YCRFS 2: New Paths Through Product Space(ycombinator.com)
ycombinator.com
YCRFS 2: New Paths Through Product Space
http://ycombinator.com/rfs2.html
22 comments
https://web.archive.org/web/20091001043956/http://ycombinato...
Online stores group products by who sells them.
One of the challenges of abstracting this layer is that who sells the product still adds or subtracts value from the purchasing experience because of those mundane details (shipping speed, reliability, returns, etc.).
There are various ways of dealing with this. Amazon [1] is already active in this space using what you might call the "user review" model combined with enforcing some exacting standards on their vendors. Already having a huge channel of course made it easier for Amazon to get vendors to accept this.
[1] Incidentally, I've never thought that Amazon did a great job with sorting products within its general categories, so I always search instead. But I do wonder, how many more people would window-shop at Amazon if their sorted product catalog was more interesting?
One of the challenges of abstracting this layer is that who sells the product still adds or subtracts value from the purchasing experience because of those mundane details (shipping speed, reliability, returns, etc.).
There are various ways of dealing with this. Amazon [1] is already active in this space using what you might call the "user review" model combined with enforcing some exacting standards on their vendors. Already having a huge channel of course made it easier for Amazon to get vendors to accept this.
[1] Incidentally, I've never thought that Amazon did a great job with sorting products within its general categories, so I always search instead. But I do wonder, how many more people would window-shop at Amazon if their sorted product catalog was more interesting?
There's also the question of "how do you arrange the products in order to maximize profit". There's also a great deal of machine/learning data mining that could be involved in this: e.g. market/basket analysis ("people who bought product X, also bought product Y"). Safeway does this, Amazon does this.
http://en.wikipedia.org/wiki/Affinity_analysis
That would make this an ideal project for people with machine learning/data mining experience, particularly from a retail/e-commerce background (Amazon, eBay, Wallmart.com or even a brick and mortar store chain such as Safeway).
http://en.wikipedia.org/wiki/Affinity_analysis
That would make this an ideal project for people with machine learning/data mining experience, particularly from a retail/e-commerce background (Amazon, eBay, Wallmart.com or even a brick and mortar store chain such as Safeway).
What would be best paths through product space?
My first thought was “doesn’t Amazon already organize things independent of vendor?” But what we’re really looking for here is a recommendation engine, and Amazon really isn’t very good at that. As I see it there are three main steps which should be done in order. The more you know about ahead of time, the further you can skip ahead. Here are the steps:
Step 1. I don’t know what I’m looking for.
Maybe I’m looking for a good science fiction book. Or a gift for a 14-year-old niece. A deep gift recommendation engine alone could probably make a good startup. The engine should ask some probing questions and give recommendations—either specific products or product categories, depending on the type of question. This experience mimic the experience of having an experienced salesperson ask questions and point you towards good options. A beautiful browsing option would be a great way to conclude this step.
Step 2. I know what I want, but I haven’t decided which brand or model to get.
Here we need a deep comparison. The ability to rank priorities would be really helpful. Let’s say I’m buying a laptop. The site should ask me how much I care about weight, screen size, processing power, etc. It can then recommend models that fit within my budget. There’s a huge amount of domain-specific knowledge required to get this right for a variety of products.
Step 3. I know what product I want.
Here we need to give comparisons among vendors. Price, shipping speed, return policies, and other general customer feedback are important. This is where it helps to be a trusted third party.
This problem is a huge undertaking, which is why no one has come close to solving it yet. Frankly, unless the domain is very restricted, I don’t think a three person team will be able to do enough specific programming and testing to make it work.
The principles and mechanics would best be discovered in a very narrow marketplace, and one full of confusion. Electronics would be the natural stating place, but a less crowded market might also work.
My first thought was “doesn’t Amazon already organize things independent of vendor?” But what we’re really looking for here is a recommendation engine, and Amazon really isn’t very good at that. As I see it there are three main steps which should be done in order. The more you know about ahead of time, the further you can skip ahead. Here are the steps:
Step 1. I don’t know what I’m looking for.
Maybe I’m looking for a good science fiction book. Or a gift for a 14-year-old niece. A deep gift recommendation engine alone could probably make a good startup. The engine should ask some probing questions and give recommendations—either specific products or product categories, depending on the type of question. This experience mimic the experience of having an experienced salesperson ask questions and point you towards good options. A beautiful browsing option would be a great way to conclude this step.
Step 2. I know what I want, but I haven’t decided which brand or model to get.
Here we need a deep comparison. The ability to rank priorities would be really helpful. Let’s say I’m buying a laptop. The site should ask me how much I care about weight, screen size, processing power, etc. It can then recommend models that fit within my budget. There’s a huge amount of domain-specific knowledge required to get this right for a variety of products.
Step 3. I know what product I want.
Here we need to give comparisons among vendors. Price, shipping speed, return policies, and other general customer feedback are important. This is where it helps to be a trusted third party.
This problem is a huge undertaking, which is why no one has come close to solving it yet. Frankly, unless the domain is very restricted, I don’t think a three person team will be able to do enough specific programming and testing to make it work.
The principles and mechanics would best be discovered in a very narrow marketplace, and one full of confusion. Electronics would be the natural stating place, but a less crowded market might also work.
The problem with steps 2 and 3 is that most of the time people aren't really sure they know what they want. They may have narrowed things down to a category, such as laptops as you mentioned, but even then you'd be hard pressed to find someone that absolutely KNOWS they want a laptop. Maybe a netbook + desktop would be a better combo? Or an iMac-style nettop. Furthermore, at any given time people are usually in the market for more than one thing, all at various stages of the buying cycle.
So now if we can simplify and treat all cases as "I don't know what I'm looking for" then follow-up to that is "I'm looking".
In that case the experience can be optimized around the looking and searching. I think the key concept in searching is, unsurprisingly, speed. Let me look at a lot of products quickly, let me research an individual product quickly, jump back out quickly, jump between categories, jump between different lists, sorts, filters, etc.
And I think pg is missing something, as the path doesn't end when a product purchase is made. You have all the nitty gritty of the actual transaction + shipping, but then you have post-purchase life as well. Most stores have you buy the product and then have you go somewhere else for support. Sure you can review the product on the store, but what if you want to hack it? or you have a problem? or you want to upgrade? etc etc. Is it up to the manufacturer at that point? What if you can't find the manufacturer's support resources?
We're doing some interesting things here at ProductWiki, but our goal is to create a great base of content and data. I imagine that one day a company will take the open data to create a really nice and innovative shopping experience.
So now if we can simplify and treat all cases as "I don't know what I'm looking for" then follow-up to that is "I'm looking".
In that case the experience can be optimized around the looking and searching. I think the key concept in searching is, unsurprisingly, speed. Let me look at a lot of products quickly, let me research an individual product quickly, jump back out quickly, jump between categories, jump between different lists, sorts, filters, etc.
And I think pg is missing something, as the path doesn't end when a product purchase is made. You have all the nitty gritty of the actual transaction + shipping, but then you have post-purchase life as well. Most stores have you buy the product and then have you go somewhere else for support. Sure you can review the product on the store, but what if you want to hack it? or you have a problem? or you want to upgrade? etc etc. Is it up to the manufacturer at that point? What if you can't find the manufacturer's support resources?
We're doing some interesting things here at ProductWiki, but our goal is to create a great base of content and data. I imagine that one day a company will take the open data to create a really nice and innovative shopping experience.
You're totally right. Guiding people in the choice between even just the various computing options is a major task. This is just a big, big nut to try to crack..
To anyone attempting it, I'd suggest again starting in a very tightly focused domain. My pick would be cameras -- the classic "whattheheckdoIwanttobuyIhavenoideawhatthesenumbersmean" product.
To anyone attempting it, I'd suggest again starting in a very tightly focused domain. My pick would be cameras -- the classic "whattheheckdoIwanttobuyIhavenoideawhatthesenumbersmean" product.
If there's too much structure for the developers alone, then we need to enable the community to contribute that structure, in addition to content. But can the application be genericized enough to cover all of the shopping domains without being completely washed out?
Think about the different kinds of web content that you read when making buying decisions. Off the top of my head, there's recommendations/searches (to discover things), consumer reviews/ratings (to evaluate them) and expert "buying guide" type articles (to gain a deeper understanding of a non-trivial decision). Currently, using all of these sources involves a lot of Googling and jumping from site to site. What if a single site could encapsulate all of them?
This site would aggregate prices and host reviews of individual products. Products could also be tagged as belonging to any number of "aisles" of varying degrees of specificity like "Sporting Gear", "Wall Sconces", "Gifts for Mom", "2 Inch Tapered Drywall Screws", "Eco Friendly", and so on. The aisles would be user created and could be just about anything. Individual users can have their own personal aisle for things they own or want. Stores can have their own aisle too (which would be a likely area for monetization). Each aisle would have a hub page showing the highest rated products, sub-aisles and related aisles, external links, and user written guides, faqs and whatever else they want to write. The idea would be to make the content model flexible enough to adapt to the needs of any special domain and to the different situations consumers find themselves in.
Think about the different kinds of web content that you read when making buying decisions. Off the top of my head, there's recommendations/searches (to discover things), consumer reviews/ratings (to evaluate them) and expert "buying guide" type articles (to gain a deeper understanding of a non-trivial decision). Currently, using all of these sources involves a lot of Googling and jumping from site to site. What if a single site could encapsulate all of them?
This site would aggregate prices and host reviews of individual products. Products could also be tagged as belonging to any number of "aisles" of varying degrees of specificity like "Sporting Gear", "Wall Sconces", "Gifts for Mom", "2 Inch Tapered Drywall Screws", "Eco Friendly", and so on. The aisles would be user created and could be just about anything. Individual users can have their own personal aisle for things they own or want. Stores can have their own aisle too (which would be a likely area for monetization). Each aisle would have a hub page showing the highest rated products, sub-aisles and related aisles, external links, and user written guides, faqs and whatever else they want to write. The idea would be to make the content model flexible enough to adapt to the needs of any special domain and to the different situations consumers find themselves in.
Etsy is a fanciful/artistic version of this RFS.
http://www.etsy.com/
http://www.etsy.com/
Shopping is broken, by Scott Adams:
Let's say you have a problem or a need or a want, and you're sure there is product somewhere in the marketplace that would help. You Google, and a handful of web sites pop up that offer promising solutions. But obviously you don't believe anything you read from vendors, so you check for online reviews. Then you wonder if the favorable reviews are planted by the vendor, and the bad reviews are planted by his competitors. Can you trust reviews from anonymous strangers?
You ask your coworkers and friends if they have ever used the product that you're interested in, and no one has. What now?
Shopping is broken.
How much more stimulated would the economy be if the people who have money, and are willing to spend it, could be reliably connected with the products that they desire?
What the world really needs is some way to connect you with the people who already use the sorts of products you want, and are willing to answer an e-mail or two about the topic.
About a year ago I had surgery to fix my voice. The information on the Internet about that particular surgery was outdated and didn't address my questions. The only way I could become a consumer of that surgery was by communicating directly with people who already had it, which I did. And since then I have answered questions for dozens of people who have the same questions that I had.
Likewise, as my wife and I make a zillion decisions for the home we are building, we prefer products and solutions used by people that we have spoken to personally. The Internet is virtually useless for any of the hundreds of product decisions we have made so far.
And what about choosing a destination for a vacation? You're much more comfortable if you have spoken to someone who visited the same place.
The obvious problem with connecting past consumers with potential consumers is that while people are generally helpful by nature, no one wants a million e-mails asking how they like their new can opener. So how do you strike the right balance?
Imagine a system that works like this: When you buy a product, you agree in advance to answer up to four e-mails from future potential customers, beginning no sooner than one year from when you make your purchase. It's totally optional, but agreeing gives you access to people who already bought the product you're considering today, to help you make your own decision. It would strike you as a fair deal.
For privacy reasons, this imagined system would disguise your e-mail address. And the system would have to be administered by some third party, not the vendor selling the product, or you wouldn't trust the strangers giving you advice.
Maybe you have a better idea for fixing shopping."
From: http://www.dilbert.com/blog/entry/shopping_is_broken/
Let's say you have a problem or a need or a want, and you're sure there is product somewhere in the marketplace that would help. You Google, and a handful of web sites pop up that offer promising solutions. But obviously you don't believe anything you read from vendors, so you check for online reviews. Then you wonder if the favorable reviews are planted by the vendor, and the bad reviews are planted by his competitors. Can you trust reviews from anonymous strangers?
You ask your coworkers and friends if they have ever used the product that you're interested in, and no one has. What now?
Shopping is broken.
How much more stimulated would the economy be if the people who have money, and are willing to spend it, could be reliably connected with the products that they desire?
What the world really needs is some way to connect you with the people who already use the sorts of products you want, and are willing to answer an e-mail or two about the topic.
About a year ago I had surgery to fix my voice. The information on the Internet about that particular surgery was outdated and didn't address my questions. The only way I could become a consumer of that surgery was by communicating directly with people who already had it, which I did. And since then I have answered questions for dozens of people who have the same questions that I had.
Likewise, as my wife and I make a zillion decisions for the home we are building, we prefer products and solutions used by people that we have spoken to personally. The Internet is virtually useless for any of the hundreds of product decisions we have made so far.
And what about choosing a destination for a vacation? You're much more comfortable if you have spoken to someone who visited the same place.
The obvious problem with connecting past consumers with potential consumers is that while people are generally helpful by nature, no one wants a million e-mails asking how they like their new can opener. So how do you strike the right balance?
Imagine a system that works like this: When you buy a product, you agree in advance to answer up to four e-mails from future potential customers, beginning no sooner than one year from when you make your purchase. It's totally optional, but agreeing gives you access to people who already bought the product you're considering today, to help you make your own decision. It would strike you as a fair deal.
For privacy reasons, this imagined system would disguise your e-mail address. And the system would have to be administered by some third party, not the vendor selling the product, or you wouldn't trust the strangers giving you advice.
Maybe you have a better idea for fixing shopping."
From: http://www.dilbert.com/blog/entry/shopping_is_broken/
Interesting, but I still don't see how the fundamental trust issue is covered by the proposed solution. The system may reduce the chance that a reviewer is a shill, but it in no way establishes their credentials. All you know is that they purchased the product - and frankly this already suggests a level of bias.
People will continue to value the advice of those they know much more highly. The challenge is to establish a comparable level of trustworthiness in online reviews.
People will continue to value the advice of those they know much more highly. The challenge is to establish a comparable level of trustworthiness in online reviews.
You have to build a trustworthy, independent brand.
To be honest, really freakin' good graphic design goes a long way.
To be honest, really freakin' good graphic design goes a long way.
...the companies that were best at mundane things like answering questions, taking orders, packing and shipping them, and taking returns...
The companies that are best at these things are best precisely because they don't consider them to be "mundane".
In customer service and fulfillment, it's a fine line between being excellent and being out of business. These can be very difficult things to do, and the thinner your margins, the better you have to be.
And very few companies handle customer returns well. They keep getting better and better until they realize the best way to handle customer returns is to avoid them in the first place.
The companies that are best at these things are best precisely because they don't consider them to be "mundane".
In customer service and fulfillment, it's a fine line between being excellent and being out of business. These can be very difficult things to do, and the thinner your margins, the better you have to be.
And very few companies handle customer returns well. They keep getting better and better until they realize the best way to handle customer returns is to avoid them in the first place.
This sounds interesting. On a very high level, a good solution to this would feel like window-shopping in a huge mall. Amazon.com is great if you know what you're looking for, (as GavinB said), but it's doesn't work for discovery. I can't meander through Amazon.com to check out products and I wouldn't go there for fun. Compare that to your local mall, where people go to just hang out.
Shopping is often a pass time and a social experience, not a mission, yet most sites are optimized for the last "stage" of shopping--the stage where you actually make a purchase, or the penultimate stage--the one where you compare a few specific products. The last stages are the most profitable and easiest to implement, so that's why they came about first. But if you can grab a consumer at an earlier stage, when they're brainstorming bridal shower ideas, or when they're bored at work and checking out new cell phones, and if you optimize for discovery as well as sales, they'll come back and buy from you.
The challenges/pitfalls/competitors are numerous, obvious, and non-obvious. Good luck to the team that attempts this!
Shopping is often a pass time and a social experience, not a mission, yet most sites are optimized for the last "stage" of shopping--the stage where you actually make a purchase, or the penultimate stage--the one where you compare a few specific products. The last stages are the most profitable and easiest to implement, so that's why they came about first. But if you can grab a consumer at an earlier stage, when they're brainstorming bridal shower ideas, or when they're bored at work and checking out new cell phones, and if you optimize for discovery as well as sales, they'll come back and buy from you.
The challenges/pitfalls/competitors are numerous, obvious, and non-obvious. Good luck to the team that attempts this!
Why does it have to be a storefront? I think it would be better if it were an advertising solution.
"Online stores group products by who sells them."
Do you really find this to be true?
1) Go to http://www.newegg.com -- while brand/vendor is one selection criteria, so are many of the technical specs and options. They even offer advanced and power search.
2) Amazon is no different: While Brands are promoted (because consumers are trained to that) they offer power on the left side here (below brand): http://www.amazon.com/Digital-Cameras-Photo/b/ref=sv_p_4?ie=...
While I don't think that the statement in the RFS is accurate, there is no doubt in my mind that the online product finding experience has dramatic room for improvement and innovation.
Do you really find this to be true?
1) Go to http://www.newegg.com -- while brand/vendor is one selection criteria, so are many of the technical specs and options. They even offer advanced and power search.
2) Amazon is no different: While Brands are promoted (because consumers are trained to that) they offer power on the left side here (below brand): http://www.amazon.com/Digital-Cameras-Photo/b/ref=sv_p_4?ie=...
While I don't think that the statement in the RFS is accurate, there is no doubt in my mind that the online product finding experience has dramatic room for improvement and innovation.
I think PG's point is that NewEgg only lists products sold by NewEgg. Amazon is a little different since they have affiliated retailers, but they still don't have everything.
Having everything is not what I want.
I go to Fry's for fun things, Banana Republic for work clothes and Neiman Marcus for gifts for the girls who care about those sorts of gifts.
No website will ever give me all three in a sane format; nor do I care for it to do that. Nor does my mother.
I go to Fry's for fun things, Banana Republic for work clothes and Neiman Marcus for gifts for the girls who care about those sorts of gifts.
No website will ever give me all three in a sane format; nor do I care for it to do that. Nor does my mother.
My sense was that he wanted a generalized version of NewEgg. In other words a way to take all the products in the world and be able to create many lenses that view them. The products seen by the lenses could be determined by whatever criteria you / your customer / your customer's customer sees fit. That is how I understood it.
Disclaimer: You have never met anyone worse at following directions than me so that could be a misread.
Disclaimer: You have never met anyone worse at following directions than me so that could be a misread.
sounds like guided navigation, faceted search and relational database stuff.
It seems to me that a great interface for presenting products would almost necessarily have to be individually designed for each store. I don't see how you could use the same kind of interface to present, say, books and perfume.
Also, I have a hard time imagining how big stores could choose to use a presentation system similar to that of other stores. Unless I misunderstood the original idea, it seems that selling presentation software is not the right way to address that problem.
Also, I have a hard time imagining how big stores could choose to use a presentation system similar to that of other stores. Unless I misunderstood the original idea, it seems that selling presentation software is not the right way to address that problem.
So, basically FriendFeed for the consumer-facing part of online retail?
Having a good understanding of co-op advertising could matter a great deal here. Retailers get paid by manufacturers for product placements... Someone going after this model could try to reach similar deals with both the manufacturers and the retail-fulfillment entities.
Having a good understanding of co-op advertising could matter a great deal here. Retailers get paid by manufacturers for product placements... Someone going after this model could try to reach similar deals with both the manufacturers and the retail-fulfillment entities.
Community Driven Web Shopping
Internet tends to be an open system where consumers are in the same way important as the product makers and sellers. Success is measured by community support and it is assured by complete transparency, trust.
Marketing your online products also cannot be done well without a strong community. A future webshop must be completely user and community driven, your duty as a webshop owner is to build trust and make your existing customers to connect to your next customers.
A Classic Store vs. Auction Based Stores
Following years will be about a 180 grades shift in e-commerce. Now you are presenting products to customers as a classic store, later customers will ask you for products and services.
Yabe.hu -- Race. For you -- is a new approach to shopping. As a customer you log in and tell your desires then vendors are racing to offer you the best solution. This is the only way to assure yourself you'll get the best deal.
You don't have to choose, listen to commercials, browse social shopping sites and do research to find what you want; just spit out a desire and let the others do the job for you.
Until this approach to shopping hits mainstream you'll have to do the classic way. Look for a shop who makes you feel comfortable, do the shopping, show your items bought and generate trust for other users and would-be-customers. Wait for your cut offered by the webshop as an affiliate program.
A Classic Webshop In Modern Clothes
Internet tends to be an open system where consumers are in the same way important as the product makers and sellers. Success is measured by community support and it is assured by complete transparency, trust.
Marketing your online products also cannot be done well without a strong community. A future webshop must be completely user and community driven, your duty as a webshop owner is to build trust and make your existing customers to connect to your next customers.
A Classic Store vs. Auction Based Stores
Following years will be about a 180 grades shift in e-commerce. Now you are presenting products to customers as a classic store, later customers will ask you for products and services.
Yabe.hu -- Race. For you -- is a new approach to shopping. As a customer you log in and tell your desires then vendors are racing to offer you the best solution. This is the only way to assure yourself you'll get the best deal.
You don't have to choose, listen to commercials, browse social shopping sites and do research to find what you want; just spit out a desire and let the others do the job for you.
Until this approach to shopping hits mainstream you'll have to do the classic way. Look for a shop who makes you feel comfortable, do the shopping, show your items bought and generate trust for other users and would-be-customers. Wait for your cut offered by the webshop as an affiliate program.
A Classic Webshop In Modern Clothes
1. Browse to your favorite online (gadget) store.
2. You are sure the frontpage is already taylored to your shopping needs (via an Automated Recommendation Engine)
3. Select your products to be bought and check for the best price gathered from other (gadget) stores.
4. Make shopping and later add comments, recommendations, how-to's, FAQs to your products bought.
5. Other users will see your value added to existing products and when buying you'll get a fair cut from every sales.
6. Too boost your revenues use widgets with your products bought on your blog or favorite social network, do the marketing and sales for the webshop you like and which pays back your efforts.