Comment testez-vous et expérimentez-vous différents modèles et scénarios CLV ?

1. Determine where your business is in its evolution 2. Ask yourself how sophisticated your data is and then how sophisticated your customers are. How long do they stay with you? What is the average lifespan? Do you have a transactional product with high turnover of users or do they stay for a while? 3. What is your customer acquisition cost? 4. Once you know how long people hang around, and what it costs to acquire a customer, then you can make estimates on the LTV. You can also then start setting goals. 5. Do not overcomplicate until you are ready. When its time to get complicated, hire someone who can manage this for you and provide the right suggestions on how to best proceed. 6. Listen to that person's advice...

Define Objectives: Clearly outline your objectives for CLV modeling and experimentation. Determine what specific questions you want to answer or insights you want to gain from the analysis. For example, you may want to understand the long-term profitability of different customer segments or optimize marketing strategies to maximize customer lifetime value. Data Collection: Gather relevant data on customer transactions, interactions, and demographics. Ensure that your data is clean, accurate, and comprehensive to support robust analysis. Depending on the chosen CLV model, you may need historical transaction data, customer attributes, and information on acquisition and retention costs.

CLV modeling is all about choosing the right approach for your business needs. Historical CLV, cohort-based CLV, and predictive CLV each have their own strengths and limitations. Consider your data availability, business objectives, and analytical capabilities when selecting a model. And hey, why settle for just one? Combine them for a truly comprehensive view of customer value!

Start with the C for Customer. Who is your customer and how do you define them? Can you tell how many customers you have? Imagine telcos offering services to households (internet & TV) and individuals (mobile). How do you combine these numbers? (answer: often not). Or banks: you're serving individuals that are also finance employees of large companies, shareholders, ... they have many different roles and should be served in different ways. It's good leadership to have a thorough discussion on who your customer is to make sure you can serve them.

Para testar e experimentar com diferentes modelos e cenários de CLV, uma abordagem alternativa eficaz seria implementar um processo de iteração contínua. Isso envolve selecionar um modelo de CLV inicialmente, definir inputs e outputs, e testar o modelo em um ambiente controlado. Em seguida, analisar os resultados e ajustar o modelo conforme necessário com base nos insights obtidos. Repita esse processo de teste e ajuste iterativamente, experimentando com diferentes estratégias e ações de CLV para identificar aquelas que geram os melhores resultados para sua empresa. Além disso, considere incorporar feedback de clientes e dados em tempo real para informar suas decisões e melhorar continuamente seus modelos e estratégias de CLV.

Testing and experimenting with CLV models is all about finding the right balance between precision and practicality. Start by clearly defining the inputs and outputs, but don't get stuck in analysis paralysis. Instead, use real customer data to create simple, actionable scenarios. Test these in a controlled environment and keep an eye on how changes impact long-term value. The goal is to build a model that not only predicts but also guides better decision-making. Remember, it's not just about numbers—it's about understanding your customers and making informed, empathetic choices.

From experience, defining CLV inputs and outputs is crucial for accuracy. Inputs typically include metrics like average purchase value, purchase frequency, and customer lifespan, along with acquisition costs. These factors provide the foundation for calculating the total revenue a customer brings over time. Outputs include the predicted lifetime revenue and profitability from a customer, which can be segmented by cohorts to refine strategies. Defining these inputs helps tailor customer engagement and retention efforts, ensuring the CLV calculation remains dynamic and aligned with business goals.

Every SaaS business should track logo retention (GRR) and net retention (NRR). Start there, and then go for the customer satisfaction metrics like CSAT or NPS. See if you can draw a correlation. After that, look at your customer base and those that renew and identify the product usage/features/license usage that results in high vs. low retention. Once you've determined that you can develop a rudimentary customer healthscore. You can then track the dollars going in and out, the customer feedback, and their product usage to get a 360 view of the customer. For the majority of SaaS businesses that will be enough to get you very far.

It's key to first identify what component of the customer journey needs to be a focus - i.e., is enablement missing? Timely updates to questions? Different models will inevitably focus on different components of the customer journey, all things being equal. Once the priority of a given experiment been decided, the path to backing into an updated model becomes more clear.

Frequency: The number of transactions or interactions a customer has with your business over a specific period. Average revenue per user (ARPU): The average amount of revenue generated by each customer over a specific period. Time horizon: The period over which CLV is calculated, such as one year, one month, or one week.

The only thing I would add to this is the testing never stops. You'll need to continue to do this at every stage in your evolution. And over time. Do not get stagnant or accept that what worked once will always work. Retention fluctuates over time. Hopefully you work for a business that is consistently growing and releasing new sellable product too so you have a chance to evolve.

Data quality assessment: Check the completeness, consistency, and accuracy of your data. Ensure there is enough historical data to support your model. Address any data issues or discrepancies before proceeding. Model performance evaluation: Use metrics such as R-squared, Mean Absolute Error (MAE), Mean Squared Error (MSE), or Root Mean Squared Error (RMSE) to assess how well your model fits the data. Compare the predicted values from your model with the actual observed values to identify any discrepancies or outliers. Validate the model's performance across different time periods or segments to ensure consistency.

Testing your CLV model involves validating assumptions and refining accuracy. Use historical data to compare projected versus actual outcomes, adjusting your model as needed. This iterative process helps identify discrepancies and improve predictive capabilities. Regular testing ensures your CLV model remains relevant and reliable, providing actionable insights for strategic planning. Consistent validation enhances the robustness of your projections, enabling more effective business decisions.

From experience, testing your CLV model and assumptions is essential to ensure its effectiveness. Start by running simulations using historical data to compare predicted CLV against actual customer behavior. Adjust for different scenarios, such as changes in customer retention rates, average order values, or acquisition costs. A/B testing can also help validate assumptions by comparing results from different customer segments or marketing campaigns. Regularly monitor and update your model based on new insights to ensure it reflects current business conditions and customer behavior trends.

Experimentation is critical with your customer base to find out what works best. But pay attention to the market, your competitors and any other outside influence that could impact how your customers experience working with you. When covid hit in March 2020, everyone had a much different set of needs on a global scale. Outside events don't need to be that big or cataclysmic, but they are a key factor. For example, I've noticed that Calendar year Q1 generally has a lower net retention rate for SaaS. That has been my experience at every business. Budgets are set at the end of the years and cuts are made then. If you have that experience, you can understand how to build out your targets/experiments moving forward.

Experimenting with different CLV strategies and actions is key to maximizing customer value. Segmenting customers based on value, potential, or behavior is a great place to start. Then, optimizing your marketing mix across channels and campaigns can help you reach the right customers with the right message. Personalization, value-added features, and soliciting feedback or referrals can also enhance customer experience. Just remember to track and analyze the results to refine your CLV model and assumptions. Boom! 💥

Explore different CLV models like RFM analysis, cohort analysis, or predictive models.Establish clear testing protocols for each scenario to ensure consistency.

Experimenting with different CLV strategies allows you to optimize customer value. Implement various retention and acquisition tactics, and measure their impact on CLV. Actions such as personalized marketing, loyalty programs, and improved customer service can significantly enhance customer lifetime value. Continuous experimentation and analysis help in identifying the most effective strategies, driving long-term growth and profitability. Embracing a dynamic approach ensures your business adapts to evolving customer needs and market conditions.

Experimenting the different CLV strategies is key to know what actions to implement in order to keep the performance high. Below are simple and effective ways to experiment your CLV strategies: ✅ Set clear goals: Define what you want to achieve (i.e. increase CLV or improve retention). ✅ Choose the right strategies: Upselling or cross-selling / Loyalty programmes / Improving customer service ✅ Segment customers: Test strategies on different customer groups to see which works best. ✅ Comparing the different strategies (e.g., loyalty program vs. no loyalty program) to measure effectiveness. ✅ Analysing the results you got: Track key metrics (CLV, retention rates, customer satisfaction) to see which strategy performs best

One of my companies reported on Net Revenue Retention (NRR) in an interesting way. Instead of using a percentage they used +/- the total MRR amount. It made sense for how their product usage fluctuated, and once explained it was very logical. That said, it made it incredibly difficult to measure the effectiveness of the CS org. So a consideration should not just be overall financial reporting and board reporting, but also how can you target/set goals for your team if they don't have an impactful and easily trackable way to understand their success. I've been at a number of orgs that have CSMs on quotas but no trackable dashboard available for them. That will not drive the right behavior.

At Sigma Infosolutions, we had a client whose initial CLV seemed average based on transactional data. However, this client was highly engaged, frequently providing valuable feedback. Recognizing their loyalty, we involved them in beta testing new products, offering a personalized experience. This strengthened our relationship, leading to increased adoption of our services and new client referrals from their positive word-of-mouth. This experience highlighted the importance of looking beyond data, focusing on building genuine relationships that can significantly enhance customer value.