The biggest data science and tech predictions for 2022 from analysts like Forbes and Gartner went out at the end of 2021. But as we near the end of Q1, are we seeing them on the ground yet? As a digital services business, here’s our take on the high-level trends we are seeing today within our customers’ data science projects.
We’re already starting to see the rise in demand for support with MLOps to operationally scale machine learning (ML) development and deployment. The level of buzz this phrase is getting takes me back to the early days of data science where everyone was talking about it, but very few knew how to do it. The main reason for implementing MLOps is to govern and manage the ML model lifecycle, minimising errors and streamlining integration and deployment.
In order to break down the entry barrier, organisations needn’t be limited by the number of tasks they can automate as part of MLOps to get into the conversation. We always advise customers to focus on automation tasks incrementally to realise the benefits for the business, rather than applying MLOps “because we can and it’s trendy”, while losing sight of actual business needs.
2. Business value from data transformation
One of the biggest gains data can deliver is through surfacing information merged from multiple sources. Further value can be gained by increasing efficiency of existing operations with techniques such as ML. For the most data-mature businesses, the data use cases are focused on incremental improvements in efficiency, conversions, and automation from continuous innovation, which requires thoughtful change management.
There’s a clear need to bring the value conversation into all ML initiatives. Quite often, customers who focus on the technical performance of ML models struggle to link it to the business value that the solution can deliver. A technical solution must be grounded in business objectives for it to be relevant, and a key to generating value and making progress on the data journey is knowing how to measure success. That, combined with an understanding of ML and AI use cases, makes a perfect data innovation incubator.
3. Customer experience
We’re already seeing customers harness AI and ML as a launch pad to realise value. Consider a retail example: as a result of the recent accelerated shift from physical stores to online sales, omnichannel operations and mindset are as critical as ever. Finding new methods and strategies for leveraging customer data to improve customer experience is a huge focus for data science teams in 2022. That being said, typical use cases of CRM, product management and pricing, document processing with OCR and classification, fraud detection and KPI forecasting will remain the mainstream focuses in maturing businesses.
Businesses like Ascent have a great responsibility to communicate, educate and demonstrate the business value of data and AutoML processes, supporting their customers through the data maturity curve. Throughout 2022, we’ll be helping our customers better understand the relationship between data and decision making, particularly implementing MLOps to achieve greater returns on their AI investments.