Revel Blog

Artificial Intelligence, Machine Learning, and Your Business

Sydney Kida | December 4, 2019 |

Industry Insights and Trends
Artificial Intelligence, Machine Learning, and Your Business

If you’re living on planet earth, chances are you haven’t missed the developments in artificial intelligence (AI) and machine learning (ML). Seems like they’re everywhere – in your personal assistants (Siri, Alexa), self-navigating cars, image processing algorithms that can analyze x-ray scans or group your photos according to family members, in military applications — the list goes on infinitely. 

What do AI and ML mean for your business?

The question is, what do AI and ML mean, and can these buzzwords affect your business? Let’s go with an intuitive definition: AI tools allow your computer, website, and/or point of sale (POS) to do things that seem really smart, things that you’d normally think require human intelligence, and sometimes much more. A subset of these tools, ML algorithms, allow computer software to scan data, learn from it, and apply valuable conclusions. Combine these tools with the advances in big data collection and analysis, and you’ve got an amazing toolbox to help you with your business.

As a restaurant or store manager, you’re probably most interested in how these tools can impact your bottom line. There are many indirect approaches here where AI can help. For example, you can use 24/7 online chatbots to improve your customer service. Better service eventually means more customers. In  this post, however, we’ll take the head-on approach. Let’s see how AI can make your marketing efforts generate more revenue!

Marketing is all about getting the right message to the right people to encourage them to buy your product. That’s why it works best when you know your audience. The more you know about what motivates your customers, the better you can personalize your offer. 

Which brings us to the topic of this post: How can AI tools help me optimize my campaigns for maximum revenue?

Let’s look at two scenarios:  

  1. You’re launching a “25% off” campaign. Who should you target? On the one hand, you might want to show gratitude to your most loyal customers. However, your most loyal customers might have already planned to come during the campaign period anyway. By rewarding them, you’ve simply lost 25% of your revenue. That’s a losing campaign! So, how about targeting members that would not have come otherwise, hoping the reward will change their minds? Great idea, but how do you find them? At this point, you’d start guessing: Maybe target members who haven’t visited for months. They probably won’t come soon. Or perhaps members that came last week. What are the chances they’ll come again in such a short timeframe? 
  2. You want to promote a new product using a “Spend over $20, and get this item for free” reward. You don’t really care who redeems this reward, but it has to be during the next week to give the new product a nice kick start. This time, it would be a waste to send the offer randomly since most customers aren’t likely to come so soon. You don’t want to spam them with an irrelevant offer. How do you find customers who are likely to come so you can reap the benefits of a high redemption rate?

The above scenarios demonstrate the challenges of matching the right reward to the right audience for the desired outcome. It’s the toughest challenge in incentive-based marketing! Luckily, this is a classic problem for ML algorithms. Let’s see what you need to make this work.

Creating a revenue-optimized campaign requires:

  1. A purchase database for extracting customer visit patterns as input to the ML algorithm
  2. A machine learning toolbox for predicting future visit probability from historical purchase data
  3. A customer database for your customers’ preferred communication method (phone number for SMS, app for push notifications, emails, etc.)
  4. A customer engagement platform for sending rewards and allowing customers to redeem them at the POS

With these resources, technical users could export the purchase data from their POS (for example, as a standard CSV file) and import it into a cloud-based ML platform (one example is Google BigQuery ML). A logistic regression algorithm would allow you to use member purchase history to predict the probability of member visits during your campaign period! Armed with this data, you could select the appropriate population to target, and import it to the customer engagement platform to send the reward.

Less technical users would probably prefer to use an engagement platform with all this magic built in, integrated with their POS. So, no tedious export/import processes involved. Just select the rewards you would like to distribute, and let the AI find the best reward for each member to maximize your revenue.

AI and ML Solutions

Como has recently launched Comillia AI, a new theme in the customer engagement platform's product for all AI- and ML-related features. Comillia AI Campaigns is the latest feature in this theme for launching hassle-free, automatic, weekly campaigns. Just select the rewards and the number of weekly members to target, and Comillia will launch your revenue-optimized weekly campaigns for you!

Learn more about Comillia AI and Como Sense to see if these AI and ML features are right for your business. 

About the Author

Danny Albocher is VP and Chief Architect at Como. He oversees Como’s technical design and system architecture, as well as special R&D tasks. Before joining Como, he served as cofounder and CTO at Keeprz, a loyalty and engagement platform for brands and businesses. Prior to that, he founded Artomania, a startup that offered DIY art kits based on digital image processing. Danny holds a B.A. (cum laude) and an M.Sc. (cum laude) from the Technion - Israel Institute of Technology, with majors in 3D graphics and surgery simulation.