Metisa's insights are modelled after decisions marketers instinctively make each day about a customer:
- How valuable is he/she? (Customer lifetime value)
- How loyal is he/she? (Churn rate)
- Are there many others like him/her? (Customer personas)
- What products does he/she like? (Recommendations)
These insights tell you what and when to send to each customer to maximize the likelihood of sales conversion. Our algorithms are predictive. They perform significantly better than traditional methods.
Customer lifetime value
Customer lifetime value (CLV) is the financial value of a customer in your store. It is the sum of future sales expected from a customer discounted to the present value. Our predictive model is based on your sales history and takes to account the nature of your industry as well as how your customers shop.
Traditionally, marketers use total sales from a customer as a proxy for customer lifetime value. Our predictive model is more accurate as it is forward looking and is based on your sales history. For instance, a customer who is new but a frequent spender could be more valuable than a inactive customer who used to buy a lot.
In customer acquisition, CLV is compared to the cost of acquiring a customer (CAC). For instance, to have a positive return on customer acquisition, a brand might want to have CAC < 0.25-0.5 × CLV. This comparison is done by channel (e.g. Facebook, Google Adwords etc.) to figure out which are the most cost-effective channels to acquire customers.
In customer retention, the goal is to maximize CLV of your existing customers. This is done by focusing on key lifecycle transitions, such as converting a one-time to an active customer, or winning back an at-risk customer. The more active a customer is, the higher his/her CLV.
Churn rate is the probability that a customer will not make a purchase in the next 24 months. Values near 0 means that the probability of a customer leaving the store is very low (active or loyal customers). Values near 1 means that the probability of a customer leaving the store is very high. We classify customers with churn rate between 0.5 and 0.9 as at-risk and greater than 0.9 as lost.
Metisa predicts the distribution of the time of next purchase for each customer. We do this by fitting the time of next purchase based on factors relating to purchasing history such as purchase frequency, first and last purchase date.
A customer persona is a semi-fictional representation of your ideal customer based on market research and real data about your existing customers.
Traditionally, the process of coming up with customer personas is a mix of gut feel from touch points with customers, research reports and creative agencies.
We take a data-centric, bottom-up approach to figuring out who customer personas should be. We group existing customers with similar buying patterns and interests and figure out which way of grouping them produces the best fit. Combining this with our predicted customer lifetime value analysis, we know how valuable each of the segments are as well as how valuable a customer in each segment is.
To learn more, read our blog post.
Recommendations are the core of creating a personalized experience. Metisa generates recommendations for each customer based on what similar customers bought. Product recommendations allow you to create content that is tailored to every customer, resulting in 2-5x increases in email click rates.
To learn more, read our blog post.
Why are my insights not available?
Predictive insights are forecasts made based on historical data. An insight is not available when we need more data to produce a reliable result. This could occur with customer lifetime value, churn rate, customer personas and recommendations. In our experience, Metisa produces results for stores with over 5,000 sales items transacted.