Fit For Purpose: How Stitch Fit Makes Personalisation Work

By: Tom Ewing, Senior Director, System1Group

For a long time now, it’s been predicted that personalisation will drive the next wave of disruption in e-commerce, as start-ups beat bricks-and-mortar stores thanks to their ability to create highly individualised offers.

But the reality has lagged a little behind the hype. The age of personalisation too often ends up amounting to unwanted retargeting offers and a coupon here and there. Even when this stuff is useful, it’s hardly a game-changing experience, more an engine to drive marginal gains.

Some companies, though, are doing a lot more, taking advantage of machine learning to create genuinely fresh and personalised experiences. One of these is Stitch Fit, whose Brad Klingenberg shared their tips and tricks at OmniShopper. Stitch Fit are a store and stylist combined, selecting and delivering items of clothing to their customers: if you don’t like a selection, you can send it  back at no cost.

Stitch Fit is a beautifully intuitive idea whose execution is fraught with difficulty. Fashion is an area where one size literally doesn’t fit all – the perfect arena for personalisation to make genuine retail breakthroughs. But to keep customers happy, their selections can’t be disappointing. You have to get it right first time, and then get it even more right second time.

To achieve this Stitch Fit rely very heavily on customer feedback. Their idea of commerce is based around feedback loops – the more they learn about you, the better the service can be. But this leads to another hurdle. A service that deals in digital goods – like Pandora or Netflix – can iterate using passive data: it knows what you’ve viewed, bought, listened to and so on. Passive data does the work.

That isn’t so for clothes. Stitch Fit knows what it’s sent out to customers, and what’s been sent back – but that’s very binary, crude data. It doesn’t know how often they wore it, where they wore it, what they thought about it. So it has worked to create a customer service culture where customers are happy to give that feedback.

How do they do it? The secret, says Klingenberg, is self-interest. If the customer trusts the brand enough to believe it when it says their feedback will improve selections next time, they will be eager to give it. It’s an order of magnitude more motivating than the generic promises that service overall will improve which most feedback forms or research surveys offer. Personalisation done right makes customers happy to open a dialogue.

Stitch Fit’s success, Klingenberg told us, is based on a combination of machine learning and a human touch. Machine algorithms come up with suggestions, which human stylists – the company employs several thousand – curate and edit using the customer’s specific feedback. Perhaps the most exciting thing about the business model is the way that the same systems of feedback loops allow Stitch Fit to evolve and optimise its exclusive clothes. In this way the service becomes not just about matching clothes to customers, but actually improving clothes for greater appeal.

Personalisation is a tricky problem for most retailers to solve, and Stitch Fit don’t have all the answers. Klingenberg happily admitted that most of what he’d been talking about was the problem of customer retention, not acquisition: the brand relies on word of mouth for reach. Also, not many sectors are as perfect for this business model as fashion, where the gap between individual fits and mass-market products is so glaring. But the basic principles – mixing machine and human expertise, and designing around feedback – are ones many brands could adopt.