Are you looking to make machine learning analytics work for your retail operation? Part two of this two-part blog series has 4 recommendations on how to avoid AI pitfalls.
While numerous large retail companies are making productive use of AI – Amazon being the prime example – it is still very much a technology, or suite of technologies in relatively early development.
This means retail leaders taking a look at introducing AI to their operation need to proceed with caution. There's a lot of hype, inflated claims and hucksterism to be wary of as numerous developers seek to promote their – largely untried – solutions.
Here are some recommendations to help prevent an expensive AI misstep:
1. Use AI for the Right Reasons
Using AI because someone in senior management has read about the competition using it is not a good reason. Instead, leaders should ask what technology will drive the most value for the bottom line.1 For example, you might ask which areas of the business have lower profits than they should, or what tasks employees are doing that are unpleasant, or where the most errors are made.2
Adopting AI might be the answer to solving these challenges, but it just as easily might not. As always, with new technologies, asking the right questions will hopefully reveal the realities of your needs and the areas where new solutions may be practical solutions.
2. Ensure the Right Leadership
With IT capabilities so central to retail success, it comes as no surprise that for many organizations the ownership of IT projects has been fragmented across numerous departments. With IT budgets increasingly spread across the enterprise, AI implementations can become pet projects of a single department, without general oversight. This often leads to stalled progress or failure since important considerations like regulatory compliance, data management, security and more may be overlooked.3
It is therefore extremely important to keep oversight of AI projects within the purview of the company's chief information officer. AI is complex and requires careful integration of many functions and data from across the enterprise. It will require planning and input from across the enterprise, with guidance from IT and senior management. Preventing departments from going rogue with their own AI experiments will reduce the risk of wasted time and costs.
3. Don't Forget the Data
Before AI will work, you have to have the data to feed it. If you are operating in an e-commerce or omni-channel environment already, you have access to reams of information. Information from your e-commerce sales, logistics processes, inventory control, marketing, and so on may all be valuable fodder for an AI system to interpret. The key is identifying exactly which data will be useful for your AI project and making sure it is captured and stored in the right format.
The good news about AI is that it can tolerate data that's not completely "clean". Data with duplicate records or even outdated and incomplete information can still be used by AI systems thanks to their ability to learn and fill in gaps.4
4. Get the right people
Part of solving the data question will involve finding staff with the skills to implement and maintain AI systems. This is a serious challenge. In fact, lack of AI talent is seen as a major barrier to the technology's adoption by many enterprises.5
There is a huge shortage of data scientists – the people who will manage AI systems – around the globe. Their skills are so valuable that numerous acquisitions of AI startups have been fuelled by the need to gain the talent behind the technology, more than the tech itself. It's estimated that graduates in the discipline will increase by seven percent a year, but demand will outstrip supply, with a 12 percent annual growth rate.6 Retailers seeking to make an AI breakthrough need to consider the cost that hiring these skilled practitioners – whether directly or through outsourcing – will add to the cost of their project.
Real Intelligence
The amount of data created from digital sensors, virtual reality applications, and smart mobile devices doubles every three years, and while the ability to store it has expanded proportionately, the cost has dropped dramatically.7
AI allows the utilization of the vast quantities of data – also known as Big Data – made available by e-commerce. It also creates an opportunity for the retailer to be constantly connected to their customers, leveraging the data that online shopping creates, to stay top of mind and make tailored offers. In fact, for now marketing and sales applications are the most common uses of AI in the retail sector, with supply chain applications ranking in second place.8
However, as we have seen, AI for supply chain is really still in its infancy. And this creates an opportunity for retail businesses that want to take advantage of AI technologies in the near future. Because AI capabilities are still in their formative stages, now is the time to look at how the information and process improvements it can afford will affect other processes and systems that the business relies on.
For companies with retail distribution operations, this means ensuring you can keep up with the increased velocity of orders coming and going from your distribution and fulfillment centres, as well as being prepared to shift on the fly as you learn more about future demand.
That's where C3 Solutions’ best-of-breed scheduling system comes in. If you want to leverage the powers of artificial intelligence to speed up your operations you need to ensure that all areas will be able to keep up. You cannot afford to have bottlenecks and congestion slowing down the arrival and departure of goods and orders at your dock doors.
Implementing a dock scheduling system will allow for the smooth arrival and departure of trucks, and help ensure that your inventory is where it needs to be at the right time. It will be a key piece of the plan when it comes time for your operation to step up to the improved speed and accuracy that AI will bring.
Think about it this way: AI will mean pinpoint accuracy in the delivery of goods for replenishment. If your docks aren't ready at the moment that inventory needs to be delivered, then what's the point in having the precision of AI to know that those SKUs are needed? Likewise, on the outbound side, whether it’s a parcel truck arriving for last-mile delivery pick-ups or a full-truckload going out to replenish a bricks-and-mortar location, if they are forced to loiter in the yard waiting for the load, then you are wasting the capabilities that you've paid for by implementing AI analytics.
But if you have a scheduling system in place, you'll be able to make sure that loads match up with time slots at the docks and you'll be leveraging your AI investment to its intended result. By using the latest in C3 scheduling technology to coordinate your dock equipment and labour with the other fast-moving parts of your retail supply chain you will be able to maximize the ROI on your IT investment.
A Cautious Gambit
Artificial Intelligence should be on your radar if you play on the retail chessboard. Just as AI can play the game of chess better than a human chess master, it will also be quickly learning how to master retail operations. That may be a little fanciful, since AI won't do anything without human creativity to do the initial programming, but once this technology evolves and matures to the point where it's commonly employed and no longer solely at the disposal of IT-native retailers or the very large, it will be a fundamental game changer.
Those that are not feeding the data they collect from their operations into the big computer brains will be at a serious disadvantage. They will be slower, less accurate, unable to predict with precision, and will lose sales. Just as in chess, the game relies on being able to think many moves ahead, retail success will likely soon come down to the ability to leverage AI to its best advantage. Ignore it and you may find your operations in checkmate.
REFERENCES
[1] "Avoid AI Novelty: Pitfalls to AI Adoption in the Enterprise (Part 1of 3)", Daniel Faggella, Emerj, April 7, 2019.
[2] "3 barriers to AI adoption across the enterprise", Chris Curran, CIO, November 1, 2017.
[3] "3 barriers to AI adoption across the enterprise", Chris Curran, CIO, November 1, 2017.
[4] "ExplAIned – A guide for executives", Ray Eitel-Porter, Accenture, September 21, 2018.
[5] "AI adoption advances, but foundational barriers remain", Michael Chui and Sankalp Malhotra, McKinsey & Company, 2018.
[6] The age of analytics: Competing in a data-driven world, Nicolaus Henke, Jacques Bughin, Michael Chui, James Manyika, Tamim Saleh, Bill Wiseman and Guru Sethupathy, McKinsey & Company, December 2016.
[7] The age of analytics: Competing in a data-driven world, Nicolaus Henke, Jacques Bughin, Michael Chui, James Manyika, Tamim Saleh, Bill Wiseman and Guru Sethupathy, McKinsey & Company, December 2016.
[8] "AI adoption advances, but foundational barriers remain", Michael Chui and Sankalp Malhotra, McKinsey & Company, 2018.