AI for Merchandising Part 2: Why Retail AI Projects Fail

In the dynamic world of retail, where trends evolve at lightning speed and making sense of the data requires more time than anybody has, the allure of AI as a silver bullet for prediction and decision-making is undeniable.

‍

Initially, it’s tempting to envision AI as an omnipotent expert, capable of ingesting vast amounts of data and churning out perfectly tailored recommendations.

‍

This perspective sees AI as a wizard, conjuring up solutions with a flourish that leaves mere mortals in awe.

‍

But what if I told you this is precisely where many AI projects stumble?

‍

Trusting AI is a lot like trusting people

Imagine forking out thousands of dollars for a consultant who requests access to all of your data, but doesn’t spend any time with your team, inquire about your upcoming range or campaigns, or discuss your plans for category extension, new demographics or investment in new materials and silhouettes.

‍

After analysing your data, this consultant leaves the room, comes back in less than a minute telling you exactly how much money they think you can make next year and how much inventory you should buy in each category.

‍

And by the way, it’s “98% certain” of the numbers.

‍

But when you ask how they got to these numbers…you get crickets.

‍

The response from your team is raised eyebrows, looks of exasperation and thoughts of:

‍

Are we really going to be bound to this magical number?

‍

How is this helping us do our job?

‍

What a waste of time, I’m going to have to do this my own way anyway.

‍

If you wouldn’t trust a consultant who operates like this, you’re definitely not going to trust an AI that operates like this.

‍

AIn’t no wizard

AI isn’t (yet) an all-powerful, all-seeing, omnipresent oracle or “wizard” who can plan your business for you.

‍

At best it’s a very, very good calculator, search engine or mimic, but is lacking the context and experience required to bring home real success for your brand.

‍

‍

The creative minds of the people you have hired are the key to how you differentiate yourself in the market: the more people outsource their thinking to artificial intelligence, the more every brand will start to think, act and look the same.

‍

The space that AI most effectively occupies within any business is to allow your team to spend more time decision making by saving time with automation, freeing them to operate more strategically.

‍

Centering the people who matter

If you could hire an absolute gun of an assistant for every key decision maker in your business, would you?

‍

Of course you would!

‍

They could do all the number crunching and research that your team don’t have time to, and could spend hours trawling data to find potential opportunities.

But at the end of the day, they are also going to require guidance and direction, and while helpful, are not going to be left on their own holding the steering wheel.

‍

You’ll also need to make sure that each assistant adapts to the working style of the person they are supporting, and be flexible enough to operate with different levels of trust.

‍

This is the best way to position AI within your business - not as an unquestionable authority but as a transparent, explainable and adaptable assistant.

‍

This approach fosters a collaborative environment where AI and human expertise complement each other.

‍

The team understands and trusts the AI's output not just because they can see the workings behind the curtain, but also because they are not bound by its limitations.

‍

They know that while the AI brings valuable data analysis to the table, their knowledge and experience are crucial in ensuring the rubber hits the road.

‍

Takeaways

Successfully incorporating AI into your business requires asking the right questions:

• What role is this solution trying to occupy in my organisation?
• Does the solution operate as a black box or an interactive process?
• Are there enough opportunities for my team to provide context and guidance?

‍

Progress update on our last post

As a follow up to our last post in this series on similarity - we have delivered.

‍

It’s now easier than ever to stop repeating your mistakes and understand the true performance of similar products you may have ranged in the past, using image recognition to quickly find “that dress” you remember from two seasons back.

‍

In addition, we’ve also enabled deep tagging functionality to give every team the power of rich attribution. Now it’s short work for merchandisers to understand trends in their range that were previously difficult to spot - do you know how white V-necks or leopard print spaghetti straps have performed over time?

‍

To learn more or if you’d like to join our pilot group, please reach out to us at
‍info@stylearcade.com.

‍

‍