LONDON, United Kingdom — Clearance sales point to a perennial problem in the fashion industry: the misalignment of supply and demand. Using traditional market research, brands and retailers are unable to predict with high accuracy what products consumers will actually purchase during any given season. As a result, merchandise that doesn’t sell is marked down, while demand for popular items goes unmet, leading to significant loss of income.
But better aligning supply and demand is a complex matter. That’s because, in trend-driven product categories like fashion, historical sales data never results in consistently better commercial decisions. What brands and retailers really require is information about what’s going to happen, not what’s already happened. But traditional fashion forecasting tools like panel-based research and trend reports are slow and unscientific, leaving buyers and merchants to make important business decisions based largely on intuition.
Now, an ambitious London-based startup called Editd — which, earlier this summer, raised a $1.6 million round of seed funding led by Index Ventures, investors in Net-a-Porter, Etsy and ASOS — is offering a realtime data monitoring and analytics platform that makes commercial decision-making in the fashion industry more scientific.
Crawling fashion retail sites, monitoring consumer opinions on social media and analysing output from key industry events, the platform blends machine-learning with human editing to turn vast amounts of raw data, captured in realtime, into the kind of actionable information that can give brands and retailers a competitive edge when making decisions like placing orders, determining pricing and managing merchandising.
BoF spoke with the founders of Editd, Geoff Watts and Julia Fowler, to find out more about how data-driven intelligence is revolutionising retail and reshaping the business of fashion.
BoF: What’s wrong with the way most fashion forecasting works today?
GW: A tangible lack of data and facts, plus the collapse of seasonal fashion is putting a lot of pressure on the way the industry works today. Most businesses have sales reporting or business intelligence to know what is selling, so they already understand the value of data at a trading level. This sales data combined with a great understanding of their customer, inspiration from trend services or their own research is what they use to make an educated guess about where things are going. But even the geniuses can’t get it 100 percent right — otherwise clearance sales wouldn’t exist because everything would sell through!
JF: Seasonal fashion is dead and speed-to-market now is the market — even on the high end. Many brands that work with us are doing 10 or more drops a year, so although the weather is seasonal, fashion is constantly variable. People expect to see new garments on every visit to a store and the production capacity is there to make it happen. Traditional forecasting isn’t a good fit when production can be so close to the market.
BoF: How can technology make this process more scientific?
GW: The cleverest businesses can know exactly what their customers want by using technology. You can measure consumers and the entire trading environment. Customers express themselves constantly online either through Twitter, on their blog, clicking a ‘Like’ button, adding a product to a basket, or buying something. The retail market is measurable — there’s never been more accurate, factual information on exactly what’s happening in realtime than now. It’s an incredible strategic advantage. But the breadth of information out there is too great for people to process and synthesise into actionable information. That’s why we developed Editd.
BoF: Last year, researchers discovered that they could predict, with astonishing accuracy, how well a movie would sell in its first couple of weekends by analysing mentions on Twitter. Can a similar analysis of realtime social data accurately predict demand for fashion products?
JF: Definitely. Though fashion is more nuanced than movie releases. People express opinions about fashion constantly — we have more than 100 million opinions sourced over the past 12 months specifically on individual garments, fabrics, prints and styles. One great example is our data on the longevity of skinny jeans — a trend that endured much longer than traditional forecasting would have predicted. The demand curve was obvious in our data. Making calls on short-term trends based on data is tremendously valuable as well. The ability to know if coloured denim, or leopard print will endure for the next 3 months is vital.
BoF: What types of data should fashion brands be monitoring to generate the most accurate predictions?
GW: Brands should get to know their competition and the full market. Your own sales reporting can’t tell you about something you never produced. Social data should be used beyond the marketing department; buyers and designers should understand what people are saying — it’s an incredibly powerful channel. But good data is useless without good execution. Last week it was 105 degrees in Manhattan and retailers had plenty of notice. Despite that, virtually all summer apparel was on sale and visual merchandising centred around coats and knitwear. It’s a perfect example of lost profit opportunities.
BoF: In a product category as emotional as fashion, to what degree should data drive design, buying and merchandising decisions? Can data-driven intelligence ever completely replace human intuition? What is the right mix?
GW: Some decisions will be handed off to technology, like when to discount, replenish, or what quantities to order. Computing can never replace human creativity, but designers and buyers should always keep their eye on the data — there’s nothing more satisfying than creating a best-seller.
BoF: Who is doing this well today?
JF: Burberry are a great example. They have strong creative direction while blurring the line between being a technology and a fashion company. There’s no doubt that they’ve directly interacted with their customers, understand social and can interpret the whole market. They have short-circuited the risk of production and holding inventory by introducing capsule collections, taking pre-orders before garments are produced, and having iPads in stores to view and order stock that’s not held on-site. Having that much data and being that close to their customers makes traditional forecasting irrelevant.
BoF: How will the rise of data-driven intelligence change the fashion industry in the years to come?
GW: One of the biggest wins will be to reduce wastage, which is an epidemic in the fashion business. We’re excited about the creative benefits too. With production capacity evolving as it is and the ability to understand consumers, we think it won’t be long before the fashion industry can be more experimental and less homogenised, while still being profitable.
Vikram Alexei Kansara is Managing Editor of The Business of Fashion