Paul Graham: The 10 Secrets of Selling Online (1997) | Hacker News Reader

Paul Graham: The 10 Secrets of Selling Online (1997)

web.archive.org

How ironic.

When we interviewed at YC, we were told that "selling products online is depressingly orthogonal to good design".

We gave a competitor's site (http://bit.ly/15mzF) as an example to compare with our site (http://stylous.com/).

We were told that "they probably understand things about user behavior that you guys don't"

I guess "High Production Values" (http://bit.ly/oC5Y8) and "Make Your Site Easy" (http://bit.ly/YYfr) don't apply anymore.

(To be fair, there were other reasons we weren't accepted.)

Pretty sure I'm gonna bail on hacker news until all the YC admission related posts die down.

This was last year.

Sorry, but I couldn't resist given the irony.

These certainly do apply.

The competitor's website is clear (and loads quickly.)

Keep working though, looks pretty good.

I had some trouble navigating your site. You guys should observe users navigating (look over their shoulders). Give them a few tasks, e.g. find a watch you like, find a shoe you like.

Your UI is simple, but not always intuitive.

Also, I kept thinking that if I scrolled down I would see more shoes on the shoes page (there isn't enough space between the bottom row of shoes and the border of the browser).

We do use the typical complement of in person user testing (headed out to do some today) and obsessing over Google Analytics data.

Right now our biggest issue is IE performance, but we also have some other UI elements that confuse people that we are trying to resolve (often think clicking on a product will add it to their favorites list).

I'd love to hear more about your navigation issues. If you are so inclined, please email me (rich@stylous.com) some more information about the issues you faced.

"We were told that "they probably understand things about user behavior that you guys don't""

Your site has good approach to design, but personally I feel lacks usability. One of the first things Im(and I guess any user) interested is price and brand. You don't show both. If I click thru some image and see its beyond what I want to pay for, Im disappointed. Instead have proper filters(price/brand/size/small description) still goes a long way(atleast for me).

EDIT: On second look, looks like you had a filter by brand/price in ur homepage, missed that.

I have to mouse over the picture of the shoe in order to see the price. That is extremely inefficient, and requires the user to have stellar short term memory.

We created the site with the assumption that the design of the products is more important to users than their specific prices (within a specific price range). So far, our users have validated this assumption.

> So far, our users have validated this assumption.

Then who cares what PG says on the issue?

Price is

Price is almost always the most important piece of data, and the potential customer wants to see it right away.

I would for sure do research before putting prices up first.

In a store, people look around for items they like and then check the price. Sure, many other stores have prices listed there, but that doesn't make it the best thing for you to do.

The current workflow is more like window shopping, which could definitely serve a large audience. This isn't selling computer parts - you want the users to become attached to items before trying to sell it to them.

When you're selling high-end handbags the distinction between $895 and $1,100 is not really the first thing on the mind of the client.

I can see why one would be interested in the price of something they like, but do you particularly care about the price of items which you do not like? After all, these are fashion items not bushels of wheat.

If you like most of the items, I can see how it would be convenient to always show the price. But if you only like a small % and can easily access their price via a mouse over, then showing the prices by default costs you visual clutter with little benefit.

Beware the self-validating assumption. Your users are your users because they like your site better than, say Zappos. By definition, your users prefer your site. But as long as your market share is only a tiny fraction of the whole market, you can't say anything about the rest of the population.

It's only when you acquire a substantial fraction of the entire market that you can claim to have validated assumptions about the market in general.

This is a great point!

While PG suggests that it's better to have a small number of rapid supporters than a large number of moderate supporters, it's important to find the maximum on the curve.

The question is how many rapid supporters would you lose and gain by adding prices in more "accessible" locations? How less rabid would your users become by such a change? These are important questions that would need to be answered to gain market share.

My gut is that the fact that you have to hover over the image to see the price is detracting. By comparison, you could show the price slightly more prominently, without having it be an eyesore (show in a light grey under the item picture, that maybe highlights to black when hovered over). This would, imo, make the price conscious happier, without getting in the way of design.

Even when you have market share, it's not clear if your assumptions are validated as it is difficult to isolate variables in the real world.

For example, is Microsoft or MySpace dominant because of their UI or despite their UI? Because of their development practices or despite them? Or was their principle asset that of being a first mover? How does one separate those variables?

We created the site with the assumption

Why not create a site with a data-gathering model that prompts updates in design, and then test the design with real-world data?

It doesn't make sense for bargain-hunting, but I actually think this is better from a botique shopping perspective.

My complaint is with the scrolling. If I go under shoes, I get this weird horizontal scrolling business, and although you've done a lot to make it feel better by mapping the vertical scroll wheel to a horizontal scroll, it's still very awkward because it works differently than pretty much everything else out there.

A few thoughts:

Your page feels extremely sluggish.

On your site, you have to hover over things to see the prices. That's not good design. On your competitor's site, you're able to quickly scan the page (and use the mouse scrolly to get more items instantly). I see the scrolly does work on your horizontal scrollbar, but it's slow, and counterintuitive. I didn't even think to use the scrolly to move back and forth. Intuitively, the scrolly is a vertical thing.

Key phrase: Form follows function. I would almost certainly not want to buy something from your site because it takes the whole point of ecommerce (find and buy quickly) and replaces it with the brick and morter approach (find something, check the price, set it down, check something else), and does so sluggishly (clicking left and right, or using the little js scrollbar takes like 5-10 seconds to load - way to slow).

Your site looks nice and clean, but that's about it. I wouldn't call it usable at all.

Make it usable, make finding something fast.

Addendum: I think the biggest problem with the usability of your site is the excessive reliance on js animation. My machine isn't exactly old (core2duo, 4gig ram), and it feels like playing a modern FPS game on a netbook.

What's wrong with just using a div with overflow:auto, rather than over-designing it.

Another thought: perhaps allow the bottom part to stretch fill the whole page, so that you can see more items at a time. Perhaps some scale-option to decrease or increase the scale of the pictures allowing more detail and less items perpage, or less detail and more images?

Sorry about the performance issues. The site is implemented in pure javascript. We are still working out the kinks in slow browsers.

Funny, as you posted that, I was adding my "excessive js animation" addendum.

The back button working even with the zooms to the next level is KEY for your audience. Well done.

It shows up as a blank page for me, with Firebug noting a Javascript error. Seems to work fine in Safari. It wouldn't hurt to make your site be able and degrade smoothly. At least show me a message that something happened, right now it just looks like you don't exist. Having it able to degrade would also let you get search engine traffic.

Indeed. Unless it has some user-agent detection for Googlebot and other search spiders, it's completely immune to search engine spiders.

I think it'd be a worthwhile experiment to add an HTML version (like Gmail) just to see the results. It should be easy enough to retain the design without all the excessive javascript.

And I'm sure a super-simple version would take much less time to develop than the current pure-js approach.

Yeah that only happens when you have FireBug -- need to fix that one

http://stylous.com is a really great design -- good job.

You approached the interview incorrectly. At the interview, the YC partners will play devil's advocate without prefacing what they are saying with "You know, I am just playing devil's advocate here". There are multiple ways to respond - you can defend your idea to the death, you can respond with reasons why you either think they are wrong, or you can indicate that you are willing to change your idea substantially. I am willing to bet that you guys just folded.

Actually, we were rejected because we held to our idea too strongly. PG thought that our feelings on the matter were too strong given our level of experience.

I don't hold that against him at all, I just couldn't resist responding to his claim that good design and success at online retailing were somehow at odds.

Sadly, I'd have to say from personal experience that this is true.

But then, I also think that "good design" or more like "stylish design" is orthogonal to "Make Your Site Easy" quite often.

Your site is gorgeous (FF/Mac). But maybe others tried it in Safari/Mac (w/JS turned off as I keep my Safari) - and got a blank home page (what I get right now)?

Maybe some safe-degrade when javascript off is appropropriate. At least a screen alert saying, 'please turn on javascript'. I see that a fair bit.

To be fair to you, I think the "bad" design that you're commenting on is much more conducive to actually creating a sale to your nifty page.

Usability and familiarity go hand in hand.

seen suspicious serif fonts, inspected some text and seen font-family:'Arial';

any chance of catering for those of us without arial font-family:Arial, sans-serif;

It should default to sans. Thanks for pointing that out.

"Good design" can mean a lot of things. In this case, I'm sure PG meant pretty design. Your site is pretty-- but it fails a lot of design tests if your goal is to sell stuff (IMO).

Selling stuff online isn't an artistic exercise-- it's an analytics and psychology exercise.

To be fair, that PG article is pretty ancient and commerce was a different beast back then (i.e. you might have needed high production values to overcome people's fear of online purchasing where you don't today). "Make your site easy" obviously still holds true. Are you contending that your site is easier than the competitors? It'd be interesting to write down a few use cases and then time user success on your site versus your competitor.

Related: part of good commerce design is SEO. A huge huge part. http://www.google.com/search?q=site%3Astylous.com

"Selling stuff online isn't an artistic exercise-- it's an analytics and psychology exercise."

Does these imply Apple does sell much online?

I do like the idea of setting up some tests and doing some measurements. The tricky thing is testing for the right thing. Guys typically think of shopping as: I want X, I search for X, I buy X. But when it comes to women and fashion it's about window shopping - not knowing what you quite want until you see it. How does one measure a UI for such a use case? Browsed items per minute? Number of items of interest found per until time?

I wasn't sure if you were soliciting feedback on your site, but since that seems to be the current direction of the posts...

Overall, I like this concept. I think the "browsing items" effect is pretty cool. As others have suggested, it would be nice to display prices at some point. At the very least, make it contextual. In other words, if someone is going to select one of the price filters, this should be an indication that price is something they are now interested in seeing.

The site looks nice in FF, and obviously has issues in IE. The fact that "iesucks" shows up in the Internet Explorer URL history actually made me smile, but pretty much invalidated my opinion of your site as a serious consumer service.

Heh oops didn't realize that would appear in the browser history. Will have to change that.

One guy named Steve Jobs once said, that design is not how it looks but how it works. He had a point.

As opposed to everyone else, I liked the menus a lot. However, the actual product selection was completely counterintuitive.

i have sites with hovering ajax and yes the performance can be slow (it's the ajax)

i don't like 'loading' message on every ajax call so what did i do? i pre-load every js and combine few images into one (not sprites due to dynamic nature) to reduce http request

in your case, it could be: sprite the shoe.jpg to watch.jpg to top.jpg sprite the pumps.jpg ... evening.jpg to middle-shoe.jpg sprite the first 6 shoes into down-shoe.jpg

so for these sprites: 1 + 4 + (10 + 6 + 6 + 4) = 31 files

or you can combine into one big file

the point is not to make user wait for the 'first impression' when (for example) s/he clicks handbags > baby

if s/he wants to click next after seeing the first items, then ajax call can be used (make it one file instead of six) but pre-load the prices into js (or <bad> put price into the file-name like shoe-2-59-49-99-29-39-79.jpg and parse -, map the prices </bad>) so you only need to call 1 instead of 2 or 6 or more http

I hate to break it to you, but with your product line, it looks like your main competitor isn't become.com, but Amazon's www.endless.com ... let me say that again: you're competing at online retailing with an Amazon.com subsidiary. Good luck.

No, we get 15% when someone buys anything on endless.com ;-)

Right, but don't you think Amazon has more resources/experience in layout/design which gets people to buy things? Or are you looking to target a niche of people who prefer a minimalist design?

Yes, the same niche that buys iPods and iPhones. Minimalism is about usability, not lack of functionality.

Why post the archive.org version when http://store.yahoo.com/secrets.html is still up?

One reason: they've pulled some of the links out of the current page. The wayback machine has the original links to the other archived sites.

Dupe: http://news.ycombinator.com/item?id=320852

But I'm glad it's got some attention this time.

I would really appreciate if the title gets changed to put [2001]...

for example: CDNOW

Reply on news.ycombinator.com