In this post, a closer look at a memorable and alarming statistic
leads to a more general discussion of things that annoy website
customers, and a bit about the questionnaire versus user trials
controversy...
Sometimes I see a statistic that I know I'm going to see quoted a lot.
An example is one from eMarketer in an article called Online Retailers
Face Four-Second Barrier
Now, as the all-important online holiday shopping season nears,
and the competition for online shoppers increases, a new report from
Akamai Technologies advises e-tailers to be on their toes — because a
few heartbeats can make the difference between a sale and a lost
customer.
The research shows that four seconds is the maximum length of time
an average online shopper will wait for a Web page to load before
abandoning one retail site and moving on to another.
"The critical takeaway from this research is that online shoppers
not only demand quality site performance, they expect it," said Brad
Rinklin of Akamai. "Four seconds is the new benchmark by which a
retail site will be judged, which leaves little room for error for
retailers to maintain a loyal online customer base."
The article is quoting a new report by Akami and Jupiter Research
(registration is needed to view the report). It looks like a good
enough piece of work the methodology section of the paper explains:
"In April 2006, JupiterResearch designed and fielded a survey to
online consumers selected randomly from the Ipsos US online consumer
panel. A total of 1,058 individuals responded to the survey.
Respondents were asked approximately 15 closed-ended questions about
their behaviors, attitudes, and preferences as they relate to buying
and researching products and services online. Respondents received an
e-mail invitation to participate in the survey with an attached URL
linked to the Web-based survey form. The samples were carefully
balanced by a series of demographic and behavioral characteristics to
ensure that they were representative of the online population.
Demographic weighting variables included age, gender, household
income, household education, household type, region, market size,
race, and Hispanic ethnicity. Additionally, JupiterResearch took the
unconventional step of weighting the data by AOL usage, online tenure,
and connection speed (broadband versus dial-up), three key
determinants of online behavior. Balancing quotas are derived from
JupiterResearch's Internet Population Model, which relies on US Census
Bureau data and a rich foundation of primary consumer survey research
to determine the size, demographics, and ethnographics of the US
online population. The survey data are fully applicable to the US
online population within a confidence interval of plus or minus three
percent."
When I hear the "new 4-second rule" being quoted at me, though I will
try to remember the following bits from the actual report:
1. "Thirty-three percent of consumers shopping via a broadband
connection will wait no more than four seconds for a Web page to
render." (From Fig 3 of the report:30% of broadband customers said
they would would wait 5-6 seconds, and a further 38% would wait more
than 6 seconds. Only 19% of dial-up customers said they would leave if
a single page took longer than 4 seconds to render).
2. From Fig 2 of the report, in which respondents were asked to
remember what annoyed them about their last unsatisfactory e-tail
experience, slow pages was the third biggest annoyance (quoted by 33%)
compared with high prices (44%) and shipping/handling issues (39%).
Annoyance at the need to register, or "website was
frustrating/confusing" came joint fourth.
3. In Fig 5, customers are asked how they would react to their bad
experience, and this was correlated with what they complained about -
interestingly, there isn't much difference in the reaction of people
who remember a crash from those who found the site too slow or
confusing (or who found the checkout process too slow). The most
likely reaction was to visit the site less often, "purchase from
another online retailer" comes third. I'd have thought that a crash
was worse than slow pages, but maybe not.
Slow-loading pages is of course not a new "annoyance" the eMarketer
article quotes a 2005 study (Taylor Nelson Sofres) in which slow pages
came out as the 5th most popular thing to be ranked by customers as
"extremely annoying" (the top 4 were pop-up ads (84%); requiring
installation of extra software to view the site (72%); Dead links
(66%) and requirement to register before viewing site (61%). I can't
think of anyone who says slow pages are good per se. As usual there's
a trade-off some sites continue to have slow pages because they reckon
the page will work so well when you do see it.
So while having slow pages is not good, it is not quite the stark 4
seconds and you are dead! But I'm going to hear this figure a lot
because while it may not be completely right it is very memorable
enough for a sort of e-marketer's 1066 And All That. At best I guess
it is a rule of a thumb that has been dipped in a pinch of salt.
Actually what seems more realistic to me is an idea from Steve Krug's
book "Don't Make me Think" - imagine the customer arrives at your
website with a reservoir of patience and good will to you and your
business. The initial size of this reservoir depends on a lot of
factors, including their previous experiences with you, their overall
mood and to-do list that day, how motivated they are to do stuff on
your site and so on. Their "good will level" goes up or down according
to how the site treats them - but the point at which goodwill=0 and
they say "aw the heck with it!" is going to vary.
This is probably the point at least to mention that old grip that some
usability study folks have with questionnaire data - that
"self-reported claims are unreliable" (sometimes this is said to be
"Nielsen's first rule"). That is - what people say they thought or
happened is not always how it seems if you watch them. This is one of
the interesting objections raised on a discussion at cre8asite about
whether it was worth putting questionnaires on a website. The "don't
bother" camp seemed to win that debate on that site, a bit to my
surprise - I'd have thought that what customers perceive as reality is
as interesting as what someone else perceives their reality is....
