Customer acquisition is harder than ever. Tight budgets, economic uncertainties and fast changing customer preferences are a challenge for retailers. In this article, we explain how 7Learnings’ AI technology enables retailers to leverage customer lifetime value (CLV) predictions to steer marketing and sales efforts and boost their overall profits.
Why is customer lifetime value a crucial business metric?
Customer Lifetime Value (CLV) is a predictive metric that estimates the total revenue a business can expect to generate from serving a single customer during the totality of their company-customer relationship.
Customer Lifetime Value is an important metric in customer relationship management: The average CLV shows how effective marketing and sales measures are and how successful customer retention is compared to the rest of the industry. As it is much easier to sell to existing customers than to acquire new customers, CLV is a key indicator of a retailer’s economic health.
What’s the relevance for everyday business decisions?
How does CLV play a part in everyday business? Our analyses can help to identify which purchases have a high probability of causing recurring or high-value purchases in the future which in return helps to create successful marketing and sales campaigns and not overspend on advertising.
If retailers, for example, know from their data that customers buying a certain product as a first purchase tend to come back for more again and again, thus, having a high CLV, they can justify spending more money on advertising for this specific product, although it might be low-priced. They might lose money on the campaign but they will recover the lost revenue in the later stages of their new-won customers’ lifetime.
How is customer lifetime value calculated?
There are many different formulas for customer lifetime value. The very basic and commonly used formula: “Customer Lifetime Value = (Average Purchase Value x Purchase Frequency) x Average Customer Lifespan” is of little practical value.
In business reality, it can be very difficult to calculate CLV. There are several reasons:
- Dynamic changes: Customer preferences and buying cycles change over time and buying patterns can differ depending on the product or segment.
- Complex calculations: Lots of variables need to be taken into account to make the calculation precise, i.e. cost of customer acquisition and retention should be included, however, they develop dynamically.
- Data quality and availability: Lots of retailers still store data in silos and do not collect all data relevant for CLV calculations. Also, data quality must be high to get reliable results.
- Data Privacy: The European rules for data protection place high hurdles on the processing of customer data.
- Resources: Retailers often lack the expertise to implement technical solutions for CLV calculations in their processes and make them easily available for business users.
How 7Learnings calculates CLV
At 7Learnings, we are experts in data analyses and as we already work with broad data sets, it was only logical to develop a CLV feature which is a great addition to our cross-marketing optimization feature.
The AI algorithm calculates CLV on a product level without using any customer data to be compliant with GDPR. If the available data history is long enough, we can now determine the average profit a retailer makes with a certain product and the product’s retention rate. And based on this information say what the future profit increase will be with each current sale.
For example, if 100 customers bought socks as a first purchase, we can identify that 50 percent returned for more and higher purchases, we can measure the average profit made off those purchases and say that with each sale of the product the retailer will increase their profit in the future by 2 euros.
This product-based metric has much more depth than the average customer lifetime value. Retailers can quantify how (low-priced) products drive future sales and adjust their performance marketing and ad spend accordingly. Returning to our example: Knowing customers tend to come back to their shop for more expensive products after purchasing a pair of socks, retailers can justify more aggressive discounts on socks that have been identified as a sales driver.
Using 7Learnings’ product-based CLV prediction, retailers prevent promoting products with little impact on revenue and focus their spending on high-impact efforts. As we use artificial intelligence, take more influencing factors into account than traditional analyses could and update all data in near-real time, retailers can be sure our CLV prediction is highly accurate.
Where is predictive pricing particularly effective for customer lifetime value?
Now that the calculation is extremely fast thanks to AI and does not tie up human resources, should retailers have CLV predicted for all their products? It’s a matter of budget and data. First, customers need to have enough historic data on transactions for our AI to work properly. That aside, we recommend using CLV predictions to increase sales in hyper-competitive channels like Google Shopping or product comparison sites. Lots of customers have their first contact with a brand on these platforms and it’s critical for a retailer’s success to know how much ad spend is reasonable.
Currently, lots of retailers work with a short-term perspective: They choose which products to advertise based on historic sales numbers or inventory needs and set prices based on experience. Often they simply try to undercut the competition to gain visibility on the platform but at a high cost. As they don’t orchestrate ad spend and product pricing, they risk losing money in the end. With CLV predictions, retailers know which products will lead to high-price sales and recurring revenue. And they can reliably prevent losing money by combining CLV predictions with 7Learnings’ performance marketing optimization feature.
Disadvantages
- CLV as a predicted value always comes with the risk of errors and skewed profit calculations which might only get obvious in retrospect and after having accumulated data over a longer period of time
- For now, CLV predictions are limited to channels like website and Amazon/eBay where customers can gather the necessary data foundation
Benefits
- CLV predictions increase performance marketing effectiveness
- Retailers can choose products to compete on Google Shopping and comparison sites based on actual data
- Retailers can plan their campaigns with foresight: They can see beyond the initial loss a product makes and take into account its long-term effect on revenue
How does 7Learnings mitigate the risks of AI?
We are well aware that the use of artificial intelligence entails particular risks and therefore take various measures to ensure the high quality of our data models. Let’s look at three of the most common concerns.
Data quality is a key factor in the accuracy and reliability of AI results. We check the data that retailers provide to us in advance: With the help of automated tests, we detect and remove bias.
Even if the data quality is good, overfitting the data model can lead to incorrect predictions later on. In this case, results of AI models are accurate as long as they work with training data. When applied to unknown datasets, however, results are poor. To ensure that our AI does better, we monitor our data models daily and retrain them if necessary.
Artificial intelligence seems like a black box to most users. No wonder they find it difficult to trust AI’s calculations and recommendations. That’s why we put particular emphasis on transparency. Our solution offers various ways to understand how AI came to its results. As a customer, you benefit from the performance of the AI without having to give away control. You can understand at any time, also with the help of our data scientists, how our solution arrived at its assessments.
Knowing the data changes the game
CLV predictions are an incredibly powerful use of AI: Knowing which products attract high-paying or recurring customers enables retailers to steer their everyday marketing and pricing activities with foresight. You can use CLV data to decide which products are worth additional promotions or which discounts are justifiable. And you can orchestrate different channels in a way that enhances overall cost efficiency.
CLV predictions are especially useful for high-competitive platforms such as Google Shopping that can easily become a cost trap. Knowing the data, it’s easy to define the right products and set the right prices. It’s the difference between gaining hundreds of new customers or spending thousands of euros in vain.
As with any AI innovation, retailers that have the courage to move first benefit disproportionately, as their competitors are not yet using this new technological support. That’s why we strongly invite you to reach out.
