏆

Customers progress through tiers (e.g., Bronze, Silver, Gold) based on spending or engagement levels. Each tier unlocks increasingly valuable rewards and perks.

• Example: Sephora’s Beauty Insider program 💄
  Tiers like VIB and Rouge, offering exclusive events and benefits.
• Ideal For: Brands seeking to gamify loyalty and reward high-value customers. If your business benefits from creating a sense of exclusivity and competition among customers, tiered programs can significantly boost engagement.

Customers desire recognition and status within the community. The drive to reach higher tiers encourages increased spending and engagement.To implement effectively, clearly communicate the benefits of each tier and provide visible progress indicators that motivate customers to reach the next level.

💡 Tiers can increase ROI by up to 80% by focusing on high-value customers while motivating others to aim higher.

3. Cashback Loyalty Programs💸

Customers receive a percentage of their spending back as cash or store credit. This type of program directly ties rewards to spending.

• Example: Rakuten offers cashback on purchases made through its platform at partner retailers.
• Ideal For: Competitive industries where immediate value matters. If your market is saturated with alternatives, cashback programs can differentiate your brand by offering direct monetary benefits.

Customers appreciate instant rewards that provide clear value for their spending. This approach caters to those who prefer straightforward benefits over complex point systems.
💡 Success tip: Transparency in cashback methods ensures trust and satisfaction.

4. Subscription-Based (Premium) Loyalty Programs📦

Customers pay a recurring fee to access exclusive benefits like free shipping, discounts, or premium services.

• Example: Amazon Prime 🚚🎥
  Offers benefits such as free two-day shipping, Prime Video streaming, and exclusive deals for a subscription fee. Amazon Prime members spend an average of $1,400 annually, compared to $600 by non-members—more than double the spending.
• Ideal For: Industries where ongoing value can be bundled into a subscription package. If your business can offer multi-dimensional benefits across different service domains, this model can enhance customer retention significantly.

5. Value-Based Loyalty Programs❤🌱

Instead of offering tangible rewards, these programs align with customers values by donating a portion of their spending to charitable causes or sustainability initiatives. This builds emotional connections by allowing customers to contribute to causes they care about through their purchases.

• Example: Edgard & Cooper donates loyalty points toward animal welfare charities based on customer purchases.
• Ideal For: Socially conscious brands seeking to build emotional connections with their audience. If your brand emphasises ethical practices or sustainability, value-based programs can differentiate you from competitors while fostering loyalty.

📊 These programs can boost retention by up to 38%, fostering emotional connections and trust.

How to Measure and Optimise the Success of Your Loyalty Program

But how do you measure the success of a loyalty program?
A loyalty program's success isn’t just about how many customers sign up. True performance evaluation involves  metrics that indicate customer behaviour, engagement, and retention. Understanding these metrics allows you to identify opportunities for growth, make strategic improvements, and ensure your loyalty efforts align with customer needs.

Let’s break down the most important ones, with examples and actionable tips to improve them.

1. Enrollment Rate 📊 :The enrollment rate is your first indicator of how appealing your loyalty program is to potential members. It measures the percentage of customers who decide to join your program out of all those who interact with your brand.

A high enrollment rate suggests that your program is attractive and easy to join. Conversely, a low rate might indicate that the benefits aren’t clear or compelling enough, or that the sign-up process is too cumbersome.

Example Insight : This chart illustrates the percentage of customers who have joined your loyalty program over a given period. It helps assess the attractiveness and accessibility of your program. In January, the enrollment rate was 20%, indicating that 2,000 out of 10,000 customers joined the program. By March, this increased to 30%, showing improved customer interest and engagement

How to Improve Enrollment

• Simplify the sign-up process: Remove unnecessary steps that make joining feel like a chore.
• Promote with targeted campaigns: Use personalised marketing to show customers how your program benefits them directly.
• Offer small incentives for joining: Things like a bonus discount or points can motivate customers to sign up.

2. Engagement Rate 🎮  :Engagement rate measures how actively members participate in your loyalty program. This includes earning or redeeming points and participating in program-related activities.

High engagement indicates that customers find value in your program, while low engagement suggests the program might not align with their preferences or needs.Ensure rewards are meaningful and desirable and make participation fun and rewarding through gamification elements like challenges and milestones.

How to Boost Engagement

• Offer desirable, personalised rewards: Customise rewards based on individual purchase patterns and preferences.
• Gamify the program: Introduce challenges, levels, or milestones to make earning and redeeming points exciting.
• Make it interactive: Encourage customers to participate through surveys, polls, or user-generated content opportunities.

When customers feel engaged, they’ll stick around for the long term, increasing their value to your business.

3. Churn Rate 📉 :Churn rate reflects the percentage of members who leave or stop engaging with your loyalty program over a specific period.

A high churn rate can signal dissatisfaction, lack of perceived value, or insufficient engagement strategies. It's crucial to identify why members are leaving to make necessary adjustments.

Example Insight : This chart compares churn rates over different periods or among different customer segments, helping identify areas for improvement. While February saw a spike in churn at 15%, we've implemented targeted outreach strategies that reduced churn back to 10% by March.

How to Combat Churn

• Re-engage inactive members: Send personalised outreach, like exclusive offers, reminders, or personalised rewards.
• Gather customer feedback: Use surveys to discover why members lose interest and fix their pain points.
• Address dissatisfaction proactively: If you identify patterns, make quick adjustments to reward structures or communication efforts.

Churn can often feel inevitable, but with data-backed strategies, it can be prevented.

4. Customer Lifetime Value (CLV) 💰 :The total revenue a customer is expected to generate over their relationship with your brand.

CLV estimates the total revenue a customer will generate over their lifetime relationship with your brand. It helps you understand which segments are most valuable and where to focus your retention efforts.

Why It Matters: CLV provides insight into the long-term profitability of different customer segments, allowing you to allocate resources effectively.

How to Increase CLV

• Reward high-value customers with perks: Introduce exclusive benefits like early access, VIP rewards, or personalised offers.
• Targeted promotions for repeat buyers: Offer loyalty discounts or rewards for repeat purchases to encourage continued engagement.
• Segment customer behaviour: Analyse CLV data to segment and focus retention efforts on high-value customer groups.

A focus on CLV enables you to invest smarter keeping your most loyal and profitable customers happy.

How We Used AI to Predict Customer Churn & Loyalty—and How It Can Help You

We explored customer churn patterns within a major retail company’s data to uncover the root causes of declining engagement with their loyalty program. Through careful analysis, we focused on answering two critical questions:

1. Who is at risk of churning?
2. What actionable steps can help retain them?

Churn tells you when your customers are losing interest, feeling dissatisfied, or simply not finding value in what you offer. It's about,

• Disengagement
• Dissatisfaction
• Perceived lack of value

If you catch churn early, you can retain customers before they disengage completely. But predicting churn is no easy task.
Using AI, analysing vast amounts of customer data to identify patterns and predict churn risk gets easier.

To tackle this, we combined machine learning with psychological insights to analyse the company’s loyalty program performance.

Here’s a detailed look at what we did, why we did it, and what we uncovered.

Step 1: Data Exploration & Pattern Discovery

Before building predictive models, we began by thoroughly exploring the customer churn data to uncover underlying patterns and trends. Our goal was to answer critical questions about customer behaviour and identify the factors most likely to predict churn.

• How many customers churned vs. stayed?
• Which features (customer behaviours, demographics, transaction history) were most correlated with churn risk?
• How does customer tenure affect churn risk?

We used visualisation techniques such as bar graphs to compare churn rates across customer groups and correlation heatmaps to identify relationships between key features. These visualisations provided a clear view of patterns and trends, guiding our modeling and analysis strategy.

This showed us that with time (in months) the churn rate seems to decrease. Customers who have been for around for 24-36 months have little to almost no customer churn patterns. Might suggest modifications in the current loyalty program.

Step 2: Predicting Churn Rate (Random Forest)

We chose Random Forest as our primary prediction model for its reliability in handling real-world customer data. Unlike traditional linear models, it excels at capturing complex, non-linear relationships between customer behaviours and churn likelihood.

Random Forest works by building multiple decision trees from the data and combining their outputs. This ensemble approach minimises the risk of overfitting, ensuring that the model performs well not only on training data but also on unseen customers. Each tree in the forest considers random subsets of features—customer engagement levels, reward redemption habits, tenure, etc.—which makes the model robust, even when certain data points are noisy or incomplete.

More importantly, Random Forest doesn't just give predictions; it offers explainability. For instance, it ranks customer attributes by their contribution to churn prediction. Through this, we discovered that factors like early reward redemption and customer tenure were strong indicators of churn risk. Customers who failed to redeem any rewards during their first few months were significantly more likely to churn, highlighting the importance of early engagement.

This gave us a clear picture of who to focus on and why they might be disengaging, enabling us to target interventions more effectively.

Feature Engineering & Model Training

We cleaned missing values, imputed numerical data, encoded categorical variables, and selected key features to train our model.‍‍

We analysed which features influenced the model’s predictions the most. This insight tells businesses which customer behaviours matter most in predicting churn.

This visualisation highlighted key behaviours like customer tenure, activity levels, and demographics as leading indicators of churn risk.

Step 3: Survival Analysis with Kaplan-Meier

What is Survival Analysis?

While Random Forest told us who might churn, we still needed to answer a crucial question: When does churn happen most often? For this, we turned to Survival Analysis—a statistical method traditionally used in medical research to study the "time until an event occurs." Here, the event we analysed was customer churn.

We applied the Kaplan-Meier Estimator (as seen in the graph below), a non-parametric method for estimating the probability that a customer survives (remains loyal) at any given time, given their tenure. The key idea is:

How likely is it that a customer continues to stay with a company/service over time?

time, survival_prob, conf_int = kaplan_meier_estimator(
    edata["Churn"].values.astype(bool),
    edata["Tenure"].values,
    conf_type="log-log"
)
plt.figure(figsize=(10, 6))
plt.step(time, survival_prob, where="post")
plt.fill_between(time, conf_int[0], conf_int[1], alpha=0.25, step="post")
plt.xlabel("Tenure (months)")
plt.ylabel("Survival Probability")
plt.title("Kaplan-Meier Survival Estimate for Customer Churn")
plt.legend()
plt.show()

The curve represents the probability of customer survival (or retention) over time, with survival defined as a customer continuing to engage with the business.

• At the beginning, the survival probability is high, close to 100%. This makes sense: most customers start their journey positively.
• However, the curve declines steadily as time progresses, showing a gradual increase in churn. By the time customers reach a tenure of 10 months, the survival probability drops to roughly 80%. By a tenure of 20 months, it falls further to around 40%.

This declining trend highlights a key observation: early churn is the most critical issue. The drop in survival probability during the initial period suggests that customers are at the highest risk of disengagement early in their journey. If businesses fail to engage customers effectively during these early phases, retention becomes increasingly difficult later.

Why Do These Drop-offs Matter?

1. Early Drop-Offs (0-5 months):
   • Suggests that a majority of customers leave because they fail to see value early on. This could indicate a need for better user onboarding, communication, or personalised first-time user experiences.
2. Mid-tenure churn spikes:
   • Some churn happens mid-way if customer needs are no longer being met. This could signal that product features, pricing, or expectations are misaligned with customer behaviour.
3. Later drop-offs:
   • These may reflect long-term dissatisfaction, shifts in external conditions (such as a customer finding a competitor's service more attractive), or disengagement.

What Insights Can You Take Away?

The combined use of machine learning and survival analysis helped us:

1. Target interventions at high-risk points: For instance, improve the onboarding experiences in the first 0-5 months as well as introduce loyalty incentives for customers approaching the mid-tenure risk window.
2. Optimise engagement strategies: Understanding these patterns allows personalised retention strategies, such as sending custom discounts or educational materials to customers during key churn risk windows.
3. Anticipate resource planning: Businesses can use these insights to allocate resources efficiently, investing time and money where churn risk is most likely to escalate.

How We Applied Psychology to Influence Customer Retention—and How You Can Too

To design a data driven loyalty solution for the company, we analysed key psychological principles that influence customer behaviour. These insights reveal why customers engage, stay loyal, and build long-term relationships. By understanding these principles, other businesses can create impactful loyalty programs that resonate with their customers.

1. Reciprocity: The Give-and-Take Principle

Reciprocity is a simple concept - when customers receive a gift, reward, or benefit, they feel a natural obligation to return the favour. This creates a cycle of goodwill that strengthens the customer-brand relationship.

Loyalty programs use reciprocity by offering small rewards like discounts, free items, or surprise gifts. These gestures make customers feel valued and prompt them to reciprocate through repeat purchases or engagement.

What We Did for the Company: Using the company data, we identified the critical time interval to maximise reciprocity. By offering small yet meaningful rewards (e.g., discounts, free items), the company could trigger this psychological principle and foster a cycle of goodwill and loyalty.

Learning for Your Business: Small, unexpected rewards can significantly enhance customer engagement. Identify key customer touchpoints—like purchase milestones or special occasions—and deliver timely gestures of appreciation.

Example: Offering a surprise discount after 5 purchases encourages customers to return, knowing their loyalty is rewarded.

2. Endowed Progress Effect : The endowed progress effect demonstrates that customers are more motivated to engage when they feel they’ve already made progress toward a reward, even if that progress is artificially introduced.

What We Did for the Company: Through survival analysis, we identified a critical window during which customers were at the highest risk of churn. To combat this, the company offered initial bonus points or “head start” rewards, creating a sense of momentum and motivating customers to engage with the loyalty program.

Learning for Your Business: Reward customers early in their journey. Whether it’s bonus points, free sign-up rewards, or introductory benefits, creating a sense of progress early on increases retention and engagement.

Example: Offering new customers 100 bonus points upon joining a program gives them an immediate incentive to continue.

‍3. Goal Gradient Effect : The goal gradient effect suggests that people work harder as they get closer to achieving a goal. In loyalty programs, the closer customers get to earning a reward, the more effort they put into engaging with the program.

‍
What We Did for the Company: By analysing customer purchase data, we identified the point where engagement increases—typically just before reaching a reward milestone. Timely nudges, such as “You’re 1 step away from your reward!”, were implemented to push customers toward completing their goals.

Learning for Your Business: Structure loyalty programs with clear milestones, and actively monitor customer progress. Send personalised reminders or incentives to keep customers engaged as they near their rewards.

Example: A program rewarding every 5 purchases can see higher engagement after the 3rd or 4th purchase when reminders are strategically sent.

4. Personalisation : Customers value unique, personalised experiences that make them feel special and understood. Personalisation enhances loyalty by showing customers that the brand recognises their individuality.
Loyalty programs that offer rewards and options based on individual preferences are more likely to engage customers. These can include personalised promotions, exclusive product recommendations, or customisable reward choices.

What We Did for the Company: We used customer data to personalise loyalty program rewards and communications. By analysing purchase history, preferences, and engagement patterns, the company could offer targeted rewards and personalised promotions that resonated with individual customers.

Learning for Your Business: Leverage customer data to create tailored experiences. Personalised rewards, recommendations, and offers show customers that you understand their needs, leading to stronger loyalty.

Example: A clothing retailer sends personalised discounts for specific product categories that align with a customer’s past purchases.

Ready to Predict and Retain Your Customers?

We’ve shared insights, research, models, and practical findings from leveraging AI in customer loyalty strategies. These insights can translate into better retention strategies, better customer experience, and better profits.

Let’s make AI work for you.

👉 Contact Us today to explore how we can tailor AI insights to your business.

We believe AI can empower every business—big or small—to make strategic, data-informed choices. Don’t wait to find out what your customers are trying to tell you.

‍