May 09

At my goal was to create a demo of a user relevant dynamic shopping experience by pulling in as much user info as possible, then correlating that with other more predictable trends (i.e. weather, similar internal customers,  …etc). This info would include previous purchses with respect to seasonal factors (i.e. Christmas cards during December should not be keyed off for July purchases). This analysis would result in an array of user relevant products of varying conversion probabilities. Entering these conversion rates, in conjunction to the expected profit margin, quantity and need for customer retention, into a multi-variate formula, then into solver would provide the optimal mix of product offerings. I glossed over the fun statistics of correlation and of fitting formulas to observed data points, as I was unable to find any free oob software that rivaled the Excel ad ins (although Google has some great graphical tools).

I decided to search out Pearl modules that could gather user client info and gather some site usage, so I ended up with these two programs:

Both are nice, but not too interesting. I found some PHP that allowed me to match each clients IP address with a known DB. At this point I decided to keep the DB performant by only running at the county level, (only 60k entries) this also has the added advantage of allowing localization, so withPHP the user will see a localized greeting (only for a few languages defined in a bit). So now I do a bit of hand waving and say that there is some correlation between consumers from some county and some products, and now an input of US equals a request for cowboy boots, more specifically here is the API input:


if ($CountryCode == “US”)
echo “Cowboy Boots “;

elseif ($CountryCode == “FR”)
echo “Wine “;

elseif ($CountryCode == “CN”)
echo “Snowshoes “;

elseif ($CountryCode == “NL”)
echo “Wool Socks “;

elseif ($CountryCode == “DE”)
echo “BMW “;

echo “Candy “;


I settled upon a Google API, then tried to use a Flikr APIwith Perl, but the hosting environment at Godaddy was limiting (no ability to install new Perl modules), so I turned to Ruby, the landed upon this great API, which is quite robust.

How does one test this site? Here are a few international proxies that will allow you to view the user relevant custom header images, greetings and product evaluation tools of this site:





List of alternates:

Comments are closed.

preload preload preload