Making Money with Mediocre AI Models: A Guide for Business Stakeholders

Making profit out of poor-performing models

In the world of AI, it’s easy to assume that only the most accurate models can bring value to your business. However, this is far from the truth. In fact, even mediocre models can be transformed into money-making machines with the right strategies. In this article, we’ll explore three real-life examples of how we turned poor-performing models into profitable assets, and share three strategies to help you do the same.

Can a Mediocre Model Really Boost My Business?

Yes, it can! Even if your model is only slightly better than random, it can still bring value. Let’s look at an example. We once built a model to predict if a certain product was going to be sold in the next week. The model’s overall predictive performance was mediocre at best, achieving only 60% balanced accuracy. However, the model showed high predictive performance when we looked at the top 10% and bottom 10% predictions. These “tails” were already optimizing millions of dollars. So, despite its overall poor performance, the model was moved to production and generated a positive cash flow.

What if My Model is Only Slightly Better Than Random?

In the healthcare domain, we had a model with 80% ROC AUC, which is reasonable but not great. However, when we looked at the extremes (cases where the model was quite sure a patient had a disease or was healthy), the predictive performance was over 90%. This allowed us to prioritize patients in severe conditions and save resources, optimizing a lot of value despite the model’s overall mediocre performance.

Can a RecSys That Fails for Newcomers Still Be Profitable?

Absolutely! We once built a recommendation system that was good at understanding the behaviors and purchase trends of recurring customers, but terrible for newcomers. Instead of trying to improve the model for the entire population, we decided to profit from the segment of recurring customers. This approach allowed us to generate additional revenue and produce a cash flow to keep working on solving the larger case.

Do you want to further discuss this idea?

Book a meeting with Kelwin Fernandes

Meet Kelwin Learn More

How Can I Transform My Mediocre Model into a Money-Making Machine?

Here are three strategies:

Strategy 1: Look at the tails

If your model isn’t as good as you’d like, look at the top and bottom cases as we have shown in this blog post. This can help you optimize value.

Strategy 2: Retarget your model

If your model is good at predicting a certain segment, retarget your model to that segment. This can bring good performance and cash flow to fund the project’s long-term vision.

Strategy 3: Ignore technical KPIs and focus on business KPIs

Instead of focusing on accuracy, precision, etc., focus on how much money you’re producing, how many customers you’re saving, or how many errors you’re preventing. This will help you see the value your model is already bringing.

Course, Templates

Data Ignite

Learn how to move from technical aspects of your AI models into a successful business implementation.

Learn More

In conclusion, don’t be discouraged if your AI model isn’t perfect. Even mediocre models can bring significant value to your business with the right strategies. At NILG.AI, we’re here to help you navigate the world of AI and make the most of your models, no matter their performance.

Like this story?

Subscribe to Our Newsletter

Special offers, latest news and quality content in your inbox once per month.

Signup single post

This field is for validation purposes and should be left unchanged.

Recommended Articles

Five Years Helping Entrepreneurs Embrace AI

We celebrate the 5th anniversary of NILG.AI with immense pride and gratitude. Over the years, we’ve embarked on a transformative journey, and today, as we commemorate this milestone, we want to take a moment to reflect on our accomplishments and share our exciting plans for the future. NILG.AI Fifth Anniversary: A Year of Remarkable Achievements […]

Read More
Classifying text using LLMs

  Text classification is one of the most common use cases in Natural Language Processing, with numerous practical applications – now easier to access with Large Language Models. Companies use text classification in multiple scenarios to become more efficient: Tagging large volumes of data: reducing manual labor with better filtering, automatically organizing large volumes of […]

Read More
NILG.AI will be at the Iqony Technology Days 2023

Innovation drives the future of the industry, and at NILG.AI, we’re committed to leading this charge. Therefore, we’re excited to announce our participation in the upcoming Iqony Technology Days 2023, scheduled for this September in Brühl, Cologne. Our enthusiasm is fueled by our recent victory in the Hands On Data (HOD) challenge earlier this year.  […]

Read More