The prebuilt customer lifetime value (CLV) notebook in AI Workbench can be used to identify, target, and grow customer transactions. The machine learning model predicts key CLV metrics for a customer segment, including:
Probability alive – Likelihood a customer will make future purchases.
Expected purchases – Predicted number of purchases in the forecast period.
Customer lifetime value – Estimated total revenue a customer will generate.
Before you begin
Make sure you have customer order data available in your customers' BlueConic Timelines.
Add a CLV notebook in BlueConic
Navigate to More > AI Workbench > Add notebook.
Choose CLV notebook from the pop-up window.
Give your notebook a name.
Save your settings.
Set the CLV notebook parameters
Select the Parameters tab.
For the CLV segment parameter, choose a customer segment to examine.
Select a prediction period in months (default is 12). This period counts the next X months from the date the notebook is run.
(Optional) Select start date and end date (in UTC format) for the transaction data the model reads.
By default, the notebook writes its results to the CLV Probability Alive, CLV Expected Number of Purchases, and CLV Customer Lifetime Value profile properties. You can customize these, but that requires updating the code.
Save your settings.
Run the CLV notebook
Select the Schedule and run history tab.
Click Run now to run the notebook analysis manually.
To schedule the import and export for a future date, activate Enable scheduling.
Click the Settings icon to select how to schedule the notebook by choosing an option from the drop-down list.
Set a time for the import. Click OK.
Save your settings.
View your results
After running the notebook, click Preview to view the results. Scroll to the bottom for a graphical summary of the CLV analysis.
The notebook displays:
Frequency of repeat transactions
Expected repeat purchases per customer
Probability of a profile being an active customer
Predicted customer lifetime value (CLV)
CLV distribution across the segment
FAQs
Once I build my CLV segments, can I only use them for web personalization? Or can I use them in other channels like email and Google?
You can use your CLV segment across channels once you’ve created them. Because your models are connected to your profile database in BlueConic which are connected to your marketing technologies, you have the ability to apply the same segment across all channels. This is a great example of why having a unified, “single” customer view is so valuable in terms of efficiency.
Does the CLV model work well with ‘high-cost, low-frequency’ products? How low frequency is low frequency? 1x purchase a year?
The CLV AI marketing model estimates purchase cadence, adjusting timeframes (days, weeks, or months) based on product frequency. For low-frequency purchases, ensure the observation period captures multiple transactions for accurate results.
What qualifies as low-frequency depends on your business, but AI Workbench allows model adjustments to fit your needs. Machine learning requires sufficient data, so if transactions are too sparse (e.g., one per year with a small customer base), consider alternative models that analyze customer attributes to predict buying behavior.
How is the CLV model trained and validated on my data?
The dataset is randomly split into a training set (for model training) and a testing set (for validation). Each set is further divided into a calibration period (first two-thirds of the dataset) and a holdout period (last third).
The model is trained using only the training set’s calibration period. It then makes predictions for the holdout period using test data from the calibration period. These predictions are compared to actual outcomes in the holdout period to evaluate how well the model generalizes to unseen data and future customer behavior.