Data science trends in retail for the post-Covid world
2020 was a difficult year for every business. The coronavirus changed every aspect of our operations. One of the biggest trends in data science caused by the Covid-19 pandemic was an acceleration of the integration of data scientists and data-driven analysis into greater company operations. Data was vital to survive and thrive in the chaos caused by Covid.
Because of the increasing importance of data science — or business science as I think it’s more appropriately considered — in retail, data-driven trends in the sector matter to every data scientist. It’s not just for those of us focused on the supply chain, pricing or other retail-adjacent applications. Data science trends in retail portend trends for the field as a whole in the post-Covid world.
Here are five critical trends that I forecast will dominate data science in retail in 2021 — and that every data scientist must therefore be prepared to face head-on in the new year.
1. Automated analytics and analysis
Almost every business uses Big Data. It’s become a basic necessity of modern retail. Next step: automation.
Over the next year, more companies than ever will automate their analytics, both in terms of data collection and analysis. Data quality doesn’t matter unless you can leverage its insights autonomously.
Business science must be autonomous, not just data-driven, to be effective.
I’ve been a tireless promoter of autonomous systems for analysis for years. Still, I’m not the only person who thinks that 2021 will be the tipping point for automation in the retail sector. McKinsey has noted it as a key trend in both consumer goods and fashion for 2021. Automation is the future for retail — which means that data scientists will need to shift their focus to automated models and analytics.
2. Maturation of AI
With the rapid acceleration of AI adoption that we’ve seen over the past year, 2021 is likely to become a watershed for AI maturation. That is, more models will deliver the optimal outcomes desired from their implementation. In terms of Gartner’s model for AI Maturity, many businesses will hit the final two stages. AI application in many organizations will finally tip into systematic or even transformational.
For data scientists, AI maturation is exciting, as they can act like real business scientists, delivering accurate, actionable insights from their models. It also means an adjustment in the approach to model development. Optimization of a mature model is a new challenge that many of us will be facing over the next year.
3. Agile models — and supply chains
More than anything else, 2020 has taught us to expect the unexpected. Supply chains are vulnerable to unanticipated disruptions, whether they come from pandemics, natural disasters, or whatever crisis we face next.
A good AI model can digest any relevant information, map micro-patterns and micro-trends, learn from its own mistakes, and continuously adjust — all automatically. Just like a human would react, but more efficiently — and also more constrained by verifiable data. That means that once armed with new data and even limited observations, AI models learn and respond to changing conditions much better than any economically viable team of human managers armed with Excel and their gut feelings alone ever could.
Data-driven decisions can better mitigate the impact of sudden deviations from historical patterns, giving companies a vital competitive advantage.
In 2021, more agile supply chains are only going to increase in importance as retailers prepare for the next disaster. As such, companies will demand more agile forecasting models from data scientists. Data scientists must prioritize creating models that can more deftly respond to data imperfections, model drift and drastic deviations from historical patterns.
More agile supply chains will be the priority of every retailer, and more agile models will be the ultimate goal of every data scientist serving them.
4. Consumer-driven products & time-lag elimination
The coronavirus officially killed the traditional, top-down “push” model of retail. Successful retailers must now let customer preferences and demands drive product offerings. A faster response time is vital to meet demand. Product lifecycles will be shorter, requiring the supply chain to adjust accordingly.
What does that mean for data scientists? Speed is more important than ever.
Models must work faster, supply chains must be optimized down to the second, and systems must recognize and respond to even the slightest change in consumer demand. For each of these developments, data scientists must work ever harder to increase data velocity (and improve all of the 5 Vs of Big Data!). Every data scientist will be pushing their models ever-closer to real-time data and analysis in 2021.
5. Self-service AI optimization apps accessible to all
During the pandemic, companies have rushed to implement AI, creating exponential growth in its adoption. Despite this, a report by Boston Consulting and the MIT Sloan Management Review reports that only 11% of companies implementing AI have seen sizeable returns on the investment. Why? Companies don’t know how to use the technology effectively.
For AI to fulfil its potential, AI must be easier to use.
One of the biggest problems in AI that the pandemic exposed is the poor user experience of AI. The average retail executive struggles to use available tools designed for other data scientists. It’s why I champion business science: data science is useless unless what you learn can be applied for practical, applicable insights.
What does that mean for 2021? Data scientists will design AI optimization tools and applications to be more user-friendly. More self-service, highly accessible apps will be released at higher rates over the next year. Simple optimization tools will no longer be a secondary goal; it will be the primary focus of every data scientist designing models to optimize any element of the supply chain.
In fact, my own company Evo has made the release of our self-service portal a priority for 2021 for this exact reason. Even smaller retailers who can’t afford personalized service and a model tailored by a data scientist deserve the opportunity to optimize their supply chain to compete in the post-Covid world. Unless these self-service offerings are accessible, AI successes will continue to lag behind adoption. Data scientists will have to change the way they design their applications to keep up in 2021.
How to get ahead of data science trends for 2021
Each of these 5 trends will critically change the way data scientists operate over the next year. It’s time to adjust to these trends now before they accelerate — and you fall behind. You may be tempted to ignore these trends as isolated patterns in a single industry, but the reality is that they exist across the business world. They will affect your work sooner, rather than later.
I’m looking forward to 2021 and a post-Covid world. But as much as the end of the pandemic, I’m looking forward to these trends becoming a reality for data scientists everywhere. Many data science trends for 2021 will help us all become better business scientists.
About the author
Fabrizio Fantini is the brain behind Evo. His 2009 PhD in Applied Mathematics, proving how simple algorithms can outperform even the most expensive commercial airline pricing software, is the basis for the core scientific research behind our solutions. He holds an MBA from Harvard Business School and has previously worked for 10 years at McKinsey & Company.
He is thrilled to help clients create value and loves creating powerful but simple to use solutions. His ideal software has no user manual but enables users to stand on the shoulders of giants.