Jakob Nielsen's
Alertbox for November 1, 1998:
Why Yahoo is Good (But May Get Worse)
Yahoo does many things right:
- Yahoo's home page has an average download time of 3.0 seconds
according to the Keynote performance index. This is one of the fastest
download times among major websites, even though it is still three times
slower than required for optimal hypertext usability.
- The page design is a model of minimalist design: plain and simple HTML
that will render correctly in even the oldest browser.
- Based on hypertext with links, links, and more links
everywhere you turn.
- Leveraging the connected nature of the Web and the many other sites
rather than trying to do everything themselves. Yahoo's city guides are a
prime example: they link to the best local content in each
city - richer than anything that could be built from scratch.
- Highly structured navigation system that emphasizes
context and a hierarchical information architecture (not that hierarchies
are the only possible information architecture, but they are easy
to understand and represent in the interface).
- Integration between search and navigation: when you
search, the hits are shown relative to their location in the topic hierarchy.
Some design critics say that Yahoo is boring, but the simplicity has its
own elegance, even if it doesn't win design awards. In fact, design awards
are rarely given to those designs that work best in real use.
The real reason Yahoo is so successful is that it embraces the new
medium and designs for its strengths rather than fighting its
weaknesses. So, since Internet bandwidth is very limited, Yahoo emphasizes
a slim design and forgets about emulating television or glossy magazines.
Instead, they base their primary service on giving people good links: the
fundamental innovation of the Web (and earlier hypertext systems) relative
to other media. Web managers often discuss whether to allow links to other
websites, with some claiming that outbound links involve a risk of losing
users who move to other sites instead of dutifully staying where they are.
One counter-argument is that users are going to leave anyway, but that
they will return if they are given good service in the form of valuable
links. Another counter-argument is that the most popular site on the Web
built its service on nothing but outbound links.
Problems Ahead for Yahoo
- As more services get added to Yahoo's home page, the page risks being
confusing and overwhelming. It's not bad yet, but beware.
- Complaints about missing listings are legion: lots of good sites are
not listed on Yahoo. The manual classification that was Yahoo's original
strength seems to have difficulty scaling to match the ever-faster growth
of the Web.
- Yahoo's search technology is very primitive:
- The search hitlist used to work but the design is insufficient to cope
with the increasing size of the database and the associated growth in
search hits
- Mistype a single character, and you won't find anything
- Unless you think of the exact keywords used to describe a site, you
won't find it: there seems to be no use of the database structure to form a
semantic model of the term space and go beyond simple keyword matching
- The integration between the
search of the primary Yahoo directory and the back-up search engine is
very poor
- There are no quality ratings to guide users to the good sites first:
many of the categories in Yahoo have hundreds of listings, so how are
people to know where to start?
- The topic hierarchy gets harder to navigate as it gets
larger. Even though cross-reference links are used to good effect to
alleviate this problem, usability will ultimately suffer unless better
classification methods are discovered.
- Statements from Yahoo executives indicate that Yahoo refuses to
think of itself as being primarily a search-and-navigation aide to
help users go elsewhere as fast as possible. Moving
beyond your roots is always difficult, but downplaying the exact reason
users come to your site seems outright reckless.
What You Can Learn From Yahoo
People from many companies are attempting to build
their own "mini-Yahoo"
as an entry-point to their intranet or other large
collection of Web information. This is a good idea, and improved
intranet navigation can save millions of dollars in most large companies.
In building a "mini-Yahoo", remember the strengths I listed above.
Emphasize speed, structure, clean navigation, and an integration between
search and the topic hierarchy. Also make sure to build a classification
hierarchy that matches your users' view of the information space rather
than your own internal model.
Avoid Yahoo's mistakes: allocate editorial resources to maintain the site
and keep it meticulously up-to-date, buy the best search you can find, and
make sure to emphasize high-quality or recommended information. Finally,
do the research to discover users' goals and design the service to support
these goals and not the things you think users ought to want.
You can also use Yahoo's statistics to
calculate the ranking of your own site relative to the rest of the
Internet.
Advertising Doesn't Work on the Web
For years, I have argued that
advertising doesn't work on the Web
and that it cannot serve as a business model for anybody except the very
largest sites.
Since Yahoo is the number-one site, it is among the few that can
survive on advertising, and Yahoo does indeed turn a profit.
But Yahoo's financial reports show that Web advertising is a very
poor revenue generator:
- For the quarter ending September 1998 Yahoo
had revenues of $53.6 million, but since $5.2M was interest earnings on
money harvested from selling high-priced stock, the true revenues from the
Web service itself was $48.4 million. During the quarter I estimate that Yahoo served
about 12.4 billion page views (estimated based on their
announced traffic of 115 million pages per day in June and 144 million
pages per day in September).
- Dividing the two numbers shows revenues of 0.4 cents per page
view.
- Less than half a cent per page view from an advertising-based service is
fine if the service has as much traffic as Yahoo. But most other sites
will not be able to survive on such low income.
I predict that clickthrough rates will be cut in half every year.
Across the Web, the average clickthrough rate was
about one percent last year and about half a percent this year. My
prediction is that clickthroughs will drop to 0.25 percent in 1999 and
approximate 0.1 percent by 2001. Even so, Yahoo's traffic will
grow faster than CPM drops, so they will probably remain in the black. It's
the smaller sites I worry about: they desperately need micropayments and other revenue
streams.
Previous: October 18, 1998: Failure of Corporate Websites
Next: November 15, 1998: 2D is better than 3D
See Also:
List of other Alertbox columns