Research

The Performance Metrics Project: Would you participate?

23 Jun 2010

0 Comment

Today at Velocity (1:00pm in Ballroom AB), I’ll be presenting important data that Strangeloop has gathered from our customers. It shows some interesting trends that correlate performance and key business metrics.

Here’s a sneak peek at one of my slides:

Web Performance Optimization: How landing page speed affects conversion rate

Customers who use our device or our service can easily opt-in to supply their performance and metrics data to help us gather exactly this kind of information. Most do, which is incredibly generous of them.

The cool thing is that it’s easy for anyone to contribute data, not just Strangeloop customers. All you have to do is insert some Javascript code into your page, the same way you do for things like Google Analytics.

Which leads me to my question:

Would you consider participating in an open project wherein contributors shared their performance and metrics data in order to gather and analyze realtime data from real websites? Would your organization? Is such a project even worthwhile? I’m keen on this idea, and I’m working to garner universal support within Strangeloop. However, opening this kind of data-gathering up to a broader audience will take time and effort. In your opinion, is it worth it?

Please give me your feedback and level of interest in the comments or by email: joshua@webperformancetoday.com.

Related posts:

By: Joshua Bixby Category: Research Tags: , , ,

From the Bottom Line to “Faster by Default”

21 Jun 2010

0 Comment

I just arrived here in Santa Clara for Velocity. I’ve got a million things to do to get ready for the sessions and keynote I’ll be presenting, but I just wanted to jot down a few thoughts I had on the plane about the evolution of web performance.

Performance has come a long way in an incredibly short time. It doesn’t seem that long ago (2007, to be specific) when Steve Souders was evangelizing the nitty gritty of the newly developed YSlow best practices.

Here’s a snapshot of how the performance landscape has changed in the past three years:

April 2007 – At Web 2.0 Expo, Steve laid the foundation by talking about the research projects that led to the development of the YSlow rules.

June 2008 – At Velocity 2008, Steve moved the conversation to metrics by asking “Fast web pages are critical to a good user experience, but what is “response time” and how should it be measured?”

June 2009 – At Velocity last year, Steve moved on to accountability and the impact of performance on the bottom line. He presented the new best practices he developed at Google, then scrutinized some popular websites to see how well they heeded the rules.

June 2010 – As you may know, the theme of this year’s Velocity conference is “Faster by Default”. What this means to me is recognizing that, given constantly permutating variables such as browser standards and dynamically delivered websites, we need to recognize the complexity of the problem and the complexity of the solution – or, more accurately, the set of solutions – needed to fix the problem. And wherever possible, we need to automate our solutions to respond nimbly and immediately.

By: Joshua Bixby Category: Research Tags: , , , , ,

Cheat Sheet: Everything you wanted to know about web performance but were afraid to ask

15 Jun 2010

42 Comments

[For a comprehensive directory of performance-related research, tools, and tips, visit the Web Performance Hub. To see some of the most compelling stats in visual form, check out these cool infographics.]

I don’t know about you, but I’m a freak for stats. Up until a couple of years ago, though, my jones for web performance numbers had to go unsatisfied, because the research landscape was pretty barren.

Fortunately, the landscape is looking a lot less bleak these days. Now everyone from Amazon to Google has publicized their findings – alongside research heavy hitters like Forrester and Aberdeen – about how site speed improvements affect user behavior and business metrics such as conversion and revenue.

For obvious reasons, I have a vested interest in collecting all this data. In my line of work, I never know when I’m going to need to pull out a stat on something like, say, how many stock brokers would jump ship on their brokerage as a result of poor mobile web performance. (Answer: 57%) I like to be ready for every question – and if I’m not totally ready, I at least want to have enough information that I can make a decent guess without sounding full of BS.

If you’ve ever found yourself trying to rationalize performance spending to your boss or co-workers – or if you, like me, have an unhealthy relationship with statistics – here’s a cheat sheet of stats and sources you can pull from for your next proposal or awkward elevator conversation.

User behavior and expectations about web performance

In 2006, the average online shopper expected a web page to load in 4 seconds. Today, that same shopper expects your page to load in 2 seconds or less. [Source: Forrester Consulting]

  • Web site performance is second only to security in user expectations.
  • 56% of online bankers and brokers expect web pages to load in 2 seconds or less.
  • Poor website performance leads to dissatisfaction more often than any other factor. Sixty-four percent of online US bankers and brokers have had a dissatisfying experience when servicing their accounts. Web performance is far and away the biggest reason for this dissatisfaction.
  • As online tasks get more urgent or complex, online users are less likely to try later and more likely to move to more expensive channels to complete the transactions. Fifty-six percent of online bankers would move to offline channels to ask a general account question; 54% of online brokers would move to offline channels to trade investments if the website was unavailable or slow to respond.
  • 48% of online bankers and brokers said that poor performance had an impact or significant impact on their likeliness to recommend a firm’s services to a friend or family member. [Source: Forrester Consulting]

Users who experience a 2-second site slowdown make almost 2% fewer queries, click 3.75% less often, and report being significantly less satisfied with their overall experience. [Source: Microsoft and Google]

  • 57% of online consumers will abandon a site after waiting 3 seconds for a page to load.
  • 8 out of 10 people will not return to a site after a disappointing experience.
  • Of these, 3 will go on to tell others about their experience. [Source: PhoCusWright]

A user who has to endure an 8-second download delay spends only 1% of her total viewing time looking at the featured promotional space on a landing page. In contrast, a user who receives instantaneous page rendering spends 20% of viewing time within the promotional area. [Source: Jakob Nielsen]

  • Nearly one-third (32 percent) of consumers will start abandoning slow sites between one and five seconds.
  • 39% say speed is more important than functionality for most websites, while only one in five rank greater site functionality as more important.

Between 37 to 49% of users who experience performance issues when completing a transaction will either abandon the site or switch to a competitor. Of these, 77% will share their experiences with others. [Source: Sean Power, Metrics 101]

Users have faulty perceptions of time. For example, the average person will perceive the amount of time it takes a page to load as being about 15% slower than the actual load time. And later, in recalling their memory of how long it took for the page to load, they will remember it took about 35% longer than it actually did. [Source: Stoyan Stefanov, Psychology of Performance]

Site speed and its impact on ecommerce metrics: Revenue, conversions, page views and downloads

Issues with application performance are affecting overall business revenues by up to 9%. [Source: Aberdeen Group]

A 1-second delay in page load time equals 11% fewer page views, a 16% decrease in customer satisfaction, and 7% loss in conversions. (In dollar terms, this means that if your site typically earns $100,000 a day, this year you could lose $2.5 million in sales.) [Source: Aberdeen Group]

  • A site that loads in 3 seconds experiences 22% fewer page views and a 50% higher bounce rate than a site that loads in 1 second. Impact on conversions: -22%.
  • A site that loads in 5 seconds experiences 35% fewer page views and a 105% higher bounce rate. Impact on conversions: -38%.
  • A site that loads in 10 seconds experiences 46% fewer page views and a 135% higher bounce rate. Impact on conversions: -42%. [Source: Strangeloop]

Shopzilla sped up its average page load time from 6 seconds to 1.2 seconds and experienced a 12% increase in revenue and a 25% increase in page views. It also doubled the number of sessions from search engine marketing and cut the number of required servers in half. [Source: Shopzilla]

Amazon found that it increased revenue by 1% for every 100 milliseconds of improvement. [Source: Amazon]

Mozilla shaved 2.2 seconds off their landing pages, thereby increasing download conversions by 15.4%, which they estimate will result in 60 million more Firefox downloads per year. [Source: Mozilla]

Performance optimization improved conversion rate by 16.07% and cart size by 5.51%. [Source: Watching Websites]

Yahoo increased traffic by 9% for every 400 milliseconds of improvement. [Source: Yahoo]

Visitors in the top ten percentile of site speed viewed 50% more pages than visitors in the bottom ten percentile. [Source: AOL]

Speed and the mobile web

“Sixteen percent of consumers have shopped via mobile phones or smartphones, but 27% of them report that it is dissatisfying due to the mobile shopping experience being too slow. One-third of consumers report wanting to shop via their smartphones in the future, with 5% indicating that this will be an important aspect of their loyalty to online retailers. [Source: Forrester Consulting]

  • More than half of mobile device users say that they expect sites to download as quickly on their mobile devices as they do on their home computers.
  • 60% of mobile web users have had a problem when accessing a website in the past year.
  • Slow load time was the number one issue faced by almost three quarters of them.
  • Three out of 5 say that poor performance will make them less likely to return to the site.
  • 40% said they’d likely visit a competitor’s site next.
  • The majority of study participants said they expect to be able complete simple transactions like checking their bank balance in a minute or less, or they will abandon the site. [Source: Equation Research]

From Nov. 1-15, the average response time for 14 industry-leading mobile retail sites was 4.73 seconds. Amazon led with a response time of 2.85 seconds. [Source: Mobile Commerce Daily]

Interesting side note: “Generation Y” generally gets slapped with the label of being impatient when it comes to page load times, but did you know that mobile web users over the age of 45 are actually the most impatient of all? The folks at Equation Research tell us this is so.

If none of those numbers convince your powers that be, then your organization may have bigger problems than just the speed of your website.

This cheat sheet is a work in progress. I’ll be re-posting it any time I make significant updates. And if you come across any new performance studies, fire me an email at joshua@webperformancetoday.com.

Related posts:

By: Joshua Bixby Category: Ecommerce, Human factors, Mobile, Research Tags: , , , , , , ,

Google, Site Speed and SEO:
We don’t know much, but we do know something

9 Jun 2010

6 Comments

Ever since Google announced that it now factors site speed into its search ranking algorithms, it’s played its cards somewhat close to its vest on what this actually means. That, of course, hasn’t stopped those of us in the SEO and performance communities from discussing the issue to bits.

I addressed the issue somewhat when Helen Overland at Search Engine People interviewed me a couple of weeks ago. Helen asked me if I could cite any specific examples of a site improving its search ranking after optimizing its performance. At the time, I said this:

Because Google’s addition of site speed to their search algorithm is still relatively new, and because they haven’t divulged the exact details of the algorithm changes, it isn’t possible yet to point to any specific examples of site speed improving search ranking.

And this:

We do know, however, that Google only considers a site fast is if its speed is in the top 20% of its class, and there’s a general assumption that any site that performs at the top of its class will get some kind of boost in the rankings. There’s another assumption that sites that perform in the middle of the class – 20% to 60% – will probably not see any change in their ranking. And sites that perform at the bottom of the class – 60% and lower – will probably be penalized somewhat. So for site owners, this is the time to take a critical look at their site’s speed – alongside their competitors’ – and figure out where they rank and what they need to do next.*

Today, I want to address the first issue:

How one site improved its speed and, as a result, its Googlebot performance

Since talking with Helen, I now have an example of a website that saw a dramatic improvement in how it performed with Google after doing nothing more than speeding up its performance.

In March, we here at Strangeloop implemented a three-month trial of our site optimization service with a client. That trial just ended and we’re now looking at the performance improvement results.

These graphs show Google’s crawl stats for the site over that time.

Crawl stats before and after web performance optimization

Notice that the first and the last graph are almost the inverse of each other. By the end of the Strangeloop trial, Googlebot was crawling about twice as many pages as it was able to at the outset, as Strangeloop halved the amount of time Googlebot needed to download each page.

As you probably already know, Google allocates either a set amount of time, or a set amount of data, for crawling each site. The more pages that Google can crawl within these limitations, the better a site’s ranking will be.

Benchmarking search rankings before and after optimizing site performance

Because we started this test before Google’s announcement, we didn’t think to benchmark this site’s search ranking before implementing Strangeloop. But given this new data, there’s excellent reason to be confident that improving this site’s performance positively affected its ranking.

In the future, we’ll look into benchmarking search ranking before and after implementing Strangeloop so that we can gather more data about how site speed improvements affect search engine rankings.

Have you seen any specific examples of a site that has improved or worsened its Google search ranking after speeding up or slowing down its performance? Let me know. If we can collect and analyze enough meaningful data from these case studies, it would make for a great white paper.

Tomorrow I’ll address my second point: how site owners who care about SEO can take a critical look at their website’s speed, calculate where they rank alongside their competitors, and assess what they need to do next.

*Credit for these assumptions goes to Eric Enge, who postulated these numbers in this post on Search Engine Land.

Related posts:

By: Joshua Bixby Category: Research, SEO, Standards and benchmarks Tags: , , , ,

Optimizing Web Performance:
Why 2010 is the year you should care

7 Jun 2010

0 Comment

The first rule of blogging is that every blog has to have an inaugural post that makes some kind of sweeping, epic statement of intent.

The second rule of blogging is that an audience of about six people will read this inaugural post. (Hi, Mom!)

That said, the sweeping, epic statement of intent is a time-honored tradition, and I’m nothing if not a traditionalist, so here goes:

I believe that 2010 is the year that will usher web performance optimization into mainstream thinking.

In other words, if you run a website that purports to meet any kind of business metric – be it sales, downloads, productivity, or ad clicks – then this is the year you will start to care about how quickly your site performs, if you don’t care already.

Bear in mind, the importance of internet speed isn’t a new issue. Let’s jump into the Wayback Machine and make a quick visit to 1999, when Zona released a report warning ecommerce retailers that they risked losing $4.35 billion per year if they didn’t optimize their websites’ loading times. And what did Zona recommend as an optimal loading time for ecommerce sites? Eight seconds.

So what else do we know about 1999?

  • The average web page contained about 15 objects – images, CSS, Javascript, etc.
  • The average page size came in at about 50k.

Back in the glorious present, what has changed?

  • The average web page contains about 75 objects.
  • The average page size is a bloated 498k.

Sites have gotten more complex. Pages have grown enormously. And don’t even get me started on how the browser wars have transformed the performance landscape. (That’s a post for another day.)

One thing that has remained the same is that far too many sites are still aiming for (and still, sadly, not always achieving) eight seconds as the Holy Grail of load times. The unfortunate thing? As far as web users are concerned, that benchmark has changed dramatically. These days, almost half of a site’s visitors will bounce after waiting just three seconds for a site to load.

Another thing that hasn’t changed since 1999: the cost of having a slow site is potentially enormous. Recent tests by Microsoft’s Bing show that slowing down a site by two seconds led to a 4.3% loss in revenue per visitor, and a survey by Aberdeen revealed that slowing down a site by just one second means a 7% reduction in conversions.

But percentages are abstract. Let’s put this into real numbers. If your site typically earns $100K a day, a one-second slowdown could be losing you $2.5 million this year.

There’s hope, of course. This is the internet, after all – the birthplace of relentless optimism. If you care about web acceleration (and I’m assuming that, if you’re still reading, you do), get ready for exciting times ahead.

And here’s where the third rule of blogging comes in: the valiant call to arms.

I’ll be using this site to discuss the issues and challenges of web performance, as well as to explore various solutions. But I’m not doing this alone. There’s a burgeoning web performance community out there. If you’re a part of it, I’d love to hear from you. Drop me a line in the comments or send me a note: joshua@webperformanceoday.com.

Related posts:

By: Joshua Bixby Category: Research, Standards and benchmarks Tags: , , ,
1 2 3 4 5 6 7 8