Artificial intelligence (AI) is everywhere at the moment. Much of this is focused around data analytics – crunching business numbers at ferocious speed and as frequently as possible. It is being discussed in terms of its impact on business, employment and human rights across all industries. But in the context of fashion brands and fashion retail, what has it to offer business decision-making? asks Malcolm Newbery.

In order to address this question – and before asking what it might cost – we have to identify what it can or may offer. Many major strategy consultancies have pondered on this. I will begin with a summary of what AI offers the fashion industry, up and down the supply chain.

The AI offer

The consensus appears to be that there are four main areas of business in which AI can advance our skills, and a number of subsets within each. They are:

• Marketing and sales to consumers;
• Understanding competitors’ offers;
• Product choice decisions;
• Supply chains and stock control.

Marketing and sales to consumers

Within this area, AI offers:

GlobalData Strategic Intelligence

US Tariffs are shifting - will you react or anticipate?

Don’t let policy changes catch you off guard. Stay proactive with real-time data and expert analysis.

By GlobalData

Learn more about Strategic Intelligence

• Smart personalisation engines, which in simple terms means that AI knows “what you like and have bought before” and also “what buyers like you have also bought.” For my grocery online shop that means a gentle reminder that “have you forgotten the cereals?” For me in a supermarket, that means a text message as I walk down the cereal aisle that says “do not forget the shredded wheat.” That is commonplace in grocery today. It has not been applied to clothing because we are less creatures of habit.
• Price and loyalty programmes. That means “buy another of this or more and get a discount or a reward (points on a club card).” Once again, this works well in continuity shopping like food or betting, but not so well in fashion.

Understanding competitors’ offers

This is an area that is already well supplied. There are numerous price comparison websites offering consumers knowledge of pricing offers in hotels, travel, insurance etc. But once again, not much on different pricing offers to consumers for the same brand by different retailers.

Product choice and sales forecast decisions

Within this area, AI (using data analytics) suggests:

• What to offer, based on the popularity of different consumer choices for products in different geographical areas;
• How much to offer based on historical popularity.

But yet again, although this works very well with continuity purchases (food, drink, cigarettes, pharmacy and beauty products), it struggles with the vagaries of seasonal and fashion products, where the purpose is to change purchasing patterns.

Supply chains and stock control

Following on from the previous subject, theoretically AI could offer advice on:

• How much to buy;
• From whom to buy it;
• On what delivery timetable;
• How much to stock;
• How to distribute it to stores;
• When to stop selling;
• When to mark down.

This of course after the choice of product, and the decision on the price at which to sell it – the most difficult decision facing fashion management.

The AI problem in fashion

By reviewing what is on offer using AI and data analytics, it becomes clearer what the problem is in the fashion industry. It also becomes clearer as to why the take-up of AI and data analytics has been slower in this sector. The simple answer is that it is the reliance of the fashion industry on the power of change. We are not interested in doing the same over and over again.

But we are interested in understanding:

• Our consumers better;
• Our competitors better;
• Our product and sales forecast choices better;
• Our supply chains better.

So what can we learn from AI to improve decision-making in the fashion industry?

AI uses in fashion in the future

If I have a limited budget to spend on AI and AI related computer systems, what should I prioritise? I suggest the following:

Marketing and sales to consumers

• Smart personalisation engines. Whether I am a physical retailer with stores, or a brand with online sales, I know from my systems what individual consumers have bought. If I know the product categories or brands they have bought in the past, I can target them with personalised messages such as “you like loose fitting jackets” or “you have bought three items from Dries van Noten in the last year” and then offer them something similar.
• Smart personalisation engines. If I know “what buyers like you have also bought”, then I can put into the consumer’s head the idea that they might be intrigued by X, Y or Z. If either of these approaches yields just a 5% increase in sales, the gross margin on an extra GBP1million sales at a 60% gross margin is GBP500,000. That buys quite a lot of software depreciated over three years.
• Price and loyalty programmes. That means “buy another of this or more and get a discount or a reward (points on a club card).” Once again, this works well in continuity shopping like food or betting but not so well in fashion. So I would not spend my AI budget on this element. The experience of the beauty industry (which is more continuity based than fashion, so it should work better) in loyalty programmes like Wowcher is not good. Consumers buy when the discount is heavy and then show that they are not loyal.

Understanding competitors’ offers

This is an area that is already well supplied. But once again, not much on different pricing offers to consumers for the same brand by different retailers. I believe this is, once again, because of the desire of both brands and own label retailers to achieve product differentiation. We do not want to be directly compared on an identical product with competitors. This is because we have always followed the principle of the four Ps of marketing. Differentiate yourself on: Product, Process, People and Price.

And in fashion, price just leads to a race to the bottom. So let’s not go there!

Product choice and sales forecast decisions

I have a real problem with the use of AI in this area. Data analytics suggests that we can offer merchandise (garments) that have been popular before. That is not the point of the fashion industry. We are offering what consumers will desire in the future, not what they are wearing now. If we harness our marketing cart to AI, we will finish up offering bland repeats of what sold well last season. We will be abrogating the responsibility of fashion. Or worse, we will be giving it away to AI.

Supply chains and stock control

If we have concluded that we should:

• Embrace AI for consumer analysis;
• Ignore it for competitor price comparisons;
• Reject it for product and sales forecast decision making;
• then the final area, supply chain, becomes the most interesting.

Assume that we continue to allow humans to decide:

• What to buy;
• How much to buy;
• From whom to buy it;
• What the selling period will be;
• And on what supplier delivery timetable.

Then we can still use AI algorithms (much faster than the human brain) to provide suggestions on:

• How much to stock (in total, in our stores and in our centralised warehouse);
• How to distribute it to stores (both initially and for replenishment);
• How to look at the stock situation against a human created selling plan;
• When, consequently, to mark down.

AI in the fashion industry: summary of conclusions

So should we be talking about how to use AI in fashion and retail? I think we should. In that case, my AI spending budget is going on:

• Specific marketing and sales activities using smart personalised engines;
• Allocation, replenishment, stock management and markdowns.

It is not going to be spent on:

• Price comparisons;
• Product choice decisions;
• Sales forecasts;
• The selling period (window);
• The buying quantities and processes.

To some in the business of selling data analytics, this will appear to be a conservative decision. I think it is a triumph for the human brain. We will decide:

• What we put in the sales offer;
• When we make that offer;
• How and from whom we buy it.

Thus, it is a victory for human buying and merchandising, whilst accepting that AI can help with a lot of the number-crunching detail.