SFIX - Stitch Fix

[Guest ajc]

 

Overview

 

Stitch Fix is an online styling service that delivers a box of personalized clothing (on a subscription basis, or a la carte). You fill out a profile on sign-up and are assigned a stylist. Based off of Stitch Fix's selection algorithms, your stylist handpicks pieces around your tastes, fit, and budget. They're mailed to you, and you keep what you like and send back what you don't. You give feedback for each item, kept or returned, on how you liked/disliked the fit, style, etc. Over time, Stitch Fix's AI and your stylist learn your preferences on every metric and deliver an ever more personalized selection.

 

The business is on track to grow FCF by over 100% YoY in 2019 and is selling for 14 times estimated 2019 free cash flow. The shares are down 70% from recent highs, because analysts have a superficial understanding of the company and compare it to Blue Apron. Stitch Fix shares also dropped a lot recently based on slightly below expectation user growth and the general NASDAQ decline.

 

The right way to think about Stitch Fix is as a new, deeply personalized, quality-first model for apparel retail. It's more like a combination of Google Search and Starbucks for clothing instead of some undifferentiated subscription box company. Spotify, Pinterest, and others try find the exact song, coffee table, etc, that fits you. Stitch Fix is changing apparel retail by doing the same for what you wear.

 

Their latest investor presentation - https://investors.stitchfix.com/static-files/be146355-1123-41f8-8128-b690d51d7b73

 

 

User and financial prospects, plus expansion opportunities

 

Stitch Fix is growing users and revenue by 20%+ annually and increasing ARPU at a good clip too. TTM revenue was $1.3B versus their current $1.7B market cap. Some of their stated long-term targets are: gross profit margin of 45%, operating profit as a percentage of revenue of 11%, adjusted EBITDA as a percentage of revenue of 12% (see investor presentation above). Stitch Fix recently expanded into the UK (their second country after the US), so this increased addressable client market should start to show up in 2019.

 

Once Stitch Fix has completed their UK launch execution, the company should be able to take a playbook approach towards similar territories. Places like Canada, Australia, Hong Kong, and Singapore, seem like sensible choices. Western Europe or Japan probably require more nuance, but are also options.

 

Stitch Fix, founded in 2011, has only now started to go beyond their female offering. Men was launched in late 2016, Plus Size in early 2017, and Kids in mid-2018. Those areas still have untapped growth potential (Kids especially since they grow out of clothes fast). Further, because Stitch Fix is the most personalized mainstream offering there's a chance for them to one day expand into big ticket, high margin items like wedding dresses, jewelry, etc.

 

 

If Stitch Fix isn't another Blue Apron, what is it?

Two comparisons....

 

First, Google Search understands your preferences and can predict them while you're typing. Until now, clothing has been mass manufactured to standard fits and patterns by manufacturers, then put on store racks. It's not personalized and the store has no memory of you. Stitch Fix builds up a profile of your exact preferences, so like Google Search the more you use it the better it gets.

 

While Amazon often suggests bizarre additions to your basket based on their algorithm, Stitch Fix actually knows what it's doing. Amazon Prime Wardrobe is seen as Stitch Fix's biggest competition, but Amazon has a long way to go even with their regular recommendation AI. Like Google Search, Stitch Fix has a virtuous circle at work. More clients means more feedback, more feedback data allows it to get more accurate at predicting preferences. A reputation for accurate prediction drives brand awareness and attracts new clients, and so on and so forth.

 

Starbucks is another worthy comparison. They originally persuaded consumers to pay a premium for overall experience, a good quality cup of coffee, and some snob appeal. Their brand positioning allowed them to not have to compete too heavily on price. Starbucks is standard today, but there was a time at the beginning when most considered it crazy to shell out $5 for their coffee. Howard Schultz figured that society's spending power would catch up and the Starbucks experience would eventually be appreciated and democratized. Lake has been smart about positioning the Stitch Fix brand in roughly the same way, so Stitch Fix has a long-term opportunity to maintain a premium valuation over its peers.

 

Stitch Fix is also uniquely innovative in that every client now has their own personal tailor. Users will often tell their stylists about the child they're expecting, wedding dates, etc, so the stylist knows exactly what to provide and the relationship can become deeper and more long-term as a result. This is reciprocal in that stylists send notes to clients with each fix too. There was a time when only the rich could afford their own private tailor, but Stitch Fix is scaling that. It's a luxurious idea, but like Netflix, Spotify, etc, consumers in future will expect fully personalized clothing offerings on size, fit, style, etc, and Stitch Fix is currently best-in-class at this.

 

 

Financial performance, economics of the business, and competitive advantages

Stitch Fix has been profitable almost since the start. It has no debt. It's unclear how long they'll keep up their current strong FCF growth, but it's inexpensive based on where FCF will be for 2019 together with their revenue and user growth opportunities.

 

They don't have the real estate expense often associated with retail. That also means no costly need to update and remodel your stores regularly. Stitch Fix does have to keep setting the standard with their core business proposition, but their model and brand positioning means they get to avoid much of the crappy economics retail is known for.

 

As mentioned previously, their overall dataset and ability to analyze it gives them the opportunity to become almost a Google Search (with mapping capability) for clothing. As long as they continue to lead there, they can build a competitive advantage around that. It'll enable them to attract more clients and they'll also be able to attract the best clothing brands since those brands will want to be in front of the most discerning consumers.

 

Stitch Fix manages inventory better than others since their data gives them a far clearer idea of what people will buy. That means there's less waste and capital tied up in inventory. While Amazon is a low-cost, quick shipper, Stitch Fix is not a discounter and has a brand known for personalization and client service. This also allows Stitch Fix to offer itself to brands as a platform that maintains some sense of exclusivity and pricing power.

 

Stitch Fix could therefore continue positioning itself as a higher-end, truly personalized platform and reap the benefits. For many consumers and brands, having a classier, higher quality alternative to Amazon would be both competitively useful and financially valuable. Stitch Fix would end up with the various economic and competitive advantages that platform status would bring.

 

 

Management, culture, and data science

Katrina Lake is a highly intelligent, driven, and grounded CEO with 10%+ ownership. You can get an idea of her business instincts and mind at work by listening to these discussions -

& https://soundcloud.com/adam-berke/target-audience-podcast-episode-4-katrina-lake. She recruited Eric Colson from Netflix in 2012 to head up algorithms and data science, as well as folks like Mike Smith, Cathy Polinsky and Paul Yee to the management team. VC's like Steve Anderson and Bill Gurley were also both persuaded to invest even though she was an unproven quantity at the time with no experience running a retail business.

 

Culturally, Stitch Fix is known for having an open, friendly work environment with an emphasis on hiring people who know their business inside out and can execute plus advance the company into new areas. They employ mostly women and are obviously female led. This might give Stitch Fix a slight advantage in recruiting top female talent, since Silicon Valley can sometimes be a slightly challenging place for women to find safe, equal working conditions.

 

Data is vital across the organization and it's used not just for clothing algorithms, but also for warehouse and inventory efficiency, as well as product testing among other things. Stitch Fix also employs more data science PhD's than any other pure apparel retailer and is a favored destination for female STEM graduates. For those interested in going into the weeds, I highly recommend the Stitch Fix Engineering and Data Science blog (https://multithreaded.stitchfix.com/blog/) to see how data is used and what it's used on across the company. Basically, every aspect of the business is analyzed in order to create greater efficiency and better customer outcomes. This focus will likely pay off nicely over the long-term.

 

 

Risks and negatives

As an investor with an owner-oriented mindset, I wish their emphasis on FCF was more front and center in their financial reporting and letters. That said, I'm glad they cover it and am happy overall with what they focus on.

 

Amazon Prime Wardrobe is their main competitor even though it focuses on delivery speed and price instead of personalization. This, together with Amazon's AI difficulties, will probably ensure that Amazon stays behind Stitch Fix in this market even if they'll have more volume. Stitch Fix clearly can't rest though since Amazon is know for plugging away at weaknesses for years until they're solved.

 

Katrina Lake owns 10%+ of the shares which is highly encouraging, but as with many Silicon Valley companies she also has super voting shares. While this is almost standard these days for many new companies in tech or related industries, I'd prefer not to see it. It does prevent Stitch Fix being bought out (and screwed up) though, for a cheap price relative to the long-term value they'll likely create.

 

Glassdoor ratings for Stitch Fix have recently trended lower from 3.8/5 to 3.3/5. While retail is a tough industry so this is understandable, I think they need to put effort into addressing the root of this.

 

Stitch Fix has also chosen to operate in an industry where yesterday is past and clients only care about how good you are today. As soon as they stop continuously improving their core offerings, or pay less attention, or any number of things, they'll create space for others. While that doesn't mean Stitch Fix should throw away their intelligent approach of growing and expanding thoughtfully, it does mean complacency is clearly a non-starter.

 

Finally, founder/CEO Katrina Lake is integral to the company going forward and so there's key woman risk. If she were hit by a bus tomorrow or her performance dropped off noticeably over time, that would materially impact the ability of Stitch Fix to continue to lead their market. The same likely goes for Eric Colson and potentially Mike Smith from the management team. Cathy Polinksy, Paul Yee, and 1 or 2 others, would also take time to replace.

 

 

Recommended reads

https://hbr.org/2018/05/stitch-fixs-ceo-on-selling-personal-style-to-the-mass-market

 

https://www.forbes.com/sites/ryanmac/2016/06/01/fashionista-moneyball-stitch-fix-katrina-lake/

 

https://www.reforge.com/blog/stitchfix-personalization-retention-monetization

 

https://www.goodwatercap.com/thesis/understanding-stitch-fix

 

https://techcrunch.com/2017/10/22/unboxing-stitchfixs-s-1/

 

 

 

[clutch]

"This, together with Amazon's AI difficulties, will probably ensure that Amazon stays behind Stitch Fix in this market even if they'll have more volume."

 

What's your basis for saying that Amazon has AI difficulties? Amazon probably hires 10x more PhDs in machine learning than this company... considering they also have more volume (data) it seems like they'd have much bigger advantage in AI.

[KJP]

Have you tried the service?  I did and found the curation poor (sending me clothing types I said I did not want), overpriced and, in several instances, not unique, e.g., a sweater from Banana Republic.  I've asked a few others who tried the service and they had the same impression.  I tried them a couple of years ago, so perhaps they've improved.

 

Also, there are other competitors, such as Trunk Club, but I'm not sure how well any of them are doing. 

[johnny]

I'm not sure I buy the data/moat argument here. I question how sophisticated such an operation can be, given how costly data point collection (Netflix can pull about 100 signals from you cruising around the carousel, at zero cost). The attached graphic in the presentation doesn't exactly impress me.

 

Data-y moats start to build when you begin to know things about the user that they don't know or aren't factoring in correctly. "Dana is 172.7 centimeters tall" is a fine enough thing to know, but it's not exactly the Colonel's Recipe, and I'm not sure how much proprietary signal they can credibly claim to have on Dana that prevents FitchStix from coming around and peeling off a chunk of their users with some novel gimmick.

 

[Guest ajc]

 

"This, together with Amazon's AI difficulties, will probably ensure that Amazon stays behind Stitch Fix in this market even if they'll have more volume."

 

What's your basis for saying that Amazon has AI difficulties? Amazon probably hires 10x more PhDs in machine learning than this company... considering they also have more volume (data) it seems like they'd have much bigger advantage in AI.

 

 

So, in 2015 Stitch Fix actually had a bigger data science team than Apple, LinkedIn, Twitter, Google or Amazon (https://www.mercurynews.com/2016/12/15/computing-out-fit-stitch-fix-algorithms-machine-learning-dress-customers/) and that team has grown by over 30% since then. Furthermore, data science and AI is part of the Stitch Fix core and is what the company was built around.

 

Amazon, while strong in that area, was really built as a shopping website backed by an innovative warehouse and logistics operation (that was my sense when I worked there). Obviously, they later opened their physical and computing infrastructure to others.

 

While I'd argue Amazon is pretty decent on the AI front, to my mind their forte was much more in building practical industrial size hardware and software systems that could scale quickly and robustly. My 2 cents is when it comes to nuance, Amazon is not really the same as a Google or a Stitch Fix. It's like comparing Noma to McDonalds. Both do what the other can't, but if you figured out a way of scaling Noma it'd sound odd to say McDonalds has the better food production techniques. It depends on what you're after and for Stitch Fix there are enough people who want a more deeply personalized, stylist-based experience.

 

I'm not sure if you're aware, but Amazon Prime Wardrobe is not like Stitch Fix even though analysts peg it as their greatest threat (incorrectly to my way of thinking). With Prime Wardrobe you click on a bunch of items you like, choose your own sizes, styles, etc, and they get shipped to you. There's no personalization and Amazon has no deep preference data for you. Essentially, it's more of a quick and easy shipping service on top of the current retail store format where there are racks full of stuff you sort through yourself. Amazon is not really in the business of intricate personalization, and as good as they are generally it might be tough to teach an adult dog such new and different tricks.

 

A few obvious examples of where Amazon falls darn short on AI ability for the consumer side, can be found here - https://worldwideinterweb.com/funniest-amazon-recommendations-gallery/. I don't doubt all of us have experienced that. Also, if you buy a toaster they'll recommend you a hundred different other types of toaster over the following month. That's another one most people can relate too. Then there's the clear problem they're currently having with fake reviews which their AI is not good enough to pick up and delete. Not to be too hard on Amazon here, but some of this stuff is just about getting the basics right.

 

Anyway I don't think Amazon is terrible or Stitch Fix is perfect, but I'd say for anyone who looks at who has the better consumer facing recommendation AI, it's clear Amazon still has some completely obvious stuff they need to take care of before they can be taken really seriously on that front.

 

This is a great presentation about where Stitch Fix's currently is on this -

 

 

 

[Guest ajc]

 

Have you tried the service?  I did and found the curation poor (sending me clothing types I said I did not want), overpriced and, in several instances, not unique, e.g., a sweater from Banana Republic.  I've asked a few others who tried the service and they had the same impression.  I tried them a couple of years ago, so perhaps they've improved.

 

Also, there are other competitors, such as Trunk Club, but I'm not sure how well any of them are doing.

 

 

Ha. I can't since I'm not from the US, but I know you're not the only one who that's happened to. Since 2013, my Stitch Fix research on the customer satisfaction and comparison side has been through blogs, YouTube videos, fashion sections of national newspapers, clothing magazines, etc. I've gone through a couple hundred of them to get a sense of the average consumer's attitude towards Stitch Fix with a focus on those articles where they'll compare a bunch of services that do the same thing. I also get a daily Stitch Fix alert from Google with the top five or ten news stories, blogposts, etc, about the company.

 

Over time I've been able to build up a pretty solid sense that Stitch Fix is the best bet in the space. Clearly the rebuttal could be I've not used it so that's a point against, but I tend to think I get to look at the numerous reviews and data instead of relying on my individual positive/negative experience. I guess there are pros and cons to both.

 

Like you say, there's Stitch Fix, Trunk Club (acquired by Nordstrom), Dia & Co (for plus size), Wantable, and others in the space. I've noticed Stitch Fix is usually the most favored by a clear margin and because of/related to that it's also been the best at scaling (see images attached below). Also, because Stitch Fix does women, men, and kids, I think it offers working mothers the chance to do all their shopping in one place. That's a practical advantage competitors might overlook.

 

To be more specific about onboarding and curation, I'd say your experience is a problem for the business model. That goes for the whole space though, not just Stitch Fix. There's an element of randomness and that is strongest in the beginning. In a perfect world, everyone would stay with the service for 6 months for their preferences to become far more obvious. Clearly there's not a chance in hell that's going to happen.

 

The approach as far as I can tell is for Stitch Fix to simply be better on average than the competition in the start and through the customer's lifetime. There will be unhappy clients who leave after mediocre experiences, but at this stage of the technology and business model I think it's more of a numbers game where you try beat the odds and maintain a strong net promoter score. From an investment point of view, it sucks to hear that happened to you and from the same point of view it's also completely understandable given there's randomness in the model and that means bad initial experiences for a certain amount of people.

 

Interestingly, this is a common thing for a lot of tech that I can think of. These days and in the past. There was a time when Amazon had Xmas screw-ups with late deliveries though they were always much better than the competition on average. Nowadays, Uber and Lyft started out with some really terrible reviews and so riders would switch services. Same with UberEats, Doordash, GrubHub, and Postmates for food delivery. Some of the reviews really lay into them and swear you should use X instead of Y. Scroll down a little and someone else is saying the exact opposite with just as much dissatisfaction.

 

I think it's somewhat in the nature of a lot of these businesses that there's a given amount of unhappiness in the start. For Stitch Fix, I'd say it's more likely because of the styling fee and the sense a client might end up empty-handed. There's two ways to look at that perhaps. One is it's better to stick to businesses where that stuff's been ironed out and the experience is good from the start almost always (say Uber these days, just for example). Another way to look at it is that Stitch Fix must be pretty good if they can succeed in a business where the first few months are actually that hard to make a customer conversion in. I tend towards the first, but I admire what Stitch Fix has achieved given the clear challenges.

 

[Guest ajc]

 

I'm not sure I buy the data/moat argument here. I question how sophisticated such an operation can be, given how costly data point collection (Netflix can pull about 100 signals from you cruising around the carousel, at zero cost). The attached graphic in the presentation doesn't exactly impress me.

 

Data-y moats start to build when you begin to know things about the user that they don't know or aren't factoring in correctly. "Dana is 172.7 centimeters tall" is a fine enough thing to know, but it's not exactly the Colonel's Recipe, and I'm not sure how much proprietary signal they can credibly claim to have on Dana that prevents FitchStix from coming around and peeling off a chunk of their users with some novel gimmick.

 

 

So, Eric Colson (Chief Algorithm Officer, Stitch Fix) worked at Netflix for 6 years and was their VP of Data Science and Engineering. According to him, Netflix doesn't come close to Stitch Fix in terms of data science & AI.

 

See 6min00secs for that quote, but anyone who wants to understand Stitch Fix should likely watch the whole thing -

 

 

 

[KJP]

 

Like you say, there's Stitch Fix, Trunk Club (acquired by Nordstrom), Dia & Co (for plus size), Wantable, and others in the space. I've noticed Stitch Fix is usually the most favored by a clear margin and because of/related to that it's also been the best at scaling (see images attached below). Also, because Stitch Fix does women, men, and kids, I think it offers working mothers the chance to do all their shopping in one place. That's a practical advantage competitors might overlook.

 

I also have personal experience with Trunk Club ("TC") and thought it was worse than Stitch Fix ("SF").  TC had the same curation issues and was even more overpriced than SF, in part because they're shipping Nordstrom's private labels, which I believe are more expensive than SF's.

 

I agree with you that curation should improve over time, which should also help margins by lowering returns.  But I was concerned about pricing.  SF's current and long-term target gross margins are around 45%.  If I recall correctly, Nordstrom, Macy's, GAP and Urban Outfitters, on the other hand, have gross margins in the mid- to high-30's.  What drives the difference? 

 

Part of it could be accounting in that the brick-and-mortar retailers may include store operating costs in COGS (I haven't checked).  But my sense from the personal experience with SF was that they were charging full price, while other retailers have to discount.  I know SF has introduced some of their own labels, and it's easier to get away with higher GMs if  people cannot price compare.  But if you're selling third-party items that customers can find cheaper elsewhere online, how sustainable is that?  Or perhaps the question is how big a market is there for that kind of service?

[bizaro86]

I could see them getting relatively significant discounts from fashion labels at a certain scale as a way to "push" product. It also seems likely they would be uniquely placed to buy fashion close outs and sell them at full price.

 

So the improved margins could be in the cost side of the ledger as well as the pricing side.

 

Thanks for the idea, I had heard of them previously but just assumed they were venture/private.

[KJP]

I could see them getting relatively significant discounts from fashion labels at a certain scale as a way to "push" product. It also seems likely they would be uniquely placed to buy fashion close outs and sell them at full price.

 

So the improved margins could be in the cost side of the ledger as well as the pricing side.

 

 

If it's a scale benefit, then GM should increase as revenue increases.  That's not the case.  They've had steady GM (44%) over the last three years despite a near doubling of revenue.  See slide 16 here:  https://investors.stitchfix.com/static-files/be146355-1123-41f8-8128-b690d51d7b73 

Also, the long-term management target model projects 45-46% GM, so no big scale benefit on the GM line even at maturity.  [see slide 19]  Instead, future profit margins will be driven by (according to management) SG&A leverage.  Thus, management sees scale benefits showing up further down the income statement, rather than via GM.

 

Your other hypothesis -- that fashion labels use them as a way to get rid of excess inventory -- is possible.  But if that's true, it would only increase my concerns about being able to sell at full price, because there's a greater risk that the same product is available much cheaper through another channel.

[KJP]

I took a quick look at the Stitch Fix and Nordstrom 10-Ks and their reported gross margins are not apples-to-apples.  The details are below.  So, is my speculation that SF is selling a higher percentage of its clothes at full price even correct?

 

********

 

Nordstrom's COGS include "buying and occupancy costs," which are defined as follows: "Buying costs consist primarily of compensation and other costs incurred by our merchandising and product development groups. Occupancy costs include rent, depreciation, property taxes and facility operating costs of our retail, corporate center, fulfillment facilities and distribution operations."

 

In contrast, Stitch Fix appears to record most of those costs SG&A.  Specifically, SF limits COGS to "the costs of merchandise, expenses for shipping to and from clients and inbound freight, inventory write-offs and changes in our inventory reserve, payment processing fees and packaging material costs." 

 

SF includes the following in SG&A:  "compensation and benefits costs, including stock-based compensation expense, for our employees including our stylist, fulfillment center operations, data analytics, merchandising, engineering, client experience, marketing and corporate personnel. Selling, general and administrative expenses also include marketing and advertising, third-party logistics costs, facility costs for our fulfillment centers and offices, professional services fees, information technology and depreciation and amortization."

 

So, on an apples-to-apples basis, the gross margin difference between SF and other retailers may not be as large as my prior post suggested, though SF must be recording in COGS much higher "shipping to consumer" costs as a percentage of revenue than Nordstrom.

[Guest ajc]

 

 

So, on an apples-to-apples basis, the gross margin difference between SF and other retailers may not be as large as my prior post suggested, though SF must be recording in COGS much higher "shipping to consumer" costs as a percentage of revenue than Nordstrom.

 

 

 

You're right. TechCrunch analyzed the same thing in their IPO breakdown - https://techcrunch.com/2017/10/22/unboxing-stitchfixs-s-1/

 

Goodwater Capital also did a detailed Stitch Fix write-up on their model and the overall industry, with comparison data included - https://www.goodwatercap.com/thesis/understanding-stitch-fix

 

I think Stitch Fix hybrid designs (in-house productions that are among their best-sellers) could potentially be a big deal for the business and margins down the road (38min30secs in this video -

).

That'd be a way for them to make themselves more self-determining like a Zara, rather than relying on outside brands. Their individual preference data for fit, style, etc, would clearly be a significant differentiator and advantage if that happened.

 

 

 

[clutch]

 

"This, together with Amazon's AI difficulties, will probably ensure that Amazon stays behind Stitch Fix in this market even if they'll have more volume."

 

What's your basis for saying that Amazon has AI difficulties? Amazon probably hires 10x more PhDs in machine learning than this company... considering they also have more volume (data) it seems like they'd have much bigger advantage in AI.

 

 

So, in 2015 Stitch Fix actually had a bigger data science team than Apple, LinkedIn, Twitter, Google or Amazon (https://www.mercurynews.com/2016/12/15/computing-out-fit-stitch-fix-algorithms-machine-learning-dress-customers/) and that team has grown by over 30% since then. Furthermore, data science and AI is part of the Stitch Fix core and is what the company was built around.

 

Amazon, while strong in that area, was really built as a shopping website backed by an innovative warehouse and logistics operation (that was my sense when I worked there). Obviously, they later opened their physical and computing infrastructure to others.

 

While I'd argue Amazon is pretty decent on the AI front, to my mind their forte was much more in building practical industrial size hardware and software systems that could scale quickly and robustly. My 2 cents is when it comes to nuance, Amazon is not really the same as a Google or a Stitch Fix. It's like comparing Noma to McDonalds. Both do what the other can't, but if you figured out a way of scaling Noma it'd sound odd to say McDonalds has the better food production techniques. It depends on what you're after and for Stitch Fix there are enough people who want a more deeply personalized, stylist-based experience.

 

I'm not sure if you're aware, but Amazon Prime Wardrobe is not like Stitch Fix even though analysts peg it as their greatest threat (incorrectly to my way of thinking). With Prime Wardrobe you click on a bunch of items you like, choose your own sizes, styles, etc, and they get shipped to you. There's no personalization and Amazon has no deep preference data for you. Essentially, it's more of a quick and easy shipping service on top of the current retail store format where there are racks full of stuff you sort through yourself. Amazon is not really in the business of intricate personalization, and as good as they are generally it might be tough to teach an adult dog such new and different tricks.

 

A few obvious examples of where Amazon falls darn short on AI ability for the consumer side, can be found here - https://worldwideinterweb.com/funniest-amazon-recommendations-gallery/. I don't doubt all of us have experienced that. Also, if you buy a toaster they'll recommend you a hundred different other types of toaster over the following month. That's another one most people can relate too. Then there's the clear problem they're currently having with fake reviews which their AI is not good enough to pick up and delete. Not to be too hard on Amazon here, but some of this stuff is just about getting the basics right.

 

Anyway I don't think Amazon is terrible or Stitch Fix is perfect, but I'd say for anyone who looks at who has the better consumer facing recommendation AI, it's clear Amazon still has some completely obvious stuff they need to take care of before they can be taken really seriously on that front.

 

This is a great presentation about where Stitch Fix's currently is on this -

 

Things have changed since last time you were in Amazon:

https://www.wired.com/story/amazon-artificial-intelligence-flywheel/

https://www.forbes.com/sites/blakemorgan/2018/07/16/how-amazon-has-re-organized-around-artificial-intelligence-and-machine-learning/#7d38a68a7361

 

The important thing is that they have all of the data, infrastructure, and talent. There's then the AWS factor. If any company wants to implement an AI/data science pipeline in a scalable manner, probably the easiest option nowadays is to implement the solution on AWS using their services and Lamda/Step functions. Even Stitch Fix uses AWS:

https://www.cnbc.com/2017/10/19/stitch-fix-amazon-competitor-depends-on-aws.html

 

Having said all these, I don't have much insight into Stitch Fix. Maybe they do have a better capability to develop AI solutions. And who knows, Amazon might end up buying Stitch Fix in the end, which would be a good outcome for the investment.

[cameronfen]

It's strange how when a company like stitch fix slaps on the title of data science people buy the company like it's making magic.  As a data scientist, it's basically just like statistics and my guess is it only helps the company on the margins for this task.  I imagine it basically works somewhat worse than a human for picking out clothes you like.  If you assume the robot saved a person 10 mins of hand curation for each order, that's 2.66 an order if you pay 2x minimum wage to a worker.  For orders over 100 dollars, that's a pretty insignificant cost advantage. 

[KJP]

 

 

So, on an apples-to-apples basis, the gross margin difference between SF and other retailers may not be as large as my prior post suggested, though SF must be recording in COGS much higher "shipping to consumer" costs as a percentage of revenue than Nordstrom.

 

 

 

You're right. TechCrunch analyzed the same thing in their IPO breakdown - https://techcrunch.com/2017/10/22/unboxing-stitchfixs-s-1/

 

 

The TechCrunch article's analysis is hard to follow.  It says that traditional clothing retailer gross margins [which the article appears to mistakenly equate to "inventory margins"] are around 45%, but the article's chart shows that the truth is closer to mid- to upper-30's.  The only ones above that are brands [Michael Kors, LVMH, Hermes, Coach] rather than pure retailers.  So, the article ought to start from the proposition that SF's reported GMs are higher than other clothing retailers and explain why that is.  But instead, the article seems to then ignore other clothing retailers and assert that SF ought to be compared to other "mature e-commerce retailers," which the article claims have GMs of 55-65%.  Unfortunately, the article contains no information about who these purportedly comparable "mature e-commerce retailers" are nor does it explain why GMs should be the same across all kinds of e-commerce, regardless of the type of product being sold. 

 

If you're concerned about whether SF's model is based on full-priced sales, then it does make sense to try to see whether (Price received from consumer - Price paid to manufacturer/brand for product) -- which the article seems to occasionally refer to as "inventory margin" -- is larger for SF than, say, Nordstrom.  But the article's analysis doesn't do that, because it only adjusts out SF's shipping costs, but doesn't also attempt to adjust out the "buying and occupancy costs" that retailers like Nordstrom typically record in COGS.  So, the article's analysis is incomplete and misleading if that's what it's trying to analyze.

 

So, at the end of the day, I don't think the TechCrunch article provides any answers to the underlying question of whether SF's business depends more on full-price selling than other clothing retailers.

 

[dwy000]

Maybe it's just me but the business model doesn't feel very robust.  Spending $25 to have someone (or a machine) pick out full priced clothing and then having to rebox and take the extras back to the post office doesn't feel like it's a massive time or cost savings (vs just shopping online).  I could see people who don't like shopping using it as an expensive timesaver but I would imagine for most people its a novelty to try and then ditch.  Would love to know the drop off rate of customers from first order to second to third etc.

 

Also not sure what the moat would be here.  A taste based algorithm that gets better over time requires a ton of orders and is impossible to compare unless you are using multiple services.

 

Maybe I'm missing the whole plot but it feels like a neat but expensive and unnecessary service that will immediately get crushed in any sort of downturn.

[Guest ajc]

 

 

So, at the end of the day, I don't think the TechCrunch article provides any answers to the underlying question of whether SF's business depends more on full-price selling than other clothing retailers.

 

 

 

I see. I think what you're talking about is being addressed by them from another angle (see https://www.businessoffashion.com/articles/news-analysis/stitch-fix-introduces-over-100-contemporary-brands).

Stitch Fix is using client preference insights and combining with brands to come up with unique lines of fashion that are only sold there.

This, together with Stitch Fix's own in-house hybrid designs, makes it look like they're solving this issue by offering more and more collections that can only be found through them.

The more offerings Stitch Fix sells that are unique and the more brands that use Stitch Fix as a preferred channel for those products, the higher up the chain they can go.

 

This is another good article on the same thing, focusing deeper on how Stitch Fix and the brands specifically benefit - https://medium.com/@medhaa/stitch-fix-and-amazon-continue-to-dominate-apparel-e-commerce-dcda1a68c0e1

Essentially, Stitch Fix seems to be aware of the risk you're pointing out and is doing what's needed to protect itself from the threat of being just another high-endish channel.

I think in some ways they're actually building leverage over competitors and the industry by using their data and personalization, instead of being on the receiving end.

 

Related, it's also probably worth flipping the issue of full-price selling and how it's looked at here. I don't doubt you're aware it definitely matters if you're undifferentiated, but if you're way more differentiated and personalized then full price selling is something many clients and brands actually want and expect. Fashion's definitely one of those weird businesses where price itself can sometimes be the most important signal so long as the cache you've built is exclusive and high-end, so I guess I'm trying to keep that in mind for this bet. Stitch Fix sort of looks like it's busy positioning itself in that way.

 

 

 

[KJP]

 

 

So, at the end of the day, I don't think the TechCrunch article provides any answers to the underlying question of whether SF's business depends more on full-price selling than other clothing retailers.

 

 

 

I see. I think what you're talking about is being addressed by them from another angle (see https://www.businessoffashion.com/articles/news-analysis/stitch-fix-introduces-over-100-contemporary-brands).

Stitch Fix is using client preference insights and combining with brands to come up with unique lines of fashion that are only sold there.

This, together with Stitch Fix's own in-house hybrid designs, makes it look like they're solving this issue by offering more and more collections that can only be found through them.

The more offerings Stitch Fix sells that are unique and the more brands that use Stitch Fix as a preferred channel for those products, the higher up the chain they can go.

 

This is another good article on the same thing, focusing deeper on how Stitch Fix and the brands specifically benefit - https://medium.com/@medhaa/stitch-fix-and-amazon-continue-to-dominate-apparel-e-commerce-dcda1a68c0e1

Essentially, Stitch Fix seems to be aware of the risk you're pointing out and is doing what's needed to protect itself from the threat of being just another high-endish channel.

I think in some ways they're actually building leverage over competitors and the industry by using their data and personalization, instead of being on the receiving end.

 

Related, it's also probably worth flipping the issue of full-price selling and how it's looked at here. I don't doubt you're aware it definitely matters if you're undifferentiated, but if you're way more differentiated and personalized then full price selling is something many clients and brands actually want and expect. Fashion's definitely one of those weird businesses where price itself can sometimes be the most important signal so long as the cache you've built is exclusive and high-end, so I guess I'm trying to keep that in mind for this bet. Stitch Fix sort of looks like it's busy positioning itself in that way.

 

I agree that differentiation is essential to full price selling.  I think that's particularly true with clothing, where there are huge price differences between substitute goods driven, in part, by branding.  So, I think what SF is doing is smart.  QVC tries to do the same thing by getting manufacturers to make versions of their product that are exclusive to QVC.

 

I initially went down the rabbit hole of GM because I was trying to figure out if SF's business model was based on full-priced selling.  After doing more reading (10-K, articles you've linked to, VIC comments), it seems clear that full-price selling is the company's core business model.  It will be interesting to see how big the business can get with that model.

 

Given the very high churn rates suggested by multiples sources (company comments, credit card data referred to in VIC comments, the Goodwater Capital presentation you linked to early), it seems to me that the more relevant financial metrics for SF are the ones you'd use for a traditional retailer, rather than CAV/LTV and related measures that make more sense for enterprise software companies.  On the surface, SF's metrics (profits, operating margin, etc.) appear to be getting worse.  Why is that?  Does the company have to expense "investments" that would ordinarily be capitalized by a traditional retailer? 

 

 

[Guest ajc]

 

 

Given the very high churn rates suggested by multiples sources (company comments, credit card data referred to in VIC comments, the Goodwater Capital presentation you linked to early), it seems to me that the more relevant financial metrics for SF are the ones you'd use for a traditional retailer, rather than CAV/LTV and related measures that make more sense for enterprise software companies.  On the surface, SF's metrics (profits, operating margin, etc.) appear to be getting worse.  Why is that?  Does the company have to expense "investments" that would ordinarily be capitalized by a traditional retailer?

 

 

 

I can't find the quote (maybe it's apocryphal), but there's a Bill Gates line that says noteworthy shifts in computing happen every three years. Maybe e-commerce is the same.

Anyway, my estimation - since I can't be sure - is Stitch Fix was similarly hitting the limits of its existing model in 2017 and 2018 based on what I can figure.

Too many experiences like the one you had, market entry by Amazon Prime Wardrobe, etc, meant they had to take a serious look and evaluate what parts of their business model had got them to public company status. They had to decide what still worked and which parts were not worth continuing with if they wanted to stay competitive and maintain the quality of their business and financials.

 

I think the end of the line for this older, more simplistic model of theirs was what caused the dip. Their new model was then a start-up within a start-up and that's the new, differentiated Stitch Fix that's emerging. I expect this to happen every few years. Basically, I'd say the old model got to its limits based on what they were willing to spend to acquire customers and they found at a certain point the new customers didn't want to spend as much, were fussier, etc.

I think they had moved past their hardcore base and so they 'scaled' and pretty much went downmarket (ie. broader and lower). At that stage, they might've seen the declines you're talking about, noticed they weren't differentiated enough (Trunk Club, seeing the Amazon offering coming, etc), and taken a step back to do a rethink of what they'd need to do for the next few years in order to stay ahead of the curve.

 

To me, this is also what you have to do if you want to be a top growth company so I expect to see these fluctuations (ie. buying opportunities of the Phil Fisher variety) and frankly I'm optimistic about them because I like seeing Lake and company do this type of thing. If they just played out their hand and never changed up significantly while the competition did, there's no way I could invest.

 

Essentially, if you go back roughly a year you see Stitch Fix starts to leave behind the less differentiated business model and announce their Hybrid Designs as well as their personalized, exclusive lines with more high-end brands that was written about in that article.

Also, Intimates was released about a year ago and Kids a little after that in order to get moms hooked by bringing their children into the mix (after all, if the kids like it and get used to it then it's likely harder for mom to one day unsubscribe).

On top of that, Stitch Fix brought on Deirdre Findlay from Google as Chief Marketing Officer and Donna Boyer from Airbnb as VP of Product in the last year. Presumably to make sure their push into a number of new, differentiated brand and design directions gets executed on.

 

I'd say Stitch Fix has really been strategically reinventing and repositioning itself over the last year and a bit (after realizing they were moving relatively inadvertently towards worse places on the value chain instead of better ones), and once that newly created core starts to take hold and permeate through the existing (ie. old) Stitch Fix business, the value of it will start to show and the financials will reflect that.

It's a little like the Bezos line about how when an analyst says 'good quarter' he's sort of surprised because really this good quarter is a reflection of the work they focused on 3 or 4 years ago.

That's my best way of thinking about it right now.

 

(Alternatively, that's all one big and extravagant (if somewhat impressive, though I say so myself) rationalization. Caveat emptor as per usual with all this stuff. Appreciate the thought-provoking discussion. Happy 2019.)

 

 

 

[SHDL]

Thanks much for the idea.  Still researching, but so far I have a good feeling about this one.  At least I think I can see the upside potential.

 

To add a bit to the ongoing discussion regarding their pricing, one thing to note is that the $20 “styling fee” is actually applied as a credit toward any purchased merchandise.  So for example if you buy a $40 sweater from them you pay $40, not $60.  (The VIC writeup for instance was a bit misleading in this regard.)  Put differently, they are essentially charging their customers for their styling services by baking the service fee into the price of merchandise — which I think explains (and justifies, at least for their core customers) their higher-than-normal markups. 

 

And to expand on this point a bit further, I think this company is probably best viewed as a combo of two businesses: (a) a clothing retailer, and (b) a personal styling/shopping service provider.  The really interesting and valuable part is of course (b), and I wouldn’t be surprised if the company’s business model changed pretty dramatically over time as they find better and more creative ways to monetize it.

 

Anyway, I think it’s an interesting company.  I may or may not buy their stock, but I will keep watching their moves regardless.

[SHDL]

On the surface, SF's metrics (profits, operating margin, etc.) appear to be getting worse.  Why is that?  Does the company have to expense "investments" that would ordinarily be capitalized by a traditional retailer?

 

I imagine their tech “investments” are a big driver here.

 

In terms of numbers, their filings indicate that their tech team grew from about 170 employees in 2017 (95 engineers and 75 data scientists, according to their S-1) to more than 280 employees in 2018 (180 engineers and 100 data scientists, according to their 10-K).  Each employee probably earned a base salary in the $100k-300k range and cost the company around $200k-900k.  So the incremental cost of growing the team by 110 was likely around $22-99m.  If so, that explains a big chunk of their $90m y-o-y increase in SG&A expenses. 

[SHDL]

So as part of my research, I tried placing an order with them.  Here’s how it went in case anyone’s interested:

 

I first had my wife create an account with them.  They made her answer a very long list of questions about her physical appearance (height, weight, other measurements, age, …) and preferences (which body parts she wants to (de-)emphasize, personal colors, favorite brands, preferred price range, …). 

 

Then I had her place an order, with a note attached telling them pretty much exactly what type of item she wanted (blouse, sweater, …) and for what occasion (work, casual, …), along with a few other special notes/requests (how conservative the office environment is, fabric quality, …).

 

It took over 2 weeks for the items to arrive.

 

The items that arrived were all of good quality and fit perfectly, and all special requests were honored.  Out of the 5 items that arrived, 2 were excellent picks (much better than what was expected), 1 was shall we say “adventurous” (a bold design that she had never tried before), and 2 didn’t work in terms of color/style and were sent back.  They asked for detailed feedback on each item as part of the check out process.

 

In total, she spent about $220 total on 3 items. 

 

I tried to do a price comparison online but that wasn’t really possible because the exact same products were nowhere to be found (including those from very well-known brands, interestingly).  But my very rough guess is that she probably could have bought comparable items for about $150 elsewhere. 

 

So was this a good deal?  For one thing it was certainly a nice time saver — it probably would have taken her at least a few hours searching around and trying things on to get the items she got from another seller.  Our personal circumstances are such that spending $70 or so to save those several hours is worthwhile, but of course not everyone will feel the same way.  Also, return shipping is really easy for us because we live in a building where we can just leave the package with the front desk concierge, but I think it’s a much bigger hassle for many other people. 

 

And most importantly:  My wife was pretty happy about the experience and wants to place another order. 

[KJP]

So as part of my research, I tried placing an order with them.  Here’s how it went in case anyone’s interested:

 

I first had my wife create an account with them.  They made her answer a very long list of questions about her physical appearance (height, weight, other measurements, age, …) and preferences (which body parts she wants to (de-)emphasize, personal colors, favorite brands, preferred price range, …). 

 

Then I had her place an order, with a note attached telling them pretty much exactly what type of item she wanted (blouse, sweater, …) and for what occasion (work, casual, …), along with a few other special notes/requests (how conservative the office environment is, fabric quality, …).

 

It took over 2 weeks for the items to arrive.

 

The items that arrived were all of good quality and fit perfectly, and all special requests were honored.  Out of the 5 items that arrived, 2 were excellent picks (much better than what was expected), 1 was shall we say “adventurous” (a bold design that she had never tried before), and 2 didn’t work in terms of color/style and were sent back.  They asked for detailed feedback on each item as part of the check out process.

 

In total, she spent about $220 total on 3 items. 

 

I tried to do a price comparison online but that wasn’t really possible because the exact same products were nowhere to be found (including those from very well-known brands, interestingly).  But my very rough guess is that she probably could have bought comparable items for about $150 elsewhere. 

 

So was this a good deal?  For one thing it was certainly a nice time saver — it probably would have taken her at least a few hours searching around and trying things on to get the items she got from another seller.  Our personal circumstances are such that spending $70 or so to save those several hours is worthwhile, but of course not everyone will feel the same way.  Also, return shipping is really easy for us because we live in a building where we can just leave the package with the front desk concierge, but I think it’s a much bigger hassle for many other people. 

 

And most importantly:  My wife was pretty happy about the experience and wants to place another order.

 

Thanks for the info.  Did the Fix include brands that your wife wasn't previously aware of?

 

My fixes did introduce me to a new brand, which was great for me.  But now that I know my fit in that brand, I buy directly from the brand's DTC website.

[Guest ajc]

 

 

I tried to do a price comparison online but that wasn’t really possible because the exact same products were nowhere to be found (including those from very well-known brands, interestingly).  But my very rough guess is that she probably could have bought comparable items for about $150 elsewhere. 

 

 

 

Part of this article addresses that - https://www.businessoffashion.com/articles/news-analysis/stitch-fix-introduces-over-100-contemporary-brands.

 

Stitch Fix is using client feedback and their data science teams to create unique offerings that only they sell. I think KJP pointed out that a few places do this, but I don't know of anyone who is actually doing it in such a tailored and scientific way. In other words, you really can't buy many of their items anywhere else and in fact they're produced specifically to appeal to client's taste preferences in ways that their competition doesn't actually have the data on or fully understand. Therefore, I'd say what your wife bought was really a more well-designed product in the sense that comparable ones for $150 would've been somewhat less appealing. At this stage it's more like the difference between a meal with ever so slightly too much spice or something like that. The thing to keep in mind is that the product differences are almost unnoticeable at this point, but a few years from now they'll become more apparent as Stitch Fix continues to deepen their understanding. If you take their Hybrid Designs as a simple example of this, Eric Colson has already noted in various talks that they were very good sellers and in this article you can see that the sales performance of some of them was extraordinary - https://qz.com/1028624/stitch-fix-let-an-algorithm-design-a-new-blouse-and-they-flew-off-the-digital-racks/. So, the differences right now in the Stitch Fix experience and clothing are mostly imperceptible, but I think it's unlikely to be that way for many more years.

 

This other article (https://www.forbes.com/sites/pamdanziger/2018/12/01/nikes-new-consumer-experience-distribution-strategy-hits-the-ground-running/) on Nike's new consumer experience strategy (which Stitch Fix made the cut for), also gives some further and different insights about where the fashion industry and brands are headed. If you combine this with the Business of Fashion article above, you can see that Stitch Fix clearly stands to benefit significantly from this trend given that their service is more personalized than everyone else in most respects.

 

Finally, Stitch Fix also took market share (https://seekingalpha.com/news/3423111-online-holiday-sales-17-percent-2018) this holiday season even though usually they do not see a bump in these months. Basically, they're a full price, year-round, personalized service, so Black Friday, Xmas, etc, isn't something they participate in, nor do they do marketing for any of this stuff.

 

Related to your experience with Stitch Fix, there's one thing I've noticed which is actually a little incredible. Twitter is obviously known for no-holds-barred commentary. If someone thinks something is a piece of shit, then Twitter (and YouTube comments) is likely going to be the place you'll find them saying that.

On the other hand, looking over the Stitch Fix feedback on Twitter these past few months it is pretty much uniformly positive and enthusiastic (https://twitter.com/search?q=stitch%20fix&src=typd). Scroll down as far as possible. I mean, for anyone who knows how rough-and-tumble that place can be, it really is somewhat crazy.

Also, having performed that search regularly over the past few years, it's clear to me Stitch Fix has actually improved their service, since there was a time a while back when there was at least a visible amount of dissatisfaction with them coming from a chunk of clients. That's not to say everyone's happy today or anything, but I think their performance on this score is worth a mention for its outlier status.

 

 

 

[cameronfen]

I think SFIX is growing on me.  Personally I think retail whether its online or offline makes no money.  Amazon never turns a large profit on it's own items and the place most of these people make money is by charging a fee to independent sellers.  The average retailer has a price to sales ratio of .5 and online retailers have no significant cost advantage as the lack of stores advantage is negated by their much higher shipping costs.  To piggy back on other people, the advantage SFIX has and where the most value imo is in their own brands that they will create by learning about the behavior of shoppers (and so the retail side is important in expanding their data moat).  These companies trade at a price to sales of roughly 2x with the high end brands (berburry, lvmh etc.) generating formidable roe.  But even something like a gilidian or a fruit of the loom or a Zara are high roe profitable companies that have withstood the test of time.