After
several months of accumulating daily search statistics (for one
day a week), I finally found time to
summarize the results. I maintained the general approach described in my
last examination (covering four days from September 15 through October
13), but, responding to the data, made some modifications
to illustrate the patterns I observed.
Description of the groupings
In subsequent
searches (for 24 hour periods once each week after the original data), a number of
new terms appeared that were not seen in the original
data. In addition, variations on spellings added more terms to be
considered. After regrouping the data, I wound up with 16 categories:
Afghanistan, Airline, American Flag, Automotive, Health, Holidays,
Internet, Maps, Miscellaneous, Music, News, Nostradamus, Osama bin
Laden, Pictures, Shopping, and World Trade Center.
This kind of detail
grouping allowed me to see some interesting patterns in the data, but
proved to be too many to summarize conveniently. Accordingly, I
developed an additional level of categorization.
Patterns in the data
What
I was primarily interested in was to identify the impact that the events
of September 11 had on the search activity on the Internet. The terms
included in Afghanistan, AmericanFlag, News,
Nostradamus, Osama bin Laden, and World Trade Center seemed to me to have some relationship to
one another, so I grouped them together into an "Event"
category.
I
also felt that I noticed a pattern in the "routine" Internet
traffic that focused on
things
related to interactive computers or internet functions,
people
looking for items to purchase, and
music
and games.
I
set these up as Internet, Shopping, and Music. All other
"routine" traffic, including terms that I could not identify
with any meaningful topic, I added together into Other.
One
of the most striking aspects of the numbers was the peak
in activity immediately following September 11. In the chart on the
left, the dark region on the top layer of the graph represents Event
oriented searches. As the chart shows, they dominated the top
search category on September 15.
Another view of the
data illustrates the relationship between the number of searches in
different categories.
Showing
the "top 300" as a percentage of the total
searches indexed provides another glimpse of the degree to which attention
was focused on a few topics. (The date range on the graph includes
from August 25 through December 1, although the data for
searches within 24 hours did not become available until September 15.)
What did I notice
about the patterns? There is a lot to notice, and more study reveals
more interesting facts. As I mentioned above, I was struck by the
degree to which the top 300 searches focused on the events of
September 11. Remember that this summary does not begin until the 15th.
What would the chart look like if I had the results for the four days
prior to that?
Another
item of interest was the number of searches that pertained to Holidays.
During this period, I encountered many searches that included the words
"Thanksgiving" or "Christmas". (I also
found "Day of the Dead", or "dia de los muertes".)
Many of these searches pertained to food or recipes, so I consolidated
them into the "Other" category. I could as well have
counted them in "Shopping".
Many news
stories about the event mention ways that people turned to the
internet for information and communication. This analysis of search data
completely ignores the volume of traffic traveling across the net in e-mail
messages. People around the entire world were resorting to their
computers to get information from and to areas where the phone
circuits were totally overloaded.
Another
interesting point is the increase in event related traffic around
October 13. This was about the time that stories of the anthrax
incident was becoming news.
I also looked at
the total volume of searches over this time period.
I was a little surprised to find that the total number of
searches increased over the period. I could not identify a
corresponding pattern in the numbers of the "top 300"
searches, leading me to conclude that the range of topics expanded
to include more terms that did not appear within the top ranked terms.
The graph does tend to exaggerate the extent of the
change, which was never greater than a 1% change in either
direction, except for October 20 and December 1, which
were larger than average changes.
Conclusion
The
trends noted in the second analysis continue to be evident in the
November data. The new information that appeared was the attention
turned toward specific information related to holidays. I
was surprised at the extent of the activity containing references
to holidays.
This
would appear to me to have substantial implications for retailers
marketing over the Internet. This is no different from print, radio
or television marketing, where we see and hear seasonal
messages to draw our attention. Internet marketing is different in
the specifics of how one draws attention to the marketing
message.
The
Internet reflects all the interests and concerns of the population. It
shows people actively seeking information about things that are
of interest to them in their daily lives, and those points of focus
change continuously. This emphasizes the importance of having dynamic
content on a web site. An effective marketing site is not a static
construction. It must changeconstantly to reflect
the current interests of the prospects for the business.
Call or write us now to arrange for a free initial consultation
(by phone or e-mail) to identify the ways that we can help you become a more modern, efficient and competitive business.
FreeFind