9 ms·

Half a second delay caused a 20% drop in traffic

106 points by quoderat 18y ago

markup 18y ago
It's pretty much a non-news -- I think everyone is aware of this -- but backing it up with actual tests and data is great. Thanks for sharing the post
- teej 18y ago
  We had a discussion about this before, where I shared some rough stats from an application I used to run. http://news.ycombinator.com/item?id=415809 http://news.ycombinator.com/item?id=415809
- sounddust 18y ago
  I agree that everyone was probably aware that speed is important, but I always thought that as long as pages loaded in less than 1 or 2 seconds, then it wouldn't make much of a difference to most users. This post changes my point of view; I think I'm going to work harder to reduce page load times as much as possible instead of stopping at something that "seems fast"
cedsav 18y ago
Sorry to bring back the table vs. css layout debate, but since Google use tables for some parts of their layout, can we assume that this is because it renders faster (and therefore makes more money for them)?
electromagnetic 18y ago
I can understand this. Walmart is the most painful website to browse on the internet, because you go on usually looking for something or the price of something and it takes what feels like an eternity to get anywhere.
- pclark 18y ago
  you should try using GoDaddy.com
pg 18y ago
I heard Google also found that increasing speed increased users' perception of the quality of the search results. The reason was that when the site was faster, users did more searches. Naturally they found more stuff. But they attributed the gain to Google's cleverness rather than the extra work they'd done themselves.
- prakash 18y ago
  I have spoken to tons of folks on websites they think are fast, websites they would benchmark against -- in the majority of cases perception of fast loading website triumphs an apples-to-apples comparison of time taken to download & render a page. Perception plays a huge role when it comes to wining the mind games in the users head regd. page download times.
  - gruseom 18y ago
    Can you give examples of how to make loading perceived as faster? All I know is progress indicators.
    - russell 18y ago
      Things that make me think a site is too slow are staring at a blank screen while waiting for doubleclick and trying to find the skip button for an intro ad. A site should display its content first and then fill in the ads later. Engage your users while all the clutter catches up.
    - patio11 18y ago
      Here's an example which made me quite a bit of money. My website basically looks like this: AAAAAA CBBBBB CBBBBB CBBBBB CDDDDD CDDDDD CDDDDD CDDDDD Most users will have their screen break a bit above the end of the B block. (By design.) The heaviest part of the page is D. Eliminating it is not an attractive option -- the overall conversion rate suffers. The most valuable parts of the conversion pathway are in B and C. The first thing the customer sees is invariably A. Solution: a bit of technical magic to make sure that the website loads element in A->C->B->D order (there is a competing technical imperative which means that in the HTML it makes more sense to order them A->B->D->C), and significant microoptimization of how quickly B and C are minimally interactive to the user. (We're talking measuring improvements in the kilobytes or tens of microseconds range.) Since customers have to actively scroll down to engage the D block, and since most do not engage the D block before first engaging the B block, the fact that D is heavy and slow very rarely impacts their experience.
LostInTheWoods 18y ago
Couldn't one come to another conclusion here? In fact several other conclusions. For one, since users were getting more results on the page they didn't need to go to a second results page, thus causing the drop-off in traffic. If you're going to come to some grand "Aha!" moment, you should define your experiment. Were these the same users? on the same day? searching for the same stuff? What browsers were they using? Did they have their windows maximized? What was their screen size? ... and so on.
- quoderat 18y ago
  Though I've never been able to find the results and complete study methodology online (can anyone?), Mayer is a CS grad from Stanford, as well as head of the UI team at Google. I'm guessing she and her group know what they are doing. I'd be more than willing to bet that they controlled for these variables you mention, else the test would be useless. Sure, it's right to be skeptical, but they do this for a living. Given Google's stake in having accurate data, they probably bothered to do it correctly. (And I am not usually a defender of Google.)
  - ojbyrne 18y ago
    Sure, but not necessarily to report it to the press correctly.
- cedsav 18y ago
  "After a bit of looking, Marissa explained that they found an uncontrolled variable." I guess that means they controlled for the other variables.
  - dhimes 18y ago
    I'm getting more questions as I start to think about this. They get revenue from ads, not searches. Did they vary the number of ads displayed? Did people look at the first page of 10, not find what they saw, then check the ads, find what they saw (which may be a page two organic result), and click the ad? If that is what is happening, then I can see why Google would lose. That is, their main conclusion would be that people tend to check organic results first. If that's true then if a company that has to pay to be page 1 is organically moved there by increasing the number of results displayed, then G loses. They check this by seeing if revenue goes up when they limit the results to 5 ads.
prakash 18y ago
Maybe PG should try this in kicking his HN habit, introduce throttling features that will increase the time taken to render HN rather than focusing on optimizations, I know it would help a lot of other folks as well ;-)
- kirubakaran 18y ago
  I hope he doesn't make the site suck for everyone just to "help" the users who seem to need baby-sitters. Suburbs suck for the same reason. I know you are joking but I am really tired of all the Nanny State type stuff. "Oh I am just doing this for your own good. You will thank me later." is one of the most evil attitudes that I know. I am not able to express clearly why I think so and why I feel so strongly[#] about it. May be it is coz I find it patronizing. Also I find it extremely unhackerly. [#] My mind is screaming obscenities and I am feeling pretty angry.
  - whatusername 18y ago
    even for kids? Or only for adults? Also - there is something to be said for wisdom.
    - kirubakaran 18y ago
      I am sorry, I didn't understand what you've said. Can you please explain? (I really tried hard to decrypt it btw)
      - whatusername 18y ago
        Sorry for being obscure. I was responding to your line: ""Oh I am just doing this for your own good. You will thank me later." is one of the most evil attitudes that I know." I completely understand your objection to that phrase in a lot of adult situations. (Safety fence after safety fence, stopping people from hurting themselves with technical products, not letting you hack stuff). I get that completely. Lowest Common Denominator design sucks. What I was asking was whether you apply that thought accross the board.. My mind particularly went to a Parent>Child relationship. Learning by experience (ie - getting burnt) is a great thing, but sometimes you can/should learn from others experience. Sometimes someone older/more experienced/with more wisdom -- can stop you doing something for your own good. Perhaps seat-belt laws may be a good example?
ardell 18y ago
Another well-illustrated example of the fact that users don't know what they want, you have to A/B Test it out of them.
- zach 18y ago
  Too true. I like the optometrist who made me read tiny letters again and again instead of asking "better like this... or like this?" Because even I can't tell the difference at first sight.
harpastum 18y ago
I think the magnitude of impact might be due to the search engine market. Results of different search engines are usually indistinguishable (to most web users), so the most important factors become design and speed. In other markets the balance may be different. I'm not saying that speed isn't important for everyone, but having that killer feature, even if it slows down your site, might be what sets you apart from your competitors.
lallysingh 18y ago
So, where does this fit into net neutrality? ISPs can certainly sell QoS services to customers. For some applications, this is important (e.g. gaming). For others, it could be a nice competitive advantage. But then with more & more of the ATM frames dedicated to QoS, that'll leave more contention (thusly, higher latency & drop) for the rest. I'm just worried about who's going to use these facts for what ends.
danohuiginn 18y ago
so, why not load the first few results quickly, then fill in the rest later through ajax? It'll take more than .5 seconds to eyeball the first ten results.

106 points by quoderat 18y ago

It's pretty much a non-news -- I think everyone is aware of this -- but backing it up with actual tests and data is great. Thanks for sharing the post

teej 18y ago
We had a discussion about this before, where I shared some rough stats from an application I used to run. http://news.ycombinator.com/item?id=415809 http://news.ycombinator.com/item?id=415809
sounddust 18y ago
I agree that everyone was probably aware that speed is important, but I always thought that as long as pages loaded in less than 1 or 2 seconds, then it wouldn't make much of a difference to most users. This post changes my point of view; I think I'm going to work harder to reduce page load times as much as possible instead of stopping at something that "seems fast"

cedsav 18y ago

Sorry to bring back the table vs. css layout debate, but since Google use tables for some parts of their layout, can we assume that this is because it renders faster (and therefore makes more money for them)?

electromagnetic 18y ago

I can understand this. Walmart is the most painful website to browse on the internet, because you go on usually looking for something or the price of something and it takes what feels like an eternity to get anywhere.

pclark 18y ago
you should try using GoDaddy.com

pg 18y ago

I heard Google also found that increasing speed increased users' perception of the quality of the search results. The reason was that when the site was faster, users did more searches. Naturally they found more stuff. But they attributed the gain to Google's cleverness rather than the extra work they'd done themselves.

prakash 18y ago
I have spoken to tons of folks on websites they think are fast, websites they would benchmark against -- in the majority of cases perception of fast loading website triumphs an apples-to-apples comparison of time taken to download & render a page. Perception plays a huge role when it comes to wining the mind games in the users head regd. page download times.
- gruseom 18y ago
  Can you give examples of how to make loading perceived as faster? All I know is progress indicators.
  - russell 18y ago
    Things that make me think a site is too slow are staring at a blank screen while waiting for doubleclick and trying to find the skip button for an intro ad. A site should display its content first and then fill in the ads later. Engage your users while all the clutter catches up.
  - patio11 18y ago
    Here's an example which made me quite a bit of money. My website basically looks like this: AAAAAA CBBBBB CBBBBB CBBBBB CDDDDD CDDDDD CDDDDD CDDDDD Most users will have their screen break a bit above the end of the B block. (By design.) The heaviest part of the page is D. Eliminating it is not an attractive option -- the overall conversion rate suffers. The most valuable parts of the conversion pathway are in B and C. The first thing the customer sees is invariably A. Solution: a bit of technical magic to make sure that the website loads element in A->C->B->D order (there is a competing technical imperative which means that in the HTML it makes more sense to order them A->B->D->C), and significant microoptimization of how quickly B and C are minimally interactive to the user. (We're talking measuring improvements in the kilobytes or tens of microseconds range.) Since customers have to actively scroll down to engage the D block, and since most do not engage the D block before first engaging the B block, the fact that D is heavy and slow very rarely impacts their experience.

LostInTheWoods 18y ago

Couldn't one come to another conclusion here? In fact several other conclusions. For one, since users were getting more results on the page they didn't need to go to a second results page, thus causing the drop-off in traffic. If you're going to come to some grand "Aha!" moment, you should define your experiment. Were these the same users? on the same day? searching for the same stuff? What browsers were they using? Did they have their windows maximized? What was their screen size? ... and so on.

quoderat 18y ago
Though I've never been able to find the results and complete study methodology online (can anyone?), Mayer is a CS grad from Stanford, as well as head of the UI team at Google. I'm guessing she and her group know what they are doing. I'd be more than willing to bet that they controlled for these variables you mention, else the test would be useless. Sure, it's right to be skeptical, but they do this for a living. Given Google's stake in having accurate data, they probably bothered to do it correctly. (And I am not usually a defender of Google.)
- ojbyrne 18y ago
  Sure, but not necessarily to report it to the press correctly.
cedsav 18y ago
"After a bit of looking, Marissa explained that they found an uncontrolled variable." I guess that means they controlled for the other variables.
- dhimes 18y ago
  I'm getting more questions as I start to think about this. They get revenue from ads, not searches. Did they vary the number of ads displayed? Did people look at the first page of 10, not find what they saw, then check the ads, find what they saw (which may be a page two organic result), and click the ad? If that is what is happening, then I can see why Google would lose. That is, their main conclusion would be that people tend to check organic results first. If that's true then if a company that has to pay to be page 1 is organically moved there by increasing the number of results displayed, then G loses. They check this by seeing if revenue goes up when they limit the results to 5 ads.

prakash 18y ago

Maybe PG should try this in kicking his HN habit, introduce throttling features that will increase the time taken to render HN rather than focusing on optimizations, I know it would help a lot of other folks as well ;-)

kirubakaran 18y ago
I hope he doesn't make the site suck for everyone just to "help" the users who seem to need baby-sitters. Suburbs suck for the same reason. I know you are joking but I am really tired of all the Nanny State type stuff. "Oh I am just doing this for your own good. You will thank me later." is one of the most evil attitudes that I know. I am not able to express clearly why I think so and why I feel so strongly[#] about it. May be it is coz I find it patronizing. Also I find it extremely unhackerly. [#] My mind is screaming obscenities and I am feeling pretty angry.
- whatusername 18y ago
  even for kids? Or only for adults? Also - there is something to be said for wisdom.
  - kirubakaran 18y ago
    I am sorry, I didn't understand what you've said. Can you please explain? (I really tried hard to decrypt it btw)
    - whatusername 18y ago
      Sorry for being obscure. I was responding to your line: ""Oh I am just doing this for your own good. You will thank me later." is one of the most evil attitudes that I know." I completely understand your objection to that phrase in a lot of adult situations. (Safety fence after safety fence, stopping people from hurting themselves with technical products, not letting you hack stuff). I get that completely. Lowest Common Denominator design sucks. What I was asking was whether you apply that thought accross the board.. My mind particularly went to a Parent>Child relationship. Learning by experience (ie - getting burnt) is a great thing, but sometimes you can/should learn from others experience. Sometimes someone older/more experienced/with more wisdom -- can stop you doing something for your own good. Perhaps seat-belt laws may be a good example?

ardell 18y ago

Another well-illustrated example of the fact that users don't know what they want, you have to A/B Test it out of them.

zach 18y ago
Too true. I like the optometrist who made me read tiny letters again and again instead of asking "better like this... or like this?" Because even I can't tell the difference at first sight.

harpastum 18y ago

I think the magnitude of impact might be due to the search engine market. Results of different search engines are usually indistinguishable (to most web users), so the most important factors become design and speed. In other markets the balance may be different. I'm not saying that speed isn't important for everyone, but having that killer feature, even if it slows down your site, might be what sets you apart from your competitors.

lallysingh 18y ago

So, where does this fit into net neutrality? ISPs can certainly sell QoS services to customers. For some applications, this is important (e.g. gaming). For others, it could be a nice competitive advantage. But then with more & more of the ATM frames dedicated to QoS, that'll leave more contention (thusly, higher latency & drop) for the rest. I'm just worried about who's going to use these facts for what ends.

danohuiginn 18y ago

so, why not load the first few results quickly, then fill in the rest later through ajax? It'll take more than .5 seconds to eyeball the first ten results.