Aptos Engage Digital 2020: The Power of AI
Username: Name: Membership: Unknown Status: Unknown Private: FALSE
In last week’s Retail Paradox Weekly, my RSR partner Paula Rosenblum discussed the importance of product lifecycle management (“PLM”) solutions in the age of COVID-19. In that context, she mentioned that the RSR team had attended the APTOS Engage Digital 2020 “virtual conference”.
Aptos is one of the latest companies to offer a virtual user conference; we’ve also e-visited conferences by Manhattan, Blue Yonder, SAS, and a couple of others in the short time since most of us have been in lockdown mode. Seeing how quickly these virtual user conferences are learning how to use the digital medium presents an interesting case study all by itself. Let’s just say that each one is better than the last, and now the challenge for companies is to not over-produce the sessions, whereas just 4 months ago the danger was in under-producing them. The Aptos event managed to find the right middle ground for the breakout sessions in particular.
Aptos Engage 2020 and AI
Let’s get down to the subject at hand. Of the Aptos Engage Digital 2020 sessions that I attended, one that I found to be particularly interesting was in the “MLM and Planning” track, entitled “Real Comparison, Real Results: The Business Case for AI in Forecasting for Initial Allocation”, that featured David Sheekey, Aptos Product Strategist. The session could have been subtitled “What AI Isn’t And What It Is”, and for decision makers who are still considering the efficacy of AI in Retail, I recommend that you have a listen to the replay.
Dave started with an explanation: “AI – what is it? Basically, it’s a disparate set of technologies – it’s not one thing… effectively, it’s computer-based decision making. Also, it’s a bit of an overworked and overused term, and it’s great marketing.” And just like that, I knew the listener was in for a pragmatic discussion.
As Dave explained, “we already have effective demand forecasting, which runs multiple algorithms and chooses the best ones… we don’t call it artificial intelligence. It’s effectively ‘clever programming’ and it does a huge amount of calculations, but it is formulaic – it doesn’t ‘self-learn’.” That’s an important distinction, and it may help a question of many business leaders’ minds, “do we really need AI?” The answer is, “no”, but self-learning models can make it a lot easier for retailers to optimize their merchandising decisions based on hyper-local trends.
In the session, Dave underlined the Aptos position that they are staying focused on delivering business value first, technical wizardry second. He reminded the audience that merchandise decision making is essentially a human activity (“Humans are … great at finding patterns in things… we simplify complexity really well….”) Dave went on to explain that experienced merchandise planners will sift through lots of transactional data to establish cadence – for example, to organize repeatable events into a buying calendar. Dave termed this activity as “planning as a project”.
But the presenter pointed out that there are limits to what even an experienced merchant can accomplish. Among the stumbling blocks that merchants face are:
- Too much information to process
- Inability to be locally expert
- Missing hyper-local trends
- Missing “tribes”
- A tendency towards risk avoidance
- Too many options
Dave summarized these challenges very simply: “there’s only so much complexity the human decision maker can absorb and act upon… the limits of decision-making cause value erosion – all the time!”
According to the strategist, the question that AI can help retailers answer is, “how can we get better?” Dave pointed out that there is a whole raft of new data that could help planners, if only they could consume the information. That new data potentially includes:
- Geospatial data
- Web data
- Personal information
- Social network data
- Sensor data
- Video; and audio data
“If we can consume more data more efficiently, we can achieve better outcomes – certainly it is true!”, exclaimed the developer. That’s where AI comes into the picture. But Dave pointed out that at least now, there is are relatively few well developed use cases that demonstrate AI’s potential, while there’s also a lack of trust in the technology, a lack of adoption partly fueled by lack of integration, a shortage of expertise, and too much hype. “We looked around at the last NRF event <in New York City last January>, and noticed there was a lot of fuss around the technology, but there were no demonstrable success stories outside of grocery and fast moving consumer goods – which isn’t our <Aptos’> primary focus.”
Dave’s advice to decision makers: “Be careful – it can get expensive… concentrate on the outcomes first, then the technology.”
Aptos is doing just that by infusing the merchandise planning process with insights derived from the use of AI to handle big and diverse data, “from the strategy, through the buying process, through the design process, through the intake decisions, to the allocation, to the replenishment, and through to the clearance decisions.” The strategist shared where and how Aptos is applying data science for prediction, selection, and decision automation (chart):
The speaker went on to describe the product direction in more detail. The session provides an excellent tutorial not only on what Aptos is doing, but what retailers should be doing by infusing AI-enabled analytics into their business processes. So, listen to the replay !
Are Virtual Conferences Becoming ‘A Thing’?
Given the risks associated with humans physically gathering in tight, airless rooms, it’s an easy bet that virtual conferences will continue for quite awhile yet. People may actually learn to like the medium. Aptos did an admirable job of making the main-tent and breakout sessions feel interactive in spite of the obvious fact that they were not. The company managed to avoid the mistake of turning the virtual conference into a collection of infomercials.
But Aptos isn’t in the conference business – its job is to provide solutions to the industry. And if the sessions that Paula and I have summarized in the last two weeks are any indication, the company is making great progress.