Sign up for weekly recaps of the ever-changing search marketing landscape.
Defining Download Time: Web Searchers Vs. Search Engines
“Download time” and “speed” are terms that often confuse SEO professionals, web designers/developers, information architects and usability professionals alike because of various interpretations. A number of items influence download time, including but not limited to:
- Server response time
- File size(s)
So when Google (or any search engine) makes an announcement about speed counting as a ranking factor, I can’t help but wonder what their interpretation of “speed” is. The reason for my concern? I do not agree with Google’s decision. Personally, I wonder if the decision is based on the mental models of Google software engineers… or those of actual searchers.
Fast download time is good… and so is oxygen
At first glance, it might seem as if I believe that a slow download time is good, which is not true at all. The truth is that I have been touting fast download time years before Google came into existence, as part of my 5 Rules of Web Design (first written in 1997) which was included in all editions of my first book, Search Engine Visibility. Rest assured, I do support a reasonably fast web-page delivery.
Yet when I see how search engine representatives spin the decision about download time, I often feel as if they are merely stating the obvious (such as “I support breathing oxygen”) instead of looking at the big picture.
For example, speed of page delivery is incredibly important from a web search engine perspective. What do I mean by page delivery on a web search engine? How quickly search engine results pages (SERPs) are delivered. A few milliseconds delay in SERP delivery does affect how web searchers feel about the search engine, and it does affect searcher behaviors.
But let’s be realistic—how many website owners run a commercial web search engine? An ecommerce site is not a web search engine. A B2B (business-to-business) site is not a web search engine. A search engine result page is not a category page or a product page or reference page or an FAQs page. What is applicable to a web search engine might not be applicable to a different type of site.
I remember at the SMX West 2010 conference, panelist Maile Ohye, Senior Developer Programs Engineer at Google, mentioned that Google does not look at how a page renders when considering a page’s download time. I found that comment to be very interesting. Because, as a web designer/developer and usability professional, I absolutely consider how my pages will render when I create them because it affects the user experience. Users/searchers expect information scent to be validated with every click.
Actual vs. perceived download time
What is more important to users/searchers: actual download time or perceived download time? A web page’s actual download time is the time it takes for a full web page (including graphic images, style sheets, scripts, etc.) to be delivered to users. Connectivity, server response time, and file size(s) will certainly affect actual download time. Perceived download time is the users’ perception of page delivery. In other words, is desired content being delivered quickly or slowly? What factors affect perceived download time?
Interestingly enough, back in 2001, usability professionals Christine Perfetti and Lori Landesman from User Interface Engineering published an article about this very topic. In The Truth About Download Time, they found that users felt that web pages downloaded quickly when the scent of information was validated. When the scent of information was not validated? Users perceived download time as slow. So when people say that speed is important, do they really mean that a web page should download within 10 seconds? Or do they mean that the website did not match their mental models?
What do users really mean when they say a web site is too slow? Jared Spool did a follow-up podcast on this subject at: Usability Tools Podcast: The Truth About Page Download Time.
Users/searchers are more concerned with their mental models and information scent being validated than actual download time. They want to see the words they type into search box appear in search results, and they want to see those same words on the corresponding landing page. On an ecommerce site, they want to see the button labeled “add to cart.” They want to see a high-quality product photo so they know what they are purchasing. And they are willing to wait for that high-quality photo to download if they see the photo rendering before its full presentation. My client data validates Spool’s findings.
So what are search engines emphasizing: server response time, the validation of information scent, or both?
When download time affects quality
Website owners have some control over some things, and no control over others. For example, it is a safe conclusion to say that most website owners do not have control over connectivity. Many website owners do not have control over web servers as many of them are in a shared server environment.
In the photography, entertainment and architecture (not IA) industries, larger file sizes for images and multimedia documents are the norm. Users who visit these types of websites expect a longer download time. Will these types of websites receive less qualified Google traffic at the expense of a faster server response? Should an entertainment site that has been usability tested and well-architected receive less search engine visibility and less search engine traffic because its page file size is 21K instead of 20K?
There are numerous examples of sites where download time has nothing to do with relevance. So I am not sure why Google made this decision. I certainly agree that download time should be minimized, but never at the expense of quality. In other words, I believe that, “oxygen is good.” But I am not sure if the powers-that-be at Google understand what users/searchers really want outside of a search engine results page.
I am interested in others’ opinions. What do you think about Google’s recent “speed” decision?
Some opinions expressed in this article may be those of a guest author and not necessarily Search Engine Land. Staff authors are listed here.