How Explainable AI Can Elevate Your Decision Making Process And Be The Factor To A Successful Business

Okay so imagine you are a founder who runs a clothes brand and you are doing well. Getting a few hundred sales in your store and on through delivery platforms throughout the day. You have a good set of repeat customers as well who buy your products multiple times.

So you now have a huge dataset of your customers and their information, what product did they buy, the size, where did they come from, whether they bought it online or offline, gender, did they avail any discounts, their payment methods and more. With this dataset you can now do magical things.

What Your Data Can Reveal

• See what kind of customers are buying what product (Eg: Men from South bombay mostly buy black oversized T-Shirts of size M and XL, while women from Noida mostly buy Blue jeans of size 34 to 36)
• See when your customers are buying the most (Eg: Just before a huge music event your sales blast as most of your customers are Gen-Z , while they are relatively less orders in months of July, probably due to high rains and less events)

And a lot more. This can help you make your marketing and sales strategies like: Increase paid ads for black oversized tees in other parts of mumbai , or have more staff in January and February due to Lollapalooza which happens in February.

Moving Beyond Predictions

Now you can go one step ahead and use your data, write an ML Algorithm (Eg: XgBoost) to predict with 90% accuracy what customer is likely buy your product again or not.

Eg: The model predicts Customer 8234: will buy your product again

Now you will know which customer is more likely to buy your product again or not. But what next? What will you do with this information?

The Power of Explainable AI

If you use ExplainableAI here, this is how your insights can change

Eg: The model predicts Customer 8234: will buy your product again with a confidence score of 80/100. As he is from Bandra and has visited your website multiple amount of times in past few days and also paid with Axis bank credit card. These metrics show a strong indication this customer to purchase from your store again.

Also, the model says that it considers "time spent on the website" the largest contributor for the prediction, followed by delivery time and location, age.

This shows ExplainableAI goes beyond traditional Machine Learning and does not only tell you what, but also "why?"

Actionable Insights You Can Use

Now this is what you can do next knowing time spent of the website is the highest contributor to a person buying your product again:

• Simulate what is the optimum time a person spends on your website to buy the product
• Once you find out the optimum time (say 7.5 minutes) you can sit with your UX team to redesign the website/ make changes in such a way that each customer spend about that much time
• Create "Counterfactuals" Eg: If customer 2847 from Pune spent 6 minutes instead of the 10 minutes he actually spent would he be more likely to buy? or if we promised 3 day delivery instead of 7, who effect would that have on the prediction?

Now you see, with the addition of ExplainableAI , the decision making process elevates and you can make good , more confident decisions to improve and scale your business.