The case for fashion-specific AI

Here’s a quick thought experiment: Imagine that you wake up one day with the ability to predict how much interest there will be in something — you become a lightweight fortune teller. But the kicker is, you have to choose: you can either predict demand decently well for many things or you can predict demand very well for only one thing. Which path do you take?

In 2020, Syrup’s co-founders Ferdi and James found themselves in a very similar scenario. Their decision? Become the world’s best at forecasting demand for the fashion and apparel space…as a start.

In honor of our recent inclusion in Activant Research’s report, The Future of Demand Forecasting Software, there’s an opportunity to discuss why we took this path, and where that path leads in the future.

There’s a short answer to the question of why, if anyone is looking for a TL;DR — to paraphrase Ferdi, there’s just no good reason to over-produce anymore. And if that’s true, we should ensure that’s being reflected in the industry most likely to benefit from corrections to over-production.

We have developed a forecasting superpower, and we’re putting it to use where there’s the greatest potential for impact: fashion.

What does it mean to apply a “vertical market focus” to demand forecasting?

In the Activant report, the authors call out Syrup as an emerging winner, specifically citing our vertical market focus. What they’re communicating is important, but nuanced. On the one hand, this means that we offer our product to brands narrowly in the fashion and apparel space—vertical focus here helps explain “who our product is for.”

But it also means that we have designed our product to specifically meet the needs of fashion and apparel brands. This was a multi-year process that can be broken down into three steps.

I. Before doing anything else, our team spent dedicated time listening to folks from across the fashion industry. Building on our team’s experience working within the retail and supply chain spaces, we sought specificity from the individuals working diligently to get clothing in the hands of their customers. Our growing understanding of the challenges faced began to paint a picture of where we should focus our forecasting capabilities.

II. Next, the data scientists got to work. Instead of developing generic models and hoping for the best, the team instead started with actual problems, and from there, identified the machine learning techniques best suited to solving them.

III. Finally, we tested and validated that the foundation we had built actually served the needs of our fashion users. Stakes are high when every season is make-or-break, and we are incredibly grateful to all of the early customers that took a gamble on us. The results spoke for themselves, but this was just the start. Additional nuance gained from practical tests of Syrup fed right back into Step 1, creating a virtuous circle of listening, building, and applying.

So that’s what a vertical focus means — what does it not mean?

It does not mean that we only care about fashion. Our broader mission is to create a world where commerce is no longer wasteful — fashion is just a great place to start.

It does not mean that we will only ever focus on fashion. The same approach we took to solving fashion-specific challenges can be applied again to other retail environments and even along the supply chain.

It does not mean that we only do forecasting. Although the report linked above refers to Syrup as a “pure demand forecasting tool,” the reality is more complex. A forecast alone is only so helpful. And if we only provided our customers with a forecast, we’d be failing to solve the specific challenges they face.

What came up, again and again, in our conversations with folks in fashion is that the planning and allocating processes remain manual and tedious. The most motivated planner or allocator will inevitably struggle if forced to constantly navigate Excel chaos. And even when planning tasks are not hindered by technological shortcomings, forecasts are frequently just good enough reflections of historical sales data.

Syrup’s vertical focus approach, then, doesn’t stop at forecasts. We’ve also designed our product to ease the day-to-day challenges faced by planning teams. To use an analogy: we don’t just predict rain — we also provide the umbrella and raincoat.

The case for fashion-specific AI

In a world quickly becoming saturated with low-effort leasing of AI models, Syrup is championing a more sophisticated approach. If you’re in the planning community and are evaluating AI as a solution, we encourage you to consider that not all AI is the same.

In fact, we have a few guiding principles that shape our approach to model development — an approach that we feel is unique, and we couldn’t be more proud to offer it to fashion and apparel brands.

• We are customer obsessed. The quality of our forecasts is measured in terms of impact on the user. Our customers’ success is the primary outcome we care about.
• We are data skeptics. As the saying goes, “garbage in, garbage out.” Rigorous qualification of data is built into our process, repeatedly.
• We are specialists. Fashion and apparel brands have unique needs that a one-size-fits-all model can’t solve.
• We are lifelong learners. AI is a constantly and rapidly evolving field. Evaluating new techniques for our customers’ specific needs is built into our operating rhythm.

‍