Hacker News new | past | comments | ask | show | jobs | submit login
Growing from 0 to 4M users on our fashion app with vertical machine learning (medium.com/aldamiz)
306 points by aldamiz on Feb 7, 2018 | hide | past | favorite | 71 comments

What is interesting to me is that this comes on a day when Bloomberg says "Americans are spending less on clothing". Full discussion here:


Combining the ideas of both posts: You can spend less if you buy more targeted-to-you and higher-quality items that you'll wear more often. Finding these items and customers is a data (and marketing) problem.

E.g. I spend less overall than I did a few years ago but each item I buy is more expensive. I've learned the value in quality clothes. I give smaller batches of more money to well-picked retailers rather than lots of small batches of money to big-name mass-market retailers.

An app that capitalizes on people coming to this realization (and knows how to market and execute it well) could make a killing.

Same. Bought a $600 leather jacket last month.

But I also only own 8 t-shirts, 2 jeans, etc. And if I get a new t-shirt it’ll be quality and it’ll probably also mean I’m throwing away one that has become worn.

I spend less money overall on clothes now than I did a while back when I’d buy a random $7 t-shirt just because I thought what it said was funny (and then wear it maybe once or twice before it getting lost in the pile of countless clothes I constantly did laundry on but never wore / me getting rid of it)

I expect the leather jacket to last decades. And I’ll actually wear it instead of it sitting in a closet.

Help me find clothes that fit me and make me look good and are of good quality (I don’t care about brand names) and you can have some of my money. I’m looking at you, ShapeScale.

Please feel free to advocate for a brand you like. I’m a 2 jeans 8 tshirts kinda guy too, and been struggling finding shirts that I’d recommend.

Everlane ($ garbage), American Apparel ($ bad with new owners), Alternative ($$ so-so). Best shirts were the BDG brand ($ soft, last 4 years with weekly wear) at Urban Outfitters but they’ve been discontinued.

If you're like most people on this site (dresses like avg male SF dev)

$$$ - A.P.C.

$$ - Naked and Famous, Muji

$ - Uniqlo U line

+1 I bought a couple of Uniqlo active t-shirts. Man, are they good despite being cheap at 20$ a pop. Wear them for running, office or even under a blazer. Super convenient. Only downside - it's made of polyester, feels great on the skin but it isn't a natural fabric.

Muji clothes are nice, but wear out crazily fast.

Personally, I’m a fan of Next Level v-necks when it comes to t-shirts. They’re quite inexpensive through Amazon. I hate doing laundry, so I usually have about a dozen. I’m 6’, 220lbs, and the athletic cut is great—incredibly soft, broader shoulders and chest vs waist (no straight cut). I typically go for the large when wearing normally, but keep a few mediums as a snug undershirt when wearing a buttoned/dress shirt. I usually get a few years’ use out of them.

Pants are almost always a pain for me—I can’t do skinny jeans as my ass and thighs refuse to cooperate with the stick-leg cut. I opt for Buckle/Salvage jeans here—not those Affliction abominations with the bedazzled ass pockets, mind you. They make a few cuts that are a stretchy type of denim, and they’re amazing. Feels just like wearing a pair of pajama pants. I usually have a couple pairs to cycle through between laundry loads.

Brands come and go and get bought/sold and change production-values on such a regular basis that what was true 2 years ago isn't true now, and I find keeping up with the new brands on a regular basis is annoying. I also hate clothes shopping. So I'm not very brand-loyal. I love my AG "Protege" pants, but I'd switch in a heartbeat.

I rather like what Stitchfix and Trunk Club are doing. Trunk Club in particular but it's pretty expensive. They only do high-quality items and they're hand-picked by a stylist who knows your style & body-type to go with your existing wardrobe. I get much more use out of my Trunk Club items than I do others. But I stopped using TC because it's so expensive and I have a solid base wardrobe at this point and know how to add new items on my own every so often.

Patagonia has some good t-shirts. They are just plain and boring, but are of great quality.


Patagucci is an overpriced joke for most things, unless you abuse their warranty policy. T-shirts are essentially free. A pair of jeans should cost you $30 and last for a few years.

Their R1 hoody and climbing pants are the two items I've worn the most this year. Their design is great, and as these pants wear out, they'll fix them for me.

The R1 hoodie is well-designed, but for $200 you can buy 2 essentially identical hoodies from UnderArmor at full price, or a quality 800-fill down parka on sale. It is literally the climber version of a Gucci handbag.

Underarmour and patagonia are VASTLY different in quality standards... I think you’re missing the point of this thread.

For t-shirts I actually really like the v-necks sold at Old Navy. The ones that go for $15-20 (don’t get their lower priced ones, they’re flimsy) have lasted me about 2 years of regular weekly wear now. They also fit my form well, which was a deciding factor. The fine-threaded cotton doesn’t ball up, and colors haven’t noticeably faded.

I'll pitch in for Uniqlo for jeans, too. I picked up a pair overseas last year and they aren't showing a single bit of wear, despite a LOT of use. Compared to my Levis, which start showing wear in the crotch after a few months, they were totally worth the ~$35 I paid.

I'm similar, and I've settled on JCrew. Partly because they offer stuff that fits my stumpy build (I have a 27" inseam), but also because it's solidly good quality.

Presumably because people were buying once every four years?

This applies not just to clothes, but shoes as well. Buy a good pair of shoes and they will last you a lifetime. 300$ for a pair of Allen Edmonds seems like a lot but they'll last for 3 years of daily wear at the very least, and then they can be resoled.

It’s a mixed bag. I’ve bought expensive Allen Edmonds and had subtle fit problems that make them largely unwearable. Another pair of Allen Edmonds fit fine but got permanently stained is salty snow.

Recently the heal and sole of a $400 pair of dress boots completely disintegrated after four months of wear. (Resoled, but still.)

I guess I’m saying that the supposed long-term thriftiness of expensive shoes is often overrated.

Why create a throwaway account to post this?

But aren't you wearing the same thing over and over? What's the point of quality clothes people see so often they think it is dirty or that you own three shirts? I just buy cheap clothes.

3 is probably too few. People don't really pay that much attention to what you wear, I promise.

I usually wear my shirts twice and take them to the dry-cleaner after. I'll only go once if it's been a hot day or if the shirt has dirt/stains. Maybe 3 times if I didn't wear it all day.

Plus I have about 7 pairs of high-quality pants that work well in an office and are comfortable for weekends. Most shirts go with most pants. So I have like 70 outfits so it really doesn't feel limiting, especially if you add in a few sweaters, jackets, and shoes to bring the combinations way up.

I did the math on this once to figure out cost-per-wear of my shirts. I kept rough track of how many times I wore a favorite shirt of mine before tossing it. Roughly 100 times. The shirt was like $120. I paid about $0.75 per wear in drycleaning. So like $1.75/wear.

(I could wash and iron for myself to bring this way down to like $1/wear but I hate ironing so I just pay the drycleaner - I also could have gone another 50 or so wears but I spilled coffee on it and I could always see the stain even though nobody else could I'm sure.)

Compare that with a cheap H&M shirt I bought a few years ago for like $40 - I've worn it twice and don't look forward to wearing it again since it doesn't fit as well and is a bit too trendy in its style for me. Cost is like $20/wear. Save your money and buy higher-quality clothes :)

Your H&M example isn't great. The reason you didn't wear it a lot isn't because of the quality, but because you bought a shirt you didn't like. Save your money and only buy clothing you like.

That's a very good point! You'll save a lot of money by being selective about what you buy.

I think I got suckered into the H&M marketing - it looked quite nice in store and it even felt nice the first time I wore it. But then I realized it only really went with one or two other things, and it didn't look quite as good after being dry-cleaned/washed, so it quickly lost most of its appeal.

> it didn't look quite as good after being dry-cleaned/washed

This is another reason to spend more money on clothes: quality materials look good as they age. I bought a Old Navy pullover which only lasted 2-3 washes before pilling like crazy.

Error was definitely spending more than $14 on a an article of clothing from h&m :P

I don't see anything wrong with a bunch of cheap shirts I only wear occasionally. Variety without too much expense. Wouldn't buy jeans from h&m.

"less clothes" could be 5-8 shirts, which is fine. You don't often see the same person outside of work every single day anyway

I recycle the same 60% of my wardrobe each week. And re-wear pants and shirts without washing them most of the time. But I do laundry every week, mostly under-garments and workout clothes.

I have some cheap clothes, which are functional, but my favorite clothes are all a step above. My 10$ Uniqlo t-shirts are much softer than my cheap multipack shirts. But I have a 40$ marino wool t-shirt that I love and one I don't like. Same with underwear. I have 2$ underwear that works, but is eh, and I have 10$+ underwear that is great, and I have 30$ underwear that I don't really like.

Yep. The main idea of the app is to help people decide how to wear their existing clothes. Many people feel they have nothing to wear when their wardrobe is full - truth is, sometimes the decision of combining some clothes is not an easy one and you simply stop using them. Or even deciding what to wear every single day can be a pain.

> E.g. I spend less overall than I did a few years ago but each item I buy is more expensive... An app that capitalizes on people coming to this realization (and knows how to market and execute it well) could make a killing.

I think that nearly every person trends the same as they age and approach their adult size, even without coming to that realization, so I'd say that you're even more true than your post implies.

Such trends may affect the big players, but for small emerging companies there is still much market share to gain from competitors.

I hope that the company will also focus on employing women and raising the numbers of women in tech! :-) Given that the user base is female, it would make sense and bring the company value to have a diverse workplace.

Here is a nice detail about their process:

We launched an extremely early alpha of Chicisimo with one key functionality. We launched under another name and in another country. You couldn’t even upload photos… but it allowed us to iterate with real data and get a lot of qualitative input. At some point, we launched the real Chicisimo, and removed this alpha from the App Store.

Just yesterday I was wondering where to begin searching for clothing related apps to see what is available in this space. Glad to see this article.

About clothing apps, here are some ideas.

There are teams focusing on the social aspect and grow via influencers. I’d think about utilities (help me do something):

- Help me decide what to wear: Chicisimo, Pinterest and (believe me) Google Images;

- Help me manage my wardobre: Stylebook, Glamoutfit;

- Help me be seen by others: Wear app, Lookbook and Chictopia (this last two worked really well on desktop);

- Help me decide what to buy: ecommerce apps obvsly, or Liketoknowit; or the second-hang category of which several are working really well, and are more widely known.

- Help me get feedback from my friends, no one really working I think, or feedback from the system (Echo Look -> Spark). And a new related category popping up: get feedback from a stylist with an in-app purchase model or even subscription; Wishi, Daam are some examples. This last category will be interesting to follow.

- Polyvore - outfits ensembling;

- Rent the Runway, Stitch Fix, Instagram obvsly. And I'm sure I'm missing many, but just trying to give you ideas of how to find inspiring apps.

I wonder if online fashion is like online music in 2005/6/7... with lots of noise, some tech focused products, and the spotify's of the world starting to be built. Fun times.

One not included in your list: https://meettailor.com/

I'm assuming you're a woman, so forgive me if I'm wrong, but do you know if these apps work well for men's fasion? I think I have OK taste, but I'd like to get good.

I'm a man:) Sorry I don't know about fashion apps for men. I feel teams are building the infrastructure focusing on the big opportunity (women's fashion). Then, expanding to other categories will be way easier.

Thank you very much.

I was idly fantasizing about a design your own app from a mass customization clothing provider, like French Rags. Put in your measurements and pick some parameters, like color, and play with it in an app.

Total fantasy territory, but that was what was running through my mind.

OT: This type of article ('how we hit the goldmine with xy') is always tempting to read but eventually a disappointment.

If someone had the plan to the goldmine why should he disclose it?

What is the current state of the art for finding similar images? Is it auto-encoders nn's, convnets and then just use some sort of distance measure?

That bullshit patent

I really love how machine learning makes something mundane but yet exhausting, into something seamless and enjoyable.

Vertical ML is so 2017, it’s all about diagonal ML

Would you accept someone who has already has a FT job and wants to make more $ by doing remote on the side?

I'm sad there's not a version of Chicisimo for guys :(

So they made a search-engine based on deep learning for a specific (but quite large) target group.

How long do they think it will take Google to catch up, if they haven't already?

Is this target market big enough to interest Google?

Yes. We're talking fashion, so a pretty big market.

We're not talking "fashion". Not the whole industry. Just the market for an app like this.

Arguably everyone interested in fashion could be interested in an app like this.

And arguably everyone interested in reading could be interested in Google Reader, but even that didn't interest Google.

Was this the main point of the story? - "At some point, we were lucky to get noticed by the App Store team, and we’ve been featured as App of the Day throughout the world"

I read an interesting story on how to build a modern data-based product.

What’s a non-data-based product?

Most products. Where you generate Gigs a day of logfile/analytics/crash-reporting data and never even look at then, instead you make changes and updates based on whatever blog post or LinkedIn update the boss or customer read most recently </cynical>

one that doesn't leverage AI/ML as a core piece of functionality

pencils, almond milk, sweatpants, ...

No. From the data they gave, getting featured in the app store added about ~5,000 to their 4 million users (957,437 unique views from featuring in the app store, 0,5% conversion).

That's kind of the impression I got too. Build a nice looking app and hope that Apple throws you a bone I guess.

They’re growing women with their app?!


Is that really of any importance here?

No, but it's important that we control and police language and shove PC culture into your head at absolutely any opportunity and just grind and grind and grind away at any artifact of traditional culture.

Case in point - see the 900 post Tesla Falcon Heavy thread, where the majority of the discussion centered around a tangential point over gender expectations.

Could you elaborate on parts of traditional culture that you think are under attack and are valuable enough to protect ?

If you're talking about https://news.ycombinator.com/item?id=16319505, that's so far from true that I can't even find the comments you're talking about.

Is this comment contributing anything?

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact