What is that Analytical AI?

During many conversations I have with business leaders about AI, most executives think only of Generative AI, i.e., Large Language Models (LLMs) or simply chatbots. The business benefits of the “other” AI, Analytical AI, are not clear to many.

Despite the use of Analytical AI in business for over twenty years, this is not surprising. Analytical AI is after all often embedded and “hidden” in business applications. These aren’t tools you can download and try out in private life. And, it never experienced the hype of GenAI, so most business users are not aware of it.

Let’s change that.

What is Analytical AI?

Analytical AI is the form of AI consisting of Machine Learning and Deep Learning. Machine Learning enables machines to learn from data and perform specific tasks like pattern recognition and prediction. Deep Learning is an advanced subset of Machine Learning, which uses neural networks that mimic the human brain to analyze complex data with minimal human input.

Machine Learning has been used by innovative companies for over twenty years. Amazon, for example, started applying Machine Learning to provide product recommendations in 1998. Deep Learning grew with increasing computing processing power over the past 15 years and is used, for example, in autonomous driving.

Most people know GenAI through chatbots such as ChatGPT or image creation tools such as Midjourney. They use these tools without any technical knowledge. Analytical AI, on the other hand, requires expert knowledge from data scientists. And, it’s often not visible to business users.

Which business problems can Analytical AI address?

Here are thirteen examples where Machine Learning or Deep Learning can help to address business challenges, sorted by business function.

MARKETING

1️⃣ Optimize marketing spend

Promotion response modeling, a Machine Learning algorithm, estimates the portion of revenues that is driven by marketing and sales activities. It identifies how much each promotional channel affects revenues and which channels are the most effective. The channel mix is then optimized, enabling higher revenues with the same investments or allowing for lower marketing spend while keeping the same revenues.

2️⃣ Increase knowledge about customers through customer profiling and segmentation

Machine Learning models evaluate potential customers based on their purchasing power, affiliations with company’s products and services, and many other factors. These models recommend the most important customers for the company, including their channel and communication preferences.

3️⃣ Avoid losing customers with churn analytics

There are signs that a customer may switch to competitor’s products or services. Churn analytics, a Machine Learning model, predicts the probability of losing each customer. These models also identify triggers driving that potential churn and recommend preventative measures.

SALES

4️⃣ Improve customer engagement with next best action

One sales rep at a B2B company is usually responsible for hundreds of customers and needs to address them through appropriate channels with fitting content. The next best action system, built with Machine Learning or Deep Learning algorithms, provides weekly or daily recommendations to each sales rep about which customers to contact that week or day, through which communication channels, and with which content.

5️⃣ Reduce sales rep travel time through route optimization

Sales reps at a B2B company visit several customers per day and many more per week. A route optimization solution minimizes the overall travel time for the week. It also provides the most efficient travel sequence for the day reducing time spent on route and maximizing customer visits.

CUSTOMER SERVICE

6️⃣ Increase revenues with cross- & upselling suggestions

Machine Learning models identify customer segments and sometimes individual customers who are receptive to cross- & upselling efforts. When these customers contact the company, customer service reps are able to suggest buying additional products without risking alienating customers.

7️⃣ Reduce customer service expenses through call center capacity optimization

The volume of customer calls is not constant over time. Machine Learning helps to predict that volume and to determine the optimal call center capacity for the next day, next week, or next month.

PRODUCTION

8️⃣ Reduce downtime and manufacturing costs with predictive maintenance

A machine needs to be maintained on a regular basis to maximize its life span, but no machine is the same. Instead of the same maintenance interval for each equipment, Machine Learning algorithms customize these intervals for each machine. This reduces expenses for equipment that can be serviced later than average. And, this avoids unexpected downtimes and disruption costs for machines that need to be serviced earlier.

9️⃣ Increase production effectiveness and efficiency through machine optimization

Machine Learning models help to maximize the yield and throughput of a machine while reducing energy consumption. In large production environments, these models reduce processing time by optimizing the factory's production flow.

🔟 Decrease operational expenses through supply chain optimization

A supply chain is often a fragile construct with many players involved, leading to many opportunities for disruptions. A simulation model identifies the optimal order quantity, taking into account several possible supply chain scenarios. These models are also used to optimize supplier selection and determine optimal stocking locations.

FINANCE

1️⃣1️⃣ Minimize losses with fraud detection

Machine Learning or Deep Learning algorithms analyze data in real-time to detect anomalies, identify potential risks, and prevent fraudulent activities. This leads to cost savings by minimizing losses.

1️⃣2️⃣ Improve product availability through increased forecast accuracy

Every company producing goods needs to plan production to meet future customer demand. It needs to forecast that demand for the next month, quarter, or year. Statistical and Machine Learning models help to improve the key forecast metric: accuracy.

1️⃣3️⃣ Optimize business processes through process mining

Large organizations usually develop processes over time that may no longer be as efficient and effective as they could be. Deep Learning models optimize these processes and that not only in finance.

Because of the many business needs that Analytical AI can address, it delivers 80% of the business value that AI can deliver to companies. This is according to McKinsey. Since GenAI is limited to text analysis, content creation, and chatbots, it can only address 20% of the business value.

AI adoption is not only about employees using Copilot or ChatGPT. It’s also about employees understanding the value that Machine Learning and Deep Learning systems provide them.