Airline reservations systems are kind of fascinating, because they were some of the first really complex computer systems, and a lot of those crusty old bits are still chugging along fifty odd years later, with newer and newer piles of code layered on top like a demented tower of jenga.
There are ones with no TPF at the bottom now. And most res systems have no old tech in the shopping part... that's more often in the booking, change/cancel and check-in flows.
The customer expectations are much lower when shopping for a plane ticket. A painfully long experience is actually expected.
Friction is absent when buying anything from Amazon (because of webperf, UX, etc) so customer expectations standard is way higher. If something goes wrong on Target, Mediamarkt or Carrefour website, the sale is gone and off the customer goes to Amazon.
As a similar example, in France the main way to buy a train ticket was through the railway website, whose UX was poor (as were travel agencies). A small team went at it and built a competing booking website (Captaine Train) whose sole promise was a quick and easy to book a train ticket. They instantly removed the above mentioned friction and became a huge competitor almost overnight.
AliExpress.com is the fastest e-commerce site I have experienced.
It was very apparent when I first browsed it - I wasn't even looking for speed, it simply jumped out at me. Especially the search.
What is particularly striking about this article is that Walmart was - in 2002 - publishing influential studies about the correlation between site speed and conversion rate:
I work for Walmart Labs. While your comment is funny, I just came here to say that this is not really how Walmart works.
The technology division of Walmart known as Walmart Labs that is responsible for developing and maintaining the website along with all backend infrastructure performs A/B testing for every new feature (including custom font) that is introduced. Any change to production systems or applications goes through a rigorous review process which involves only engineers (not managers).
No PM in Walmart can define a non-negotiable feature like this. Walmart Labs has a strong engineering culture and feature definition and testing is largely driven by engineers. Overriding a developer's recommendation the way you describe would be found to be in very poor taste.
The report in this article was created prior to the current redesign of the Walmart website. Back then a lot of technical debt had accumulated. The website has been redesigned now and a lot of that technical debt has been eliminated. The website is much faster now. Please give it a try.
You had me with this: "Any change to production systems or applications goes through a rigorous review process which involves only engineers (not managers)."
I'm going to spend some time looking around the site today.
> Walmart is one of the top USA e-commerce retailers. In 2016, they were the second after Amazon by sales.
> In e-commerce, the conversion is directly affected by how fast the site loads.
Maybe site performance matters a whole lot less than developers think it does. At least at this scale. Is the claim that users get tired of waiting and head over to Amazon? Seems like there would be a lot of other factors in a decision like, most of which would be far more important than load time.
Or do they get tired of waiting and just decide not to purchase something? Do they end up just purchasing it later? If they don't, then I guess people didn't need whatever they were trying to buy in the first place? If that's the case, maybe slow page load times are a good thing for society as a whole. Perhaps we should be celebrating them as a check on empty materialistic impulse buying.
FWIW Personally I prefer Walmart's site. I find it easier to use, I like that I can sort by best sellers, and there seems to be less advertising noise in the results. However it has a high minimum to get free shipping, a smaller selection of goods, and some things seem to be more expensive.
But overall I was surprised by how good it turned out to be - give it a whirl if you've never tried it. Amazon is ahead of the curve, but I'm starting to wonder if they'll manage to hold onto that lead. A few missteps and things could look completely different in a decade.
I don't think it's as simple as "user closes the window after 1.87 seconds thus you lose a sale". I don't even think most people can move that fast. But if one site is regularly faster then it's less frustrating and you'll use it more often. That's true of commerce sites, text editors, IDEs, etc. Unless the faster one is frustrating in other ways.
Wouldn't it be better to show the intended font at first page (using font-display: swap or fallback instead of optional) instead of a potentially uglier fallback font until second page?
With swap (or fallback) the page is visible early using a fallback font but you still end up with your "goal font" without having to reload or visit another page
I joined the Edge Foundation team @WalmartLabs about a year ago. We help manage the HTTP ingress flow (CDNs, WAF, Tier 1/2 proxies and LBs, etc) for all Walmart eCommerce properties. Our custom proxy also does some basic and not so basic frontend optimizations (js and css compression/minification/bundling, image resizing, url rewrites, etc) and handles a large portion of routing requests to backend servers.
This is a great write up from earlier this year, and from the screenshots it looks like it was before our new site design launched. That was a pretty massive modernization effort that included some site performance improvements. We also now have a sizeable team (100+) focused on some of the other details from the article. Brotli for instance went into A/B testing this month.
There are a lot of moving parts and changing priorities at a company the size of Walmart, and development and deployments can be slower than at smaller shops. But there definitely is active progress on improving our site speed.
And if you like working on any of these kinds of systems we are always hiring!
Guessing you haven't worked on a large distributed system or been responsible for revenue-producing LOB software deployments managing hundreds of millions in revenue. "Shit" like this is important. Like lose-your-job important.
Glad to hear it. I find Walmart's web site to be so slow on my PC to be almost unusable. I hate to think what the experience must be like for users on old computers or really low end phones.
On older PC, dsl (rural), Windows 10, and latest Chrome. It is more than acceptable for me because it works. Lots of cute websites don't because of broken JavaScript or slow library downloads.
Some good points, but I would suggest using "font-display: swap;" instead of "font-display: optional;" This will avoid the problems outlined and will actually replace the fallback font with the custom one without having to reload the page.
30 comments
[ 2.9 ms ] story [ 101 ms ] thread- bestbuy.com
- homedepot.com
- wayfair.com
- target.com
- nike.com
- hm.com
http://thecodist.com/article/how_flight_reservations_work_pa...
http://thecodist.com/article/how_much_is_that_seat_by_the_wi...
Friction is absent when buying anything from Amazon (because of webperf, UX, etc) so customer expectations standard is way higher. If something goes wrong on Target, Mediamarkt or Carrefour website, the sale is gone and off the customer goes to Amazon.
As a similar example, in France the main way to buy a train ticket was through the railway website, whose UX was poor (as were travel agencies). A small team went at it and built a competing booking website (Captaine Train) whose sole promise was a quick and easy to book a train ticket. They instantly removed the above mentioned friction and became a huge competitor almost overnight.
http://www.webperformancetoday.com/2012/02/28/4-awesome-slid...
Get back to the basics guys!
Dev: faster page loads make us more money!
PM: we NEED to use this custom font. This is non-negotiable.
The technology division of Walmart known as Walmart Labs that is responsible for developing and maintaining the website along with all backend infrastructure performs A/B testing for every new feature (including custom font) that is introduced. Any change to production systems or applications goes through a rigorous review process which involves only engineers (not managers).
No PM in Walmart can define a non-negotiable feature like this. Walmart Labs has a strong engineering culture and feature definition and testing is largely driven by engineers. Overriding a developer's recommendation the way you describe would be found to be in very poor taste.
The report in this article was created prior to the current redesign of the Walmart website. Back then a lot of technical debt had accumulated. The website has been redesigned now and a lot of that technical debt has been eliminated. The website is much faster now. Please give it a try.
I'm going to spend some time looking around the site today.
> Walmart is one of the top USA e-commerce retailers. In 2016, they were the second after Amazon by sales.
> In e-commerce, the conversion is directly affected by how fast the site loads.
Maybe site performance matters a whole lot less than developers think it does. At least at this scale. Is the claim that users get tired of waiting and head over to Amazon? Seems like there would be a lot of other factors in a decision like, most of which would be far more important than load time.
Or do they get tired of waiting and just decide not to purchase something? Do they end up just purchasing it later? If they don't, then I guess people didn't need whatever they were trying to buy in the first place? If that's the case, maybe slow page load times are a good thing for society as a whole. Perhaps we should be celebrating them as a check on empty materialistic impulse buying.
FWIW Personally I prefer Walmart's site. I find it easier to use, I like that I can sort by best sellers, and there seems to be less advertising noise in the results. However it has a high minimum to get free shipping, a smaller selection of goods, and some things seem to be more expensive.
But overall I was surprised by how good it turned out to be - give it a whirl if you've never tried it. Amazon is ahead of the curve, but I'm starting to wonder if they'll manage to hold onto that lead. A few missteps and things could look completely different in a decade.
With swap (or fallback) the page is visible early using a fallback font but you still end up with your "goal font" without having to reload or visit another page
This is a great write up from earlier this year, and from the screenshots it looks like it was before our new site design launched. That was a pretty massive modernization effort that included some site performance improvements. We also now have a sizeable team (100+) focused on some of the other details from the article. Brotli for instance went into A/B testing this month.
There are a lot of moving parts and changing priorities at a company the size of Walmart, and development and deployments can be slower than at smaller shops. But there definitely is active progress on improving our site speed.
And if you like working on any of these kinds of systems we are always hiring!
Why would you put a better compression algorithm into A/B testing? That doesn't make any sense at all.
>deployments can be slower than at smaller shops
No shit, if you do things like the above.
It's on the level as the (in)famous A/B testing of shades of blue that most people cannot distinguish.
And I bet it was made by just one guy in a few weeks too.
1. No dead ends from distractions or on purpose. Amazon usually forces you to screens you didn’t want to see. 2. Checkout is faster or just as fast.
And if it’s wrong, it easy to return the same day for credit.