Researchers Say Web Visitors Could Bolt After Just 250ms Delay

New research suggests that web site vistors are even more impatient than we believed. How impatient? How about they are likely to leave after just a 250 milliseconds delay in loading?

read more

USAJobs.gov relaunch hampered by performance issues

When the federal government launched an updated version of the USAJobs.gov web site recently, it was hampered by performance issues and problems with the search engine, issues that should have been picked up in pre-launch performance testing.

read more

9 Seconds to Make an Impression

If it’s true that you only have 9 seconds to make an impression on your website, you better be sure that you make the most of it from a website performance perspective.

read more

Amazon Testing New Web Site Design

Amazon is reportedly redesigning its web site. When a site gets redesigned like this it requires a lot of pre-monitoring and testing before the launch, and careful analysis in the days following the live launch.

read more

Open Source Web Site Performance Tools

Louis St-Amour wrote a very helpful comment on my “The Complete List of End-User Experience Monitoring Tools” post – helpful enough that I thought it deserved a whole post. OWA, indeed commonly means Outlook, so it’s kind of unfortunate naming clash, but in this case I’m referring to Open Web Analytics from http://www.openwebanalytics.com/ which has what it calls, “domstream recording”. This lets you see exactly what a user sees in their browser and where their mouse goes, though I’m not sure how embedded it is for performance timing or error catching. A similar feature is in piwik as a plugin, as I understand it. This brings up an interesting point that I heard a number of times at this year’s Velocity conference – when will web analytics start providing better web performance data?  Yes, I know that Google Analytics has a Site Speed Analytics Report – but that is just a start. PageSpeed is not just a Firefox/Chrome extension  It’s a web service now too! Seriously though, what I was referring to was mod_pagespeed, the Apache module, which beyond all its other fun goodies has an “instrumentation” feature that — you guessed it: measures “the time the client spends loading and rendering the page, and report that measurement back to the server”http://code.google.com/speed/page-speed/docs/filter-instrumentation-add.html Google helpfully mentions that their instrumentation feature is not needed if you already have one of your own and cites Boomerang, Episodes or Jiffy: – http://yahoo.github.com/boomerang/doc/ – http://stevesouders.com/episodes/paper.php – http://code.google.com/p/jiffy-web/ It is interesting that I have not found enterprises leverage these wonderful free tools (if you work for a large enterprise and use these tools – please comment!). I believe the main issue for enterprises is that there is a lot of work that one has to do in order to build a more complete solution from these free tools – a lot of it is around displaying the data in a meaningful way. Sampling is indeed the worst part about Google Analytics … sigh. Sampling is a big problem when it comes to identifying problems. I see a lot of the new free end user experience monitors which only sample end user response times. Thing is – it gives their marketing department what they need – but it leaves their customers guessing when problems arise. The reason problems with IT systems are so hard to catch is that they are typically erratic and unpredictable. Without a complete data set it becomes difficult to prioritize and troubleshoot. Also, I would suggest that “real end user experience” tracking is something you’d need to be Google Chrome to get. After all, as Google notes about its own filter: “Note that the data reported by this filter is only approximate and does not include time the client spends resolving your domain, opening a connection, and waiting for the first few bytes of HTML.” Of course, that’s where the highly technical Speed Tracer comes in http://code.google.com/webtoolkit/speedtracer/ I agree that most vendors do not do “real end user experience tracking” – they typically only track the datacenter’s contribution in an accurate way. So that is definitely something to look out for – especially if the applications you are responsible for serve company employees. Often times just being able to identify that a slow desktop is the reason a user is unhappy can go a long way. Are there any other Web Site Performance tools worth mentioning? Please leave me a comment and I will add it to the...

read more