Mastering Data Segmentation and Personalization: Boost Conversion Rates with User Personas and Targeted Use Cases

Context about the Data (Visitors), Segments, and Use cases 

Understanding website visitors' preferences is crucial for online businesses in today's competitive landscape. There are two main types of visitors: Users/Customers and Anonymous visitors. Businesses have focused on the customers in the past, who are ~2% of the total traffic on a brand’s website and have optimized the CX and conversion rate to the fullest. On average, 98% of the website traffic is anonymous, and marginal improvement in CX and thus conversion rate will have a disproportional improvement in overall brand conversions. To improve conversion rates, businesses need to understand visitor preferences of anonymous traffic. Despite 79% of consumers valuing personalization for brand loyalty, 77% feel businesses fall short in using personalization tactics effectively.

Businesses can collect data such as contextual data (location, referral source), browser data (user agent, IP address), clickstream data (pages viewed, links clicked), and user-level data (collected through logins, cookies, or opt-in forms) of anonymous visitors. 

This data allows businesses to personalize the visitor experience by showing relevant products, reminding them on-site about abandoned carts, and offering targeted discounts. Personalization can shorten the conversion cycle and lead to faster sales. 

To understand anonymous visitors’ preferences, businesses need to collect data across multiple visitor sessions, which involves recognizing the anonymous visitor across various sessions. Recognizing anonymous visitors across sessions requires a probabilistic model that calculates the probability of the anonymous visitor with certain parameters. If the likelihood of matching visitor parameters across sessions is high, models conclude that the anonymous visitors are the same. This technique is loosely called fingerprinting.1  Fingerprinting allows businesses to stitch anonymous sessions together. This data becomes the foundation for segmentation and creating user personas – fictional representations of individual visitors on your website within distinct segments.  

Lego leverages its understanding of customer behavior through 6 personas, ranging from superfans (Lead Users) to first-time buyers. They focus on the top 3 segments (Lead Users, Connected Community, and 1:1 Community. 

• The first segment, Lead Users, had highly engaged product enthusiasts. Lego adopted a co-creation strategy with them wherein Lego actively involves them in product design through online platforms.
    

• The second segment, called Connected Community, had known customers with deep brand connections. Lego interacts with them through online communities and social networks, providing personalized attention. 
  
  

• 1:1 Community, the third segment, comprises recent buyers who have visited Lego experiences. For them Lego tailors content to their interests based on purchases and Lego shop/park visits. 

This strategy has helped Lego capture a 7% market share and become the 4th largest toy manufacturer with a high growth rate and strong profits.

Segmentation and Clustering

To truly understand the audience, businesses must segment the audience into distinct groups. There are three main approaches to user segmentation:

• Static Segmentation: This traditional method relies on pre-defined criteria such as demographics (age, gender, location) or purchase history. It's straightforward but may miss nuanced user behavior. Example: A customer with 2 orders between 1 April 2024 and 30 April 2024. 
  
  

• Dynamic Segmentation: This approach utilizes more sophisticated techniques to create fluid user groups. We'll leverage data points such as user behavior (clickstream data), browsing patterns, and real-time interactions to segment our audience dynamically. Example: Customers with 2 orders in the last 30 days. Dynamic segments are different from static segments with respect to the rolling time window of 30 days.  
   

• Deep-Learning-based Segmentation: There might be situations where clear segmentation criteria aren't readily apparent. This is where clustering comes in. Clustering is an unsupervised learning technique that groups similar data points together without predefined categories. Clustering algorithms analyze your data and identify natural groupings based on similarities within the data points. This is particularly useful when you have vast user data and want to discover hidden patterns in their behavior. Ex: Clustering users based on purchase intent. A model based on data points such as browsing behavior, time spent on product pages, and mouse movements can predict purchase intent with high accuracy and group visitors with similar purchase intent together.

Segments can then be created at varying levels of granularity, ranging from highly specific representations of individual visitors to broad categories encompassing larger segments of users. This flexibility allows organizations to tailor their personas to suit their needs and goals.

Granularity of Segments

At the most granular level, a persona can represent a single visitor or user. Granular personas provide a deep understanding of individual user needs, pain points, and motivations, allowing for highly personalized experiences and targeted marketing efforts. However, segments at the individual visitor level become unmanageable and inefficiently expensive.

On the other end of the spectrum, broad personas represent larger segments or groups of users with shared characteristics. For example, a persona could be "Tier-I customers with iPhones," encompassing a significant portion of the user base that meets certain criteria, such as being high-value customers and using a specific device type. Such broad personas are also not useful as they represent an overly broad and diverse set of audiences that don’t have common needs and preferences.

Businesses need to employ moderately granular segments to reach the right set of homogenous audiences with common needs and preferences. The goal of both segmentation and clustering is to create mutually exclusive and collectively exhaustive (MECE) groups. MECE segmentation ensures that every visitor is placed into a distinct bucket or segment, ensuring correct targeting while maintaining consistency in communication and brand authenticity. It prevents wasted marketing efforts. Resources are allocated towards solutions that resonate with specific visitor groups, maximizing their impact.

Overlap of visitors - assigning priority within the use case or segment

However, it is impossible to rigidly follow the MECE framework since a visitor has multiple data points attributed to them, which may place them in multiple segments. Here's an example:

Imagine a streaming service business that wants to segment users based on their viewing habits.

Potential Segments:

• Movie Watchers: Users who primarily watch movies.

• TV Show Bingers: Users who tend to watch entire seasons of TV shows.

• Documentaries & Specials: Users who favor documentaries and special interest content.

A visitor might enjoy all three categories. They might watch a movie one night, binge a TV show the next, and enjoy documentaries on weekends. This visitor fits into all three segments, creating overlap.

Handling Visitor Overlaps across segments

Overlap can be handled using the following methods:

• Prioritization: If overlap does occur, establish prioritization rules to determine which bucket a visitor belongs to. Like in the above example, "Movie Watchers" is given priority while classifying users who watch more movies even if they also enjoy other content.
  
  

• Weighted Attributes: Consider assigning weights to different data points. In the above example, a user who watches mostly movies (70%) with occasional documentaries (30%) might still be classified as a "Movie Watcher" due to the higher weight.

Handling Visitor Missing Segments

There are situations where visitors might not neatly fit into any existing segment, creating "missing segments."

For example, there might be segments: first-time visitors, visitors who abandoned carts, and buyers. A visitor who has visited the product page multiple times but has not added anything to the cart and made a purchase yet might be missed in the segmentation. To deal with this, the following approach can be utilized:

• Catch-All Segments: Establish a catch-all segment or persona that captures visitors who do not fit into any of the defined segments or personas. As mentioned in the above example, if a visitor does not fall into a particular category, a new category called product viewers can be created that fills the gap of the previous segmentation.

Hypothesis at the root of Segments and thus User Personas

Segments are created based on the unique combination of data attributes, and this combination must have intrinsic hypotheses around visitor behavior.  For example, females aged 24-35 residing in Tier-I cities and carrying iPhones will also be good candidates for Premium Modern daily wear Silver Jewelry.

Based on the collected data, hypotheses are formulated about the potential characteristics, motivations, pain points, and preferences of different groups or segments of visitors. These hypotheses aim to capture the distinct problems or needs each segment faces. Using the hypotheses as a foundation, organizations create well-defined user personas that represent distinct groups of visitors with shared problems, needs, or behaviors. The entire process is centered around understanding the visitors through data-driven hypotheses and then developing tailored solutions (use cases) to address those specific issues or requirements for each segment or persona. By following this approach, organizations can ensure that their marketing efforts, product development, and user experiences are not generic or one-size-fits-all but tailored to their diverse visitor base's unique problems and needs. Segmenting your customer base and building personas also unlocks valuable insights. Take Metlife, an insurance giant, for example. By tailoring their sales process to the behaviors and needs of different customer segments, they achieved a staggering $800 million in annual savings targets.

Hypotheses should be based on the identified characteristics and the business’ understanding of the target audience. For example:

• Hypothesis 1: Males aged between 18 and 34 years from Tier-I cities with iPhones who have demonstrated an interest in sports shoes will likely purchase premium sneakers like Nike Jordans.

  
  

• Hypothesis 2: Returning customers from Tier I & II cities with a history of high-value purchases might be receptive to exclusive early access to new product launches

These hypotheses can now be used to create user personas, which will help online businesses tailor communication strategies that resonate with each distinct visitor. While forming hypotheses, it is important to remember that hypotheses should lead to testable actions that can be used to improve customer experience.  

Use Case Creation using Hypothesis

Once businesses have developed hypotheses for the different user segments and created user personas, the next step is to formulate specific use cases that align with those hypotheses. Use cases are scenarios or examples that illustrate how businesses plan to engage with each segment through various marketing and communication channels.

Here's an approach to forming use cases around hypotheses:

1. Identify Marketing Channels and Touchpoints: List the various marketing channels and touchpoints available to the organization. These could include pop-up ads, banner ads, email campaigns, social media ads, retargeting campaigns, push notifications, in-app messaging, and more.
   
   

2. Map Channels to Segment Hypotheses: For each user segment hypothesis, determine which marketing channels and touchpoints would be most effective for reaching and engaging with that segment. Consider the segment's characteristics, behaviors, and preferences outlined in the hypothesis.
   
   

3. Define Use Case Objectives: Establish clear objectives for each use case based on the segment hypothesis and the marketing channel being used. These objectives could include increasing brand awareness, driving website traffic, promoting specific products or features, encouraging sign-ups or purchases, or fostering customer loyalty and retention.
   
   

4. Craft Tailored Messaging and Content: Develop tailored messaging, content, and creative assets for each use case, aligning them with the segment's needs, motivations, and preferences described in the hypothesis. For example, if the hypothesis suggests that a segment values convenience, the messaging for a pop-up ad use case could highlight time-saving features or streamlined processes.
   
   

5. Define Success Metrics: Establish key performance indicators (KPIs) and success metrics for each use case. These metrics should align with the objectives and help measure the effectiveness of campaigns and the validity of the segment hypotheses.

Examples

1. Hypothesis: Special discounts and free shipping increase conversion rates among first-time visitors.

   
   User Persona: First-time visitor

   • Data Attributes: First-time visitor, browsing history (indicates interest in specific products), location (optional), device

   • Needs: Attractive prices, convenience, positive first impression

   
   Use Case:

   • Targeted pop-up ad upon first visit offering a welcome discount and free shipping on the first order.

   • Banner ad highlighting free shipping and friendly return policy.

   
   Metrics to Track: Conversion rate, CTR on the discount code within the pop-up, average order value
   
   

2. Hypothesis: Mobile users prefer shorter forms and streamlined checkout processes.

   
   User Persona: Mobile Visitor

   • Data Attributes: Device, Operating system, browsing behavior (indicates purchase intent), check-out funnel drop-off point

   • Needs: Quick and easy checkout experience, mobile-friendly interface

   
   Use Case:

   • Implement a mobile-optimized checkout flow with fewer form fields and auto-fill capabilities.

   • To reduce form-filling requirements, offer guest checkout or integrations with digital wallets (e.g., Apple Pay, Google Pay).

   
   Metrics to Track: Time spent on checkout for mobile visitors, mobile cart abandonment rate
   
   

3. Hypothesis: Returning visitors who previously viewed specific products are more likely to convert with targeted product recommendations or promotions

   
   User Persona: Returning Visitor

   • Data Attributes: Visitor status (new vs. returning), product viewing history, browsing behavior, cart abandonment data, and purchase history.

   • Needs: Personalized product suggestions, reminders about viewed items

   
   Use Case:

   • Automated email campaign triggered by a returning visitor viewing a specific product. The email showcases related products, highlights key features, and potentially offers a discount or limited-time promotion.

   • Push notifications featuring promotions or discounts on the specific products the visitor has shown interest in.

   • Implement product recommendation widgets or sections on the website, tailored to display products previously viewed by the returning visitor.

   
   Metrics to Track: Open rate, CTR on CTA mentioned in email, revenue generated from conversions influenced by these use cases
   
   

4. Hypothesis: Users spending more time on a product page are more likely to convert if they are presented with additional product information based on their browsing behavior.

   
   User Persona: Visitor

   • Data Attributes: Time spent on product pages, product page interactions (e.g., image zooms, video views), browsing behavior, and purchase history.

   • Needs: Personalized product suggestions, reminders about viewed items

   
   Use Case:

   • Implement dynamic content modules or recommendation widgets on product pages, displaying additional product information, guides, or related products based on the visitor's browsing behavior and dwell time.

   • Offer live chat or chatbot assistance on product pages for visitors who spend an extended time, to address any queries or provide personalized guidance.

   
   Metrics to Track: Conversion rate, video views, CTR on additional information links

Impact of Use Case definition with segmented audiences

Defining specific use cases and analyzing metrics for each use case can provide more granular insights and opportunities for optimization. The impact of the use case definition can be observed through:

• Targeted Optimization: With use of case-specific metrics, businesses can optimize user experiences, product offerings, and marketing campaigns for each specific scenario, addressing unique pain points and preferences of different user groups.
  
  

• Prioritization: By understanding the impact and potential of different use cases, businesses can prioritize their efforts and resources toward the areas that offer the highest potential for improvement and revenue growth.
  
  

• Hypothesis Testing: Use case-specific metrics to provide a foundation for testing hypotheses and validating assumptions around user behavior, preferences, and the effectiveness of various personalization strategies.

For example, in a use case that implements a “Compare Products” feature for visitors who browse through similar product pages and spend time on each product page businesses could track to see if time spent on product pages has lowered after implementing the feature, the number of clicks on the comparison feature, etc., whereas if for the business implements a “Top Picks” carousel for visitors exhibiting similar behavior as mentioned above they could track CTR on the product recommendation, number of unique product pages viewed by visitor. Ultimately the business could compare the add-to-cart rate and purchases to decide which use case to focus efforts on.

Costs Involved

Implementing this robust system of data collection, segmentation, and personalization to enhance user journey involves various costs, including:

1. Technology and Tools: Implementing data segmentation and use case analysis may require investing in specialized software or tools for data management, analytics, and personalization. These tools often come with licensing or subscription fees.
   
   

2. Infrastructure Costs: Investing in server infrastructure, storage, and computing power to handle large volumes of data and complex analytical workloads. Costs associated with cloud computing services (if utilized) for scalable data processing and storage. Maintaining and upgrading the underlying infrastructure as data and user demands grow.
   
   

3. Data-Related Costs: Expenses related to acquiring third-party data sources or enriching internal data with external data sets (e.g., demographic, geographic, or behavioral data) for more comprehensive user profiles. Costs associated with data cleaning, transformation, and integration from multiple sources. Cost for implementing data governance and security measures to ensure compliance with data privacy regulations (e.g., GDPR, CCPA). Utilizing website analytics tools, customer relationship management (CRM) software, and marketing automation platforms can incur subscription fees.
   
   

4. Training and Skill Development: Upskilling existing teams or hiring new personnel with expertise in data analysis, user experience design, and personalization can involve training and recruitment costs.
   
   

5. Marketing Automation Tools: Integrating segmentation data with marketing automation platforms can involve additional costs.

By bridging the gap between anonymous and known visitors through data collection, cleaning, fingerprinting, ID stitching, segmentation, clustering, and hypothesis testing, businesses can tailor user experiences and improve conversion rates. This process requires a combination of technical expertise, data analysis skills, and a deep understanding of customer behavior. Continuously refining and iterating on these techniques will enable companies to stay ahead of the curve.