How eSalon makes the right investments on their most profitable customers

How a subscription box company used Ocurate to optimize lifetime value and increase profits

A women’s beauty subscription box came to Ocurate looking for a way to retain its 250,000 customers and convert high-value leads. Their existing method for predicting lifetime value (LTV) was only 75% accurate, and they enlisted Ocurate to improve accuracy and unlock profitable growth.


Subscription brands don’t make money on a customer’s first order. The initial customer acquisition cost is offset if that customer becomes a loyal subscriber. The north star metric for leading subscription brands is a customer's lifetime value (LTV).

This is no different for a particular subscription company that sells women’s beauty products. Prospective customers fill out an online questionnaire which helps the brand both personalize product recommendations and gather important data. Based on the anwer patterns, the brand can predict the prospective customer’s LTV. 

If the prospect filled out a questionnaire but didn’t make a purchase, the brand retargeted that prospect on major ad platforms — but only if the data indicated the prospect had high predicted LTV. This is all part of the brand’s strategy to invest more in ads, promos, and customer service for leads and customers they know will generate a high LTV.

However, the brand’s standalone customer data and limited data science resources prevented them from employing that strategy to the fullest extent: They were only able to identify a small slice of their highest value customers. This meant they ignored untapped, highly profitable revenue opportunities.

The brand encountered three major obstacles:

  • Limited pre-purchase data made it hard to identify, appropriately invest in, and convert a large percentage of high-value leads
  • Some subscribers churned regardless of intervention, and it was unclear which subscribers were at risk of marginal churn 
  • It was unclear which churned subscribers were worth the investment to win back

Their executive team attributed 18% of lost gross profit to these obstacles.

Results with Ocurate

Ocurate used machine learning paired with the brand’s historical customer data to identify high-value leads and customers at an individual level. 

Ocurate identifies:

  • High-value leads based on conversion probability and predicted lifetime value
  • Current subscribers most likely to upgrade plans
  • Customers at risk of margin churn that are worth the investment to retain
  • Lapsed customers worth reactivating

With 93% accurate LTV predictions from Ocurate, the brand identified thousands more high value customers at the lead level. Not only did they more accurately predict LTV for those who completed surveys, but also for the hundreds of people a week that abandoned the quiz. 

For leads alone, the brand projects Ocurate’s insights will increase gross profits by as much as 24% because of the ability to make the right investments to convert more high value leads.

beauty product high ltv with Ocurate

“I am bullish on Ocurate’s ability to help us focus on the right customer based on their superior data and AI tied to predicted value. Initial tests have shown a much smaller error in Ocurate's model compared to our existing solution.”
Andrew Grant, VP Growth.” 

Ocurate’s unique benefits

Superior data

We took five years of R&D to build a future-proof database that maps every adult American’s foundational personality, behavioral, and attitudinal traits with explicit consent.

Better AI

PhD-level data scientists from Stanford, Wharton, Columbia, and Cornell built our data models, which are more than 90% accurate.

Seamless integration

Insights are tailored to your industry, business, and customer base by examining historical customer data (like transaction history, production data, and customer ID). We integrate real-time predictions into your existing workflows via API — no matter where or how your customer data is stored.

Ocurate will soon release open-source software for lifetime value modeling. Want to see it in action yourself? Sign up here to join the waitlist.