top of page
Maximize sales efforts
by using AI for identifying
Upsell opportunities
It is a sales technique designed to persuade customers to upgrade to a higher tier or better product. By using modern AI techniques, sales departments can be helped to make the best use of their resources to get as much out of their efforts as possible. We will go into detail about how you can create your own upselling model with zerocode.ai in this article.
The Challenge
Companies only have limited resources, and knowing where and to whom to target upselling is key to using resources efficiently. This not just makes the sales department more efficient, but also leads to happier customers, as each customer gets the product that best suits their needs. Current manual upselling techniques are not data-driven, and as more companies adopt these AI tools for sales, manual techniques will become even less important.
The opportunity for deep learning
Because sales today is entirely digital, or at least involves digital interactions, many data points are captured in an automated way. And many different variables and data points are key to a problem for which AI could be the solution. Not only can AI be used to automate tasks that a human could do, but it can also help find patterns that would not be visible to the human eye.
AI model to use
For this use case, our model of choice is: Tabular data classification
How does the model work?
Out of the tabular data we have, we need to pick out some columns that we know about the customer, the product or the context. These columns are used as input and are usually called input features. At that point, we need to select a column that we want the model to predict, which is the label feature in the training dataset.
What columns to include in the table data depends on what is accessible and what can be found out by trying different variations of zerocode.ai. It is also useful to be creative when thinking about what features can be used.
As an example, the tabular data would consist of one row per customer. The features could be divided into those that relate to the customer, those that relate to how the customer uses the product, and other features. The customer characteristics could be the type of user (novice or advanced), how experienced the person is in another area, or where they are located. Customer characteristics with the product could be how many days have passed since signup, how many times the customer has interacted with the product in the last month, how many interactions there have been in the last week, what parts of the product the user has used, whether the company has had any other interactions with the customer, and more. The different characteristics are the hardest to find, but they can be things like the date, summer, or a Friday night.
A column in the tabular data should be the one to predict. A good flag for this use case would be a column that says "Yes" or "No" depending on whether the user agreed to upgrade within a month. This column would then be the one the model could predict based on the other features.
Data requirements
To be successful, we need historical data about our customers in a tabular format like Excel. How many data points or rows are needed is hard to say without trying, because it depends on the task and the variety of data points. A good starting point is a table with at least 100 data points.
Model results and success
On the platform, you can see the performance of the model in the evaluation view after you run your first experiment. For this type of experiment, it is important to look at the model accuracy and the number of false positives and false negatives.
Where to learn more?
If you want to build this yourself with zerocode.ai, a suggestion would be to follow our tutorial on how to predict with tabular data. And then use your own dataset in a similar way as described. The tutorial also goes into the details on how to measure model performance and what to do to improve it even further. Have fun!
bottom of page