September 07, 2005
Mint
Mint is a new, cheap and rather nice looking stats package that runs on PHP and MySQL. I'm still looking for the perfect stats package. I'm not convinced that WebTrends is it. There are lots of choices out there...
Posted by Michael at 09:00 PM | Comments (0)
March 02, 2005
10x10
Yet another intriguing way to present the latest news: 10x10 presents an hourly ranking of news stories in photos and keywords.
Their website describes it best: "Every hour, 10x10 scans the RSS feeds of several leading international news sources, and performs an elaborate process of weighted linguistic analysis on the text contained in their top news stories. After this process, conclusions are automatically drawn about the hour's most important words. The top 100 words are chosen, along with 100 corresponding images, culled from the source news stories. At the end of each day, month, and year, 10x10 looks back through its archives to conclude the top 100 words for the given time period. In this way, a constantly evolving record of our world is formed, based on prominent world events, without any human input."
There are some great notes for developers as well.
Posted by Meghan at 05:56 PM | Comments (2)
February 21, 2005
A case for the database
Databases are abstract, boring concepts to those unfamiliar with them. Companies expect data retention, and the ability to retrieve said data. Such an overly simplistic view is dangerous because it ignores the massive power inherent in the DBMS. There are a few simple reasons to get excited about databases:
1. Databases will allow for complex queries.
A database is far more than a server storing Excel worksheets—it can store relations, or relationships between distinct objects. The basic relational model states that there are entities, or primary objects. These are things like orders, people, books, etc. While tables of entities are certainly useful, they are more so when relations between entities can be expressed and stored in the database.
Good DB design can answer questions like, "Which books belong to Jane?", or "Who ordered the most books this month?" In order to answer the last question, attributes must be assigned to both relationships and entities. For example, people typically have a name, address, social security number, etc. Books often have titles and authors. Now, consider a relationship between people and books called books_owned, which relates a person to the books they own. Suppose we also want to store how long a given person has owned a book—the only place that makes sense to store such information is as an attribute of the relationship. This simple design allows us to track a single book's ownership history across many different people.
Apart from the above toy examples, databases can break down sales figures by region, state, date range, product category, or all of the above in a few simple lines. Extremely complicated reports can be automated, so that human effort is largely spent on interpreting data, rather than constructing the report.
2. Through complex queries, insights may be gained.
There's no point to storing all sorts of data if it cannot be put to use. The flexibility and power of modern database software allows for data mining, or the automatic search of large stores of data for patterns. Data mining is often used in CRM, in order to understand how customers use a web site. Online retailers like Amazon use the technique to identify buying trends, or to suggest books for customers to purchase. Direct marketers can employ data mining techniques to learn how to effectively target those most receptive to a particular pitch, given prior history.
3. Decreasing storage costs and increasing computer speed mean that more data can be gathered and analyzed than ever before.
The data-gathering industry is constantly growing. Several years ago, web statistics were a very simple affair, nothing more than processing log files and determing hit totals. With more advanced tracking and personalization techniques, software can follow a user's path through a web site and determine the precise amount of time spent on each page. Click-through rates can be analyzed, and web sites can be updated constantly to better cater to the end-user. Product usage data, defects, complaints, and all other customer-related communication can be stored for later analysis. Even simple things like store locator searches by ZIP code can yield useful information: by tracking the submitted ZIP codes, one can determine the geographic location of a given user. Correlating regional usage data with marketing campaigns can help to measure the impact of multi-media marketing.
Two caveats (as usual)
It is important to note here that massive data gathering has a serious detraction: consumer privacy. While no one should have any expectation of privacy viewing a company's website, a privacy policy should be clearly available from every single web page. Customers have a right to know what information is being gathered, and with whom it is being shared. In order to continue recording customer behavior, it is essential that we respect an individual's privacy rights by keeping them informed.
In addition, if sensitive information is being collected (such as personal records or financial account details), security is of paramount concern. Unintentional disclosure of private information (by the company or an attacker) can ruin customers' hard-won trust. Whenever a database solution is being deployed, security issues must be considered and addressed.
Conclusion
Data collection is an increasingly important aspect of interactive development. Like many other pursuits, proper marketing is a battle for information: whoever has the best information first and uses it wins. Technology allows us to exceed past expectations about customer tracking—it's time to take advantage.
Posted by Matt at 10:57 PM | Comments (0)
February 15, 2005
Looking at Data Differently
This site uses a distinctive (read: "frickin' cool") visual heirarchy to accentuate relationships/common biases in up-to-the-minute news feeds.
And this site visually maps photos posted on Flickr to highlight the sense of location in users' images.
What is exciting about these cases is that they are such fresh and well-paired marriages of design and content. Most sites have rich stores of information, whether submitted by users or of product info/inventory, etc. But these examples underscore how well-thought-out design can add value and even alter perceptions of that information.
(Yeesh, I think I'm waxing Tufte)
Posted by Andy at 11:05 AM | Comments (4)