A few things to keep in mind when considering launching into Predictive Analytics, with or without the help of Artificial Intelligence
You’d like to better understand your customers and anticipate their moves?
You feel uncomfortable knowing you are sitting on a significant amount of untapped data?
You’re not alone, for this is the plight of most business owners. You’ve also read that Artificial Intelligence or Machine Learning could help you?
Then you’re almost ready to launch your first Predictive Analytics project. The following checklist could help to prepare you:
#1: Start by outlining areas you want to understand more
* What explains my Net Promoter Score?
* What drives uptake of this product?
* Who are my non satisfied customers and what explains their dissatisfaction?
* Which customers are likely to churn next month and why?
It’s important to choose areas that can be looked at in depth. For instance, in the case of the last question above, you could look at the following:
Obtain a clear customer profile: are there clear warning signs that a customer is about to churn?
Example of signals:
· unsubscribed to the newsletter within last 30 days
· contacted the customer centre within 3 days of monthly invoice
Identify a customer’s intention based on behaviour: do I receive signals I can act upon to help increase my business through sales & marketing activities? A list of at-risk customers will help you to target these customers with specific offers, thereby reducing your churn rate.
Even if you are not yet in a position to undertake sales & marketing activities, an accurate customer profile will already be of value, by helping you to better understand your customer base and identify your most important customers and those among them who are at risk. Not all organisations are able to do this.
#2: Available data: start with what you have, now
You’ll be surprised with the results you can achieve with the data you already have at your disposal. The ongoing gathering of data can continue in the background even as you start to mine the data you currently hold. This is one of the major advantages of using an AI-based approach compared to a classic Data Science approach. I’ll elaborate on this in a separate blog.
#3 You know your job but understand little about Data Science? No worries, leave it to the machine
Achieve real, tangible results in only a few hours or days, working on your own or with a business analyst. This process no longer requires months of time investment.
Stick to the business you know without becoming embroiled in complex and sometimes irrelevant technical questions. With Predicsis.ai we can affirm that a couple of days’ analysis is sufficient to provide a clear first level of understanding of the business question posed. This can then be reinforced and improved upon over time.
#4 The real life test
The final stage is to generate Sales & Marketing activities based on the perceived customer intention, then to run tests on the results to obtain a clear view of the benefits of the Predictive Analytics output.
Give control back to the business through the use of the Predictive Analytics solution without recourse to Data Science platforms. Don’t waste time and effort trying to gather all possible data, and ignore people talking jargon (neural networks, random forest, gradient boosting…): focus on the business input you require. Ask for results in days, not months, to avoid wasting time and money when you and/or your company might not be ready for Predictive Analytics.