Stitch Fix to Lay off About 1,400 Employees in California | Hacker News Reader

Stitch Fix to Lay off About 1,400 Employees in California

wsj.com

149 pointsseibelj118 comments

This is more interesting than all the other layoff stories in that they're axing 1,400 California jobs but adding 2,000 in lower-cost locations like Dallas, Pittsburgh, Cleveland, Minneapolis and Austin.

It's like offshoring -- but off-Californiaing.

I wonder if they gave the employees the option to relocate with out of calofornia salary. Not sure if that’s legal but it would have been worth giving them an option, some might as well take it, everything’s too expensive in cali, one cannot easily start a family there

They gave them the option to stay on if they relocate. Doesn’t say they keep their CA salary though.

>“ Most of the layoffs will take place in September and those affected will have the opportunity to relocate and stay with the San Francisco-based company.”

The two Stitch Fix stylists that I've known both worked remotely and I got the impression that most of them did.

Is that not true? If it is true, how do you relocate a remote worker?

In this case perhaps rehire with slashed salary. Some people in need would take it

And resent you forever, as well as likely to leave very soon. It would be a liability.

That isn't a problem at all in a lot of industries; sure, you may have to coddle tech workers for now, treat them like actual human beings and all that nonsense, but imagine a future where you can treat just like your warehouse staff. Wouldn't that be neat? /s

On a more serious note, I know a few companies who do treat their tech people like disposable inventory, on par with the lower-prestige roles, and it works out just fine for them. They don't get highly motivated world-class talent, but they don't need that sort of talent; just someone to take care of their relatively modest IT/dev needs. It's not very efficient across a lot of metrics, but low salaries probably compensate for that.

This seems to be a trend across all sorts of industries, less and less jobs that pay well, less employees that get coddled and treated well, commoditize the hell out of every last role etc. – just because it's not a big thing in much of tech yet probably doesn't mean it's an impossibility.

It’s already happening IMO. Tech wage growth has been pretty slow esp outside of the FAANGs, and many agreed upon “best practices” are commonplace in the industry now, leaving less room for differentiation among workers

> but imagine a future where you can treat just like your warehouse staff. Wouldn't that be neat?

I think that day is closer than many imagine. If I take out well paying / no physical labor part I am as a Jira slave and daily standup chump are lot closer to warehouse worker than silicon valley type pampered engineer.

Thats normal, in most industry, the worker resent their employer. They come and leave all the time.

A company purpose is to make money, not to take care of its employees.

"We are going to cut your salary in half. You should move to a cheaper city to make up for it."

Lower minimum wage floor in those areas.

> "We are going to cut your salary in half. You should move to a cheaper city to make up for it."

To think that less than 4 months ago People were ridiculing Andrew Yang and other proponents of UBI. I admit having it be, initially anyway, State sanctioned bothers me a great deal; but the fact is gainful employment is no longer a reliable aspect of modern Life for many in Society. This will continue to cut deeper and deeper, and for anyone not paying attention these protests and riots in the US have a much deeper sense of resentment than just Police Brutality against black males. And it's felt throughout the spectrum of underemployed, highly indebted underclass. Its like the bottled and pent up anger from the Occupy movement coming back.

In many ways this reminds me of the London Riots of 2011, when London police killed a young black male named Mark Duggan [1] and it exposed a myriad of sociological, and socio-economic issues that quite frankly still remain to this day.

1: https://www.washingtonexaminer.com/opinion/three-riot-lesson...

Context matters. The current economic climate is in no way representative of normalcy as we just shut down a large chunk of it indefinitely.

That aside, the idea that you can finance life in the highest cost living areas of the USA as a “stylist” (picking items for a clothing subscription) is lunacy. Clearly that work can be done anywhere and it’s only a matter of time till it’s pushed to the lowest cost areas. I wouldn’t be surprised if the cost of such labor goes to near zero as they attempt to start “paying” people with publicity instead.

The job of a stylist will largely be automated away with AI. There will be a few needed to keep training the machines but it’s a career path akin to the horse and buggy maker.

Actually Stitchfix is an interesting example of “human in the loop” AI. The AI makes recommendations to the stylist, who assembles outfits. Presumably the stylist is still adding value or they wouldn’t have them.

Either way, if people can be scammed^Wconvinced into using their own car to make deliveries for a few bucks that barely covers depreciation, then you can definitely find someone to sit on their computer and click through pictures of clothing for less than an SF city livable salary.

I'm not convinced of this. We could easily have used software to procedurally generate outfits from a selection of components since the 1980s. Why hasn't this already happened?

The same reason that Jobs (and Ive) were the arbiters of what Apple shipped for many years; you can't program taste or style.

Putting someone’s job title in quotes like that is super condescending. It’s bad taste, especially when you make more and you’re commenting on a post where 1,400 of them got laid off.

> Putting someone’s job title in quotes like that is super condescending. It’s bad taste, especially when you make more and you’re commenting on a post where 1,400 of them got laid off.

It's in quotes with an inline description because, while it's the actual job title for the positions in question, it's also a common English word that in most circles would refer to a "hair stylist".

Ascribing negative intentions to minor punctuation and making up biographies about your fellow commentators says more about your bad taste and biases than mine.

You clearly don’t know much about how style jobs work. Sometimes it can be done anywhere, but there is a difference. Like most of the nicer things in life, if you can’t tell the difference, then it’s not for you.

I suppose you think it’s lunacy that hair dressers think they can live in the city as well.

UBI = "universal basic income" here

The requirement to relocate makes me wonder if the true reason is to avoid some current, or likely future California labor law.

Why not just offer a straight salary cut, since they are already, as you say...remote?

What are the implications of this in terms of taxes? My understanding is that the CFTB https://www.ftb.ca.gov/ will continue to tax you as a CA resident as long as your employer is based in CA. The laundry list for determining residency is long and in many cases you would end up paying taxes even if you don't live in CA.

https://www.forbes.com/sites/robertwood/2019/10/22/now-calif...

https://moskowitzllp.com/leaving-california-and-the-franchis...

The article I read said that they were going to give their current employees the opportunity to relocate. It did not specify if there was going to be a change in their income though.

It doesn't make sense to do it if they don't cut their pay.

Apparently they’re not engineers, but stylists. Can the difference in pay for stylists be very large across states?

If there were no difference, what would be the motivation to fire in CA and hire elsewhere? I guess office rent is a factor, but that's probably small compared to salaries.

Nothing compared to salaries for engineers. Office space averages $17k a year a head, and it’s probably higher in area like CA and NYC where land is expensive.

Wow, 17k per head outside the valley and nyc? Seems exceasive, its the first time i har of this figure. Any idea how much that would be in california/nyc?

When I was doing some budgeting in London, England several years ago, I found that the monthly rent for a small apartment was similar to the the cost of a desk in the office, so 17k per head sounds about right to me in large urban areas.

> 17k per head sounds about right to me in large urban areas.

I don't know about the US, but in France, in large urban areas like Lyon or Toulouse, you can easily find office space for 100 to 200 €/m²/year. Which means, translated in $, that $17k per year would grant you 75 to 150 m² (800 to 1600 sq. ft.).

That would be a very very nice office, far from the shoulder-to-shoulder openspace model, even if you add other expenses than pure rent :-)

There are some American cities where prices are similar. The data is old, but Dallas was about $10k cheaper per head than NYC was in 2015.

Of course, in America rents in NYC and SF are going to drive up the averages by a lot, with companies like FB and Google willing to pay huge amounts for prime real estate and luxury build outs.

So the interesting question for you is: how much are rents in Paris, and how does that affect the averages of France as a whole?

Office space is both cheap and shockingly expensive. It’s cheap because it’s usually a few dollars a square foot, often as low as $1-2. It’s expensive because you need a lot of square feet, plus room for expansion. Your average employee needs 100-150ft in order to not feel cramped, including personal and communal space.

Here’s an article from 5 years ago that showed a $10k per employee per year cost difference between NYC and Dallas. https://www.marketwatch.com/story/heres-how-much-your-compan...

Keep in mind that office prices are going to vary wildly based on location. High rises in the downtown with good access to public transit are going to cost a lot more than an office park that’s just barely within city limits. So these averages per city are going to hide huge standards of deviation.

> Your average employee needs 100-150ft in order to not feel cramped

Haha, tech companies in the Bay Area use open office designs that pack people like sardines. Each employee probably gets closer to 30-50sqft.

they can cut their pay and their employees could still be better off since they are going to be expaning in Austin where there is not state income tax.

Also, the company will be able to save more money since everything is cheaper from taxes to real estate in other states compared to California.

It might be worth it to keep them at their current salary and just not pay SF real estate costs.

> Not sure if that’s legal

Why wouldn't it be legal?

You don't need to wonder. It literally says that in the article. Even above the paywall.

I was curious about that. They keep popping up with job openings lately and I was t sure what’s going on there.

>It's like offshoring -- but off-Californiaing.

This is what "Going full remote" post-covid is going to look like for a lot of companies.

Good!

It makes sense. If you're not looking for super high-end tech talent, there is no magical advantage to paying California salaries so that people can spend every dime on housing.

I skeptical of most of the tech stuff too -- fewer tech roles require the skillsets/social connections only found a big tech hub than in the past. For many years, IBM was able to run successful tech operations in places like Minneapolis, Austin, Binghamton NY, Hudson Valley NY, North Carolina, Tampa, etc.

The median home price is Pittsburgh is literally 90% less than many California locales. Even with a 40% pay cut, in many professions moving to one of these places will be a neutral in terms of money and probably a positive impact on the employee's quality of life.

> The median home price is Pittsburgh is literally 90% less than many California locales.

Can you give some data around this? 90% seems like a hyperbole. Or are you comparing the most expensive zipcode in the bay area (e.g Atherton) with the cheapest in Pittsburgh?

My friends just moved to Pittsburgh and bought a great house for 108k. This is literally 95% less than a comparable home in Silicon Valley.

and he probably got a backyard with it too. In CA, you don't even get that. Why pay 1 to 2 million dollars in CA for a house when you can't even have a backyard?

I didn't do a study of real estate valuation. I googled "real estate palo alto" and "real estate pittsburgh" and grabbed the number from the first Google results. (realtor.com)

I can tell you that I live in a small northeastern city, and a 3 or 4 bedroom mid 20th century, single family house is around $130/sqft in a nice neighborhood. In the burbs, it's about $150/sqft. When your mortgage+taxes is <$15k/year, you don't need to make crazy salaries.

I don't know bay area real estate, so looking at Gilroy, CA, which would be a hella commute to SFO, you're looking at $400-450/sqft for 3/4br single family. That's alot of cash.

I moved to Cali from Illinois. These numbers look correct. A similar, but smaller property was MORE than 10x greater in California.

Note: These are stylists and not engineers.

I've tried StitchFix twice and found their stylists pretty lacking in terms of value they add. They don't listen to your requests and I'm sure most of them just have templates of stuff they send out w/o looking at what you liked / disliked. I'm fairly certain it's a role that can be very easily filled regardless of the Zipcode.

In fact, I was pretty convinced until now that the stylists were just algorithms but it looks like I was wrong.

My assumption has always been that it's a supply side problem to solve and their style choices are limited. If I made a business like this, I would source unsold/unwanted clothing at a discount and try to pair it up with people who might actually buy it. I've just assumed that's what they're doing... anyone know either way?

I also had this conception, then learned most if not all of my items have been from Stitch Fix’s internal brands and that’s why I didn’t recognize any of them. I thought they were a curator of brands and styles I might want to check out, but it seems to me they’re running it like a D2C high-volume fashion label.

So I’m guessing by owning the brands themselves they get even higher margins than if they were buying unwanted brands in bulk at a discount. And each of the brands goes for a certain generic style, e.g.: “Even Tide brings the vacation to your closet with a mix of coastal colors and mid-century modern inspired lines.” [1]

Kind of a let down for me as a subscriber, but I still buy quarterly as I’m so bad at picking a wardrobe out on my own.

[1] https://www.stitchfix.com/men/blog/the-service/exclusive-bra...

I tried them out for 4 months. I kept one garment per delivery on average. I buy my clothes from thrift shops all the way to Barney's/Neiman Marcus. You can find quality at any price point. However, when their internal brands began showing up more and more in my boxes I started suspecting they were offloading "junk" on me at a premium. I thought about whether the stylists make a commission or have flexibility in pricing the garments. There were many items that I thought were overpriced for theor quality. They sent me a "flannel" shirt that had the feel of paperboard.

I thought about applying for their ML positions, because it's definitely an interesting business problem. But I got a strong feeling they don't actually use ML in their product. They're just a retailer after all.

Likewise. The one time I used them, I was sent some pretty generic stuff that had little to do with what I specified within my profile.

It's almost as if it didn't matter what I put in my requirements, I was going to be given the same stock pack du jour. I even got this heavy winter coat from them that maybe would have made sense in Minnesota, but had no place in the Bay Area. They were not geography-aware at the time, from what I could tell.

Never used them since, it's easier to go into the store myself.

Generic in terms of your taste or generic in the larger population? IMO the whole reason to have human helpers is to avoid both, while algos can do both kinds of generic well.

> I was pretty convinced until now that the stylists were just algorithms but it looks like I was wrong

Katrina Lake talks a little about the way they've structured this during her How I Built This interview[1]. They seem to have a pretty large engineering / data science org, and it feels like the stylists are effectively human curators of algorithmic recommendations.

1. https://www.npr.org/2019/09/10/759594143/stitch-fix-katrina-...

"Today (Sep 2019), it has about three million customers and brings in more than a billion dollars in annual revenue."

Wow!

It always amazes me what people will pay for that seems to have zero utility from my perspective.

You mean, like movies and video games? Or food other than rice and beans?

Most people in first world countries are well above the subsistence level, and much of what they spend their money on isn't useful. It does, however, improve their quality of life. Clothes are no better or worse than my own hobbies, and I suspect, yours.

I meant paying someone else via a subscription service to do it. Clothing is important for signaling, and there is definitely utility there, but I would think there are other ways that are cheaper/easier/more reliable.

In the face of a billion dollars of revenue, what do you think is more likely: 1) everyone using stichfix gets "zero" utility from it or 2) You don't understand the thing you also have an opinion about for some reason?

For some people, buying clothes is a really stressful and difficult psychological process. You really don't know anyone who does not like shopping for clothes and would be willing to buy away that stress from a personal shopper/algorithm ?

Maybe you personally don't find value in that, or your tastes don't agree with Stichfix's style, but it seems quite understandable to me that many people would want this.

Of course people are getting utility from it. But I was stating if I was presented with StitchFix as an investment opportunity, I wouldn’t be able to imagine how it would be valuable to people, so I might pass on it. I’m not denigrating anyone who does get utility from it.

That seems to be a systemic problem in tech. Most VCs are dudes and they can’t see the value in startups that do things men are less interested in, making it harder to get funding to solve problems women have. Might be why some folks are (rightfully) sensitive about the topic.

Katrina Lake definitely had this problem getting her first funding rounds - the VCs she talked to thought it wouldn't work, because there's a perception that most women like to go shopping, or just not seeing the value for themselves.

VC funding comes down to gut decisions, and I suspect there are a lot of other billion-dollar opportunities out there in spaces that are overlooked by the demographic of wealthy white american males between the ages of 25-50.

I kind of get what you're trying to say, but it's important to remember that the definition of 'utility' is not universal.

One example comes to mind from the recent thread[1] about hiring for older programmers - someone gave an example of an older friend of theirs who started getting substantially more offers once he changed the way he dressed:

> It began with a friend who was in the job market as a 50+. More on the hardware side. This guy has some cool experience. He gets lots of interviews, they go well, but no offers. As his frustration grows, he grows desperate to try something different. He dyes some color back into his hair, gets some tinted glasses, and lets his daughter take him shopping for some more hip interview clothes. A month later and he's in bidding wars for who to hire him. He said the difference was night and day. He was now pointing out his age in interviews "are you sure I'm not too old?" and the interviewers were like "no way man."

In a case like that, fashion has clear social utility and a very real, tangible value to that person.

1. https://news.ycombinator.com/item?id=23320974

I agree fashion has utility. But if I needed help with it, I would go look at some manakins or clothing websites or whoever you want to emulate and copy that. I wouldn’t see a reason to pay someone else to do that for me.

My wife is a customer, and initially I was irritated that she had signed up, since I viewed Stitch Fix as akin to the subscription services that send food, etc.

But being smart, I decided to avoid expressing my skepticism about how she was spending her money. Turns out that for her, it's a great idea. She hates to shop, hates to spend time shopping, but has to look professional for work. Like many women, she's not a size 0, so finding stuff that fits as well as looks good can be a challenge.

With Stitch Fix, they always send the right size. And the styling (as much as I can tell as a dude) looks fine. If you don't like something, you just send it back.

Prices are higher than I would have expected, but again, that's because I'm a dude who buys cheap stuff. Women's clothing is both more expensive, and less sturdy than much of what men wear.

> In fact, I was pretty convinced until now that the stylists were just algorithms but it looks like I was wrong.

Maybe the stylists really exist to train an algorithm that can replace stylists in the future?

I would be very surprised if an algorithm did not make suggestions that the stylists would confirm, deny or modify.

Even for stylists, the choices they've made for the models on the front page are... tacky, to say the least. I don't consider myself fashionable in the slightest but that the outdated camo jacket on a sky blue tee is a freakin eyesore. It seems like just over half of the photos are appealingly color-coordinated.

If that's what their stylists are coming up with then... ew

Same, shocked they have this many human stylists when they seemed to make out they were using ML to predict taste.

I agree. Tried the service a few times and didn't keep a single item. A lot of their recommendations were actually from their own Stitch Fix clothing line, which felt like at best Gap quality for Gap full retail prices.

I gave StitchFix a couple of shots. It was okay, I got a couple decent items out of it, but didn't amaze me.

But I canceled when I very specifically stated in both my profile and my message to the stylist that I did not want any white/tan colored items (I'm fair skinned) or polo shirts (just not my style).

My next order contained a white polo shirt and a tan knit beanie (in the summer!). smdh.

I, too, figured it was just some algorithm dressed up as personal stylists, but knowing that it is actual people just makes it even more unforgivable.

it always amazes me how many human filled jobs are very very robotic. It is often indiscernible on if a job is filled by a human or robot

Anecdotal, but my wife is a stylist and is very much listening and providing individual feedback. She feels so bad when someone doesn't like what she picks so it's her goal to make them happy. Not all are bad.

I believe this. I love the idea of StitchFix but have not had great experiences with them. I finally had a stylist explain that they just don't get much inventory that matches my style preferences. This makes perfect sense, since they need to stock clothes that will fit/suit the widest variety of people, which leads to a lot of baggy, shapeless choices.

Don't forget drab colours and ridiculous lapels and pleats!

That much is true, she often gets frustrated that she can't pick from their entire inventory.

I love Stitchfix. In my experience, however, this is accurate.

I've received orders from good stylists and bad ones. The bad experiences were exactly as described: my preferences were clearly not read, they specifically picked things I had already rejected on the style-carousel-rating thing. It felt like they just dumped a bunch of excess crap on me. When this happens, the pieces don't even fit right. Very frustrating.

On the other hand, I have gotten shipments that were clearly respecting my requests and preferences. Those interactions absolutely send me over the moon, and make me much more likely to buy something.

This is just a wave of low cost shoring. I work for a SV company and we are “offshoring” all our support from Midwest to Canada. I can see this trajectory play out for Stitch Fix and other companies. Pretty stylists here are commodity jobs and plenty available in Canada. Best part is that it is an English speaking nation as well

> all our support from Midwest to Canada

Is Canada lower cost than Midwest?

Yes, because of Canadian dollar

Another story about this, which isn't paywalled: https://www.msn.com/en-us/money/companies/stitch-fix-is-layi...

recent somewhat related article discussion:

Remote work means anyone can take your job https://news.ycombinator.com/item?id=23337857

They can if your skills aren't unique.

Very few people are unique, so what's the point in mentioning that in a tiny sentence if you're not going further with it? Perhaps the insight comes... one reply later?

Most employers aren’t quite sure what they’re paying for so it hardly matters if your skills are unique anyway.

Very few people in any industry have a truly unique set of skills. And in jobs that can be easily remoted away like software, that quantity is even less.

There's not too many of us with truly unique skills. And even those can usually be covered by different combinations of employees.

Unique in this context doesn’t mean absolute uniqueness, it means your skills aren’t widely available.

The world needs experienced Go developers for an example. Being able to hire one remotely may increase the available pool of Go developers you can hire, but it doesn’t make them cheaper. That’s because hiring remotely also increases the number of employers chasing the same few Go devs.

Even those without "unique" skills can typically be taught and the company still comes out ahead.

For small businesses and small startups it's still non-trivial to hire out of state, in terms of legal and tax overhead.

Ignorant here and sure this varies per state, but what are some of main the legal or tax challenges?

Some states have worker’s compensation monopolies so you have to buy it from the state.

All states require that you register as a company there and pay taxes.

There are PEOs like Justworks that act as the employer and file taxes for you. But you still have to register first and then assign the company to the PEO.

In short it varies a lot and that doesn’t take into account things like local laws on compensation.

I say this as a founder of a remote-first startup registered in 12 states.

Interesting not what I'd expect with a PEO. Any sense on whether JustWorks requirement is because of law or because is best for JustWorks?

It’s because of laws. PEO handles some things but states being what they are have exceptions.

It’s unfortunate because this system encourages hiring remote offshore contractors.

Yeah, I was surprised when I found out this is how it works. When I first signed up with a PEO I assumed it was all taken care of. But then they handed me a big list of states I needed to register with and all the very antiquated processes each of them uses.

What's the situation like if your employee opens a company and invoices you like a consultant ? I do this in Europe (admittedly I don't work for a single client exclusively) but it seems like the path with least friction for tech workers.

In that case it's very simple for the company paying but much more work for the employee since they are the ones responsible for taxes. It's not necessarily a bad thing, but I've found many employees don't want to do it. They need to make sure they are either paying their estimated taxes quarterly or be saving a significant portion of their income.

That also doesn't address that it is significantly cheaper for group health insurance in the US vs. individual. As a company I can be paying $400/mo vs. the employee who would be paying $2K for similar coverage.

You need a healthcare plan or plans that covers employees based there (some HMOs are regionally focused for routine in-network care).

This is fairly surprising since most ecommerce businesses have been doing well because of coronavirus shelter-in-place rules across the country.

Speaking as a Stitchfix customer (and nerdy engineer who loves their service and hates shopping for clothes), I semi-permanently paused my subscription a couple of weeks ago. I haven't left my house in two months, so there's not much point in buying new clothes.

I know of at least 1 luxury clothing startup that is doing extremely well due to the increased online activity. Maybe it's just stichfix's market segment?

Which one?

https://archive.md/zQh5e

Kind of interesting they picked Austin as a low cost alternative. I get it's a tech hub, but surely there are other cities less expensive?

Austin is ridiculously expensive by Texas standards. Super cheap by California standards.

Lots of Texas is cheaper, but executives don't want to drive three hours in a rental car after their flight.

Dallas, Houston, and San Antonio are cheaper than Austin and do not require a 3h drive from the airport.

I'm sure you're right, but I'd like to better understand how a city surrounded by sparsely inhabited wasteland could have a shortage of anything of which those other cities have plenty. Austin is practically a suburb of San Antonio anyway, and has no obvious qualities that aren't shared by ten other cities and big towns in Texas.

Other Texas cities do not have the cultural amenities or pleasant weather of Austin, nor do they have the political sensibilities of Austin that make it more attractive for younger workers. It's the city in Texas where you will have the least difficulty finding friends outside the MAGA and hardcore evangelical groups.

Quality vs expense. If you can half the cost for similar quality in Austin or 80% less cost for zero quality in some nowhere town, you go for Austin...

I live in Austin. There's "Austin", meaning the actual city limits, parts of which are getting more expensive (although still a fraction of SV prices). Then there's the Austin area, large parts of which are half the cost of central Austin, itself half the cost of SV. There are also still low-income, low-cost parts of literal Austin, which I think has not been true in SV for a while (although I haven't lived in SV since 1992 so I could be wrong there).

I'm sure cost isn't the only factor. You still have to relocate to some place that will attract talent, either right out of college (like UT), because it's already a big city or because people want to live there. It would be even better if the state is offering tax breaks, or has lower taxes than where you're coming from.

Pure horse shit, about "lower cost areas". I suspect there is a right wing shill hiding somewhere here. The founder's appearance on Shark Tank has the smell of it. Why else would you go for such a ridiculous headline.

I hope this doesn't impact my first box just ordered last night.

I’m surprised they had that many stylists — most of this could be done by AI and a handful of tastemakers.

Unless by stylist they also mean customer care, which they probably do as this article mentions hiring 2k people in lower cost US cities.

https://www.cnbc.com/2020/06/01/stitch-fix-is-laying-off-140...

My understanding is that it in large part was done without AI because a lot of the input requests are too complex to summarize in any form other than language.

“I work in an semi-casual office but I want to try adding some new colors to my wardrobe to spice things up. Next month I’m going to a wedding and I need a new bracelet and a pair of gold earrings for that. I don’t need any more scarves this fall but I’d love a nice knitted hat in cool colors. Stripes don’t flatter me.”

And I could go on for a while longer talking about a wide range of clothing choices. Could you give options for all of that? Yes, with some work. Would that be a good user interface that keeps a wide range customers returning? I doubt it.

Plus their user base just prefers at least the illusion of having an individual stylist, so you receive a detailed note with each package talking about the ways that the clothing choices fit your style, wardrobe, body type, and activities.

You are right about the input requests being too complex. From this microservices presentation[0], they do use machine learning to generate recommendations which are then curated further by the stylists, specifically for this reason.

But I've always wondered about the grouping and parity between stylist and customer. IIRC, they say it in the presentation but they do not say how much bucketing happens. Is Person of archetype-A grouped with 100 other people and given to Stylist A? 1000 other people? Or do they only do additional curation if you reject so many items and write so many input request words?

Also, is there a good track record for such a request like "I need a pair gold earrings" working? and working specifically because of human curation and not pattern matching on color and item description?

0. https://youtu.be/E8-e-3fRHBw?t=332

Reply on news.ycombinator.com