Twitter’s Crawl

A company can have the best product around, but if the pages are too sluggish, if the product suffers recurring outages, if the user-product interaction is varied and inconsistent, the product’s overall Usability can, and does, suffer.

Quick-UX provides for the rapid, simple and quantifiable assessment of a product’s User Experience (UX). Among the various components that define a product’s Usability, as well as Quick-UX‘s, are Accessibility, Consistency, Recognition, Navigation, and Page Load Time.

In answering the question of Usability, "Can I use it?" the sub-category of Page Load plays an instrumental role. Page Load, often obfuscated or connected with other perceived causes of a product’s dissatisfaction, ultimately, either positively or negatively, presents an unquestionable influence on a product’s overall Usability.

Example: Poor Load Time (value = 0.0)

Twitter is fast becoming, and for some already is, an essential communication tool.

Yet, Twitter earns a Page Load Time variable value of 0, due to its intermittent slow performance, but more so contributing to this value are the constant outages felt through the year, month after month.

If the page doesn’t load, if requested action takes an interminable amount of time, if the likelihood of the next user action failing is constantly looming, the overall Usability of a product takes a terrible toll.

In 2009, according to Pingdom, Twitter experienced a total of 20.82 hours of downtime.

Outages of Twitter were not isolated to merely the entire site being unavailable, but also consisted of sub-sections, or sub-features not working or resulting undesirable or unexpected behavior. Contributing to the pervasive problem of Page Load Time is both the inaccessibility of the product as well as the inability of the users to obtain key information (missing updates, etc) and other bugs leading to incomplete or otherwise incorrect Page Loads.

A Quick Study

I quickly examined and compiled a list of incidents that affected the Page Load Time of the Twitter product, distinguishing between total downtime, and partial downtime and information inaccessibility, based upon the public posts on Twitters blog.

http://status.twitter.com/archive

I did my best to not double count any problems, but it was difficult since many of the problems occur so frequently, and it is often difficult to distinguish, from these status blog posts alone, between a persisting problem being experienced or fixed, from that of a new emergence of a similar or same problem. Furthermore, I also excluded the impact on Page Load Time arising from scheduled maintenance/downtime – periods of time over which the user expectation would be most aligned with the product’s promise of Page Load Time.

Some of my notes regarding my review of Twitter’s 2009 product Page Load Issues…

That said, a clear picture of the Page Load Time experience felt by the Twitter product’s user base quickly emerged.

Approximately 14% of all days in the year experienced delays and disruptions, directly altering the Page Load Time of the product. And, another ~10% of the year’s days experience pages loading with missing information, resulting in a total number of days experiencing disruption at around 24% of the year or 86 days! (note: there may be some day overlap that is not taken into account in these numbers)

Note: Data for December is complete (only goes through December 21, 2009)

Should Do & A Clear Flight Path

When using Twitter, tweets, responses, searches can and sometimes do occur quickly and without incident. However, with such consistency of problematic service, fail whales, site latency, etc. Twitter earns no more than a value of 0 for Page Load Time; but with a clear path to improvement…

• first, focus on the reliability of the Page Load, drastically reducing downtime,
• then, focus on the missing data and other inconveniences, some of which are touched upon in my table of notes above.

Next…

Over the next several weeks I will be providing real-world examples of Page Load Time values…

Poor Load Time (value 0) [Twitter, Twine]
Delayed Load Time (value 0.5) [Conversation Pieces]
Prompt Load Time (value 1) [Facebook]

Subscribe now (click here) to make sure you don’t miss any part of this series exploring the Usability and Page Load Time of Quick-UX, the quick and easy method of generating quantifiable and comparable metrics representing the understanding of the overall User Experience of a product, as well as other insightful posts from The Product Guy.

Enjoy, Discuss & Tweet!

Jeremy Horn
The Product Guy

Add to Social Bookmarks: