Data science has the potential to inform critical decisions, allowing companies to automate or scale processes and improve their business.
This week, I’ll cover five ways we’ve recently used data science to help our clients improve their bottom line. Some of these methods involve the creation of data products while others involve automating or optimizing processes in place.
1. Improved marketing efficiency – we’ve had a lot of success using customer segmentation to help our clients understand their customers better. This has a direct impact on how our clients market to their potential customers. By identifying customer segments, we help our clients understand the types of customers they have and develop marketing messages that speak to them directly. It results in a coupling of customer needs to product value proposition. The value in improved marketing efficiency is in lower cost to acquire a customer and overall better customer lifetime value (LTV), since attracting the right customers avoids early and expensive churn.
2. Personalized customer experiences – customers now expect personalized customer experiences. If they don’t get it from you, they’re gonna go somewhere else. We’ve had a lot of success helping our clients deliver real-time product recommenders that delight their customers. And the best approaches come from testing and constant iteration. We use offline data to generate a first-approach model to predict customer preference. From there, we stage improvements and test them in real-time, using KPI’s to assess progress. The value to personalized customer experiences is generally greater customer LTV, as customers come to rely on your brand to provide an engaging, fresh experience.
3. Incident detection – sometimes things go wrong. Recently we’ve worked with a client that tracks data related to construction jobs. We used millions of rows of this data to accurately detect when an incident will happen. We’re deploying the model country-wide to help them flag potentially dangerous incidents before costly mistakes are made. This is more substantial (in terms of risk and value) than simple customer churn detection, but very closely related. The value in incident detection can be the saving of a potentially lost customer or the saving of a life when a job site is operating in an unsafe manner.
4. Match producers to consumers – It may sound trivial to match resumes to requisitions, but this space is filled with nuance, such as finding categories or tags that apply to resumes and requisitions even when the word was not used in either, but implied. This type of work is largely powered by Natural Language Processing (NLP). The latest language models powering NLP make it possible to do things I could only have dreamed of 10 years ago, such as using a language model to finish sentences and paragraphs with a level of coherence that make it indistinguishable from what a human might write. The value in matching producers to consumers is important to fields from online dating to college admissions, court proceedings, and PR pitches.
5. Enhanced internal decision making – we do this chiefly through data dashboards, and to a lesser extent through financial projections. Data dashboards are an essential part of a healthy business. They’re the speedometer that tells organizations how quickly they’re zipping along. It’s important to have senior analytics professionals oversight in their creation, as it’s all too easy to “lie with statistics”. The value here is obvious. Only with a strong command of company KPIs can a business understand how it is fairing and when changes must be made.
These are only some of the services we at Bennett Data Science have helped clients with recently. If you’re looking for help in any of these areas, please hit reply and we can set up some time to talk.
Have a wonderful week!
Listen to ‘new’ A.I.-Created Nirvana Song Drowned in the Sun
The ‘new’ Nirvana song Drowned In The Sun is a composition created by A.I. as part of an initiative raising awareness of mental health issues in the music business. It was created by feeding MIDI files into Google’s A.I. program Magenta, which learns how to compose in the style of given artists by analysing their past work. Vocals for the song are provided by Atlanta musician Eric Hogan, frontman of a Nirvana tribute act. Read more here.
Facebook Tries to Make A.I. Fairer by Paying People for Data
Artificial intelligence systems are often criticized for built-in biases. Commercial facial-recognition software, for instance, may fail when attempting to classify women and people of color. In an effort to help make A.I. fairer in a variety of ways, Facebook is rolling out a new data set for A.I. researchers that includes a diverse group of paid actors who were explicitly asked to provide their own ages and genders. Facebook hopes researchers will use the open-source data set to help judge whether A.I. systems work well for people of different ages, genders, skin tones, and in different types of lighting. Read more here.
Why A.I. Knows who you Find Attractive Better Than you Do
When it comes to earning social currency, being attractive is as good as gold. A team of scientists from Finland has now designed a machine learning algorithm that can plumb the depths of these subjective judgments better than we can and can accurately predict who we find attractive via our unique brainwaves — and even generate a unique portrait that captures these qualities — with 83 percent accuracy. Read more here.