MORE POSTS

Customer Segmentation – How Machine Learning Can Set You Apart

September 27th, 2019   •   Allen Klein   •   Min Read: 2

Excel and structured spreadsheets have been a fantastic tool for marketers and business owners for years. It brings structure to our data and helps us summarize and understand the reality of our data.

But, the limitation is obvious, these tools only help us understand the past, what about predicting which customers will be valuable next? This is where Machine Learning can help us shift from the rear view mirror and help predict our next best customer.

We’re going to focus into the travel and hospitality industry and put on the hat of owning a hotel. Let’s imagine that you have 5 years of customer level data with attributes ranging from the number of travelers, duration of stay, characteristics of those stays, etc. What is the right mix of how to weight those attributes for how they lead to higher lifetime value (LTV)? How do you know which attributes lead to high LTV and what if more than 3 of those lead to high LTV? I’ll be honest, that math isn’t easy and while possible, tough to repeat, and I can’t speak for you, I can’t learn that nuances that quickly and at scale.

We built a model that considered the following:

1. Identifies characteristics and behavior of high spending customers
2. Classifies existing customers into high, mid or low value customers
3. Predicts whether a new customer will become a high value, mid or low value customer

Currently, our model only works with the data structure from RoomMaster from Innquest. But, it works like a charm and learns over time and becomes more customized to your data over time. We can adapt it to other data sources, if you’re interested, read below.

Here is how we’ve seen our clients apply this segmentation:

1. MVP list at check in for hotels to have a special greeting for customers that are expected to be high lifetime value.
2. Targeted email campaigns for each of the LTV buckets with different offers in each.
3. Ad targeting by bucket on major advertising platforms.

The Results: The first campaign drove an incremental $35,000 in revenue for the client.

At Kurvv, we’ve learned that 0 to 60mph on your data is better than 60 to 120mph as it gets our clients on the right path to leveraging their data set and creating more value.

If you have CRM data and want to explore applying this model to your business, drop us an email at yo@kurvv.ai.