AIAIO: Our Blog

Many sites have been doing some form of personalization for years. This often includes relatively “wide” customer segments like new vs repeat visitors, and relatively static discount or product promotions. Now imagine doing promotions that are customized with every page view, to the specifics of each individual visitor. That’s dynamic personalization. It’s the next killer app in ecommerce.

Dynamic Personalization

In a blog post a few weeks back, I wrote about dynamic personalization as one of the top strategies for next-generation ecommerce. Dynamic personalization is a hot topic at Ai, and our clients are leveraging it with great results.

Many sites have been doing some form of personalization for years. This often includes relatively “wide” customer segments like new vs repeat visitors, and relatively static discount or product promotions. Now imagine doing promotions that are customized with every page view, to the specifics of each individual visitor. That’s dynamic personalization.

What It Is

Have you ever been on Amazon and seen “customers who bought X also bought Y” after you added an item to your shopping cart?

That’s dynamic personalization. Specific product (or content) recommendations based on your unique behaviors. These behaviors include what you have browsed and searched for, the products you’ve viewed, your ratings and reviews (if any), and your purchase history.

How it Works

All these behaviors are compared in real-time with the behaviors of other shoppers.  The result is a match between things you’ve implicitly expressed an interest in, and the interests others have expressed. This type of dynamic personalization uses collaborative filtering, a fancy way of saying that it uses the wisdom of the crowd, and represents the most common method by which dynamic personalization works.

The D in Dynamic

As customers continue to visit a web site, browse and search, view products, and purchase, this data is fed into the personalization database. As the site owner adds products, content, and promotion, this information is fed in as well. In this way, the dynamic personalization system continues to evolve as it learns more about what is popular and relevant right now.

If a hot new toy is released, this product will come to the fore in recommendations as a result of customer interest. As it fades in popularity, it will be replaced by something more popular.

This adaptive learning and re-adjustment happens thousands of times per day without any explicit setup or maintenance. Additionally, the type of recommendation presented can be dynamically customized to the type of product displayed. Quantity discounts might be appropriate for cleaning supplies, but not for dvd players. Feature-based upsells and accessory cross-sells might work for a TV, but not for a book.

Why It Matters

A very small percentage of what you have available on your site is relevant to any one customer at any one time. Getting users to view what is relevant to them–making them aware that you have what they need, that they are in the right place and (by the way) there are other options they may not have thought of that are also relevant to them–is a sure way to greater sales and customer retention. Traditional methods of updating and optimizing your site through analytics and testing are great, but rely to a certain extent on guesswork and can be too slow to react to a trend.

Dynamic personalization relies on what your customers are actually telling you about what they prefer. If you don’t have the time to update your site thousands of times per day and provide live customer service chat to every customer, you should probably look into it.

Over the last six weeks, Ai has had a team participating in a Team Trivia quiz league run by the New York City Social Sports Club. You may remember a previous post from the beginning of the season, but all that has come to an end after last night’s playoffs. The outcome? Ai won 2nd place overall in the league.

The season started off slow, with team Ai holding rank at 17th place out of 24 teams. As the weeks went by, we gradually made our way up the ladder to 13th place, lingering at 11th place, and ending up in 9th place before the finals. At each quiz night we would generally rank at about 5th or 6th place, however the spread of points between teams was very thin and most times there were still eight or more teams ahead of us tying for 2nd, 3rd, or 4th places. There are also sportsmanship points that count against each team member not wearing their league shirt. We always had our shirts.

All 24 teams attended the playoffs, but only the top 12 teams were competing for the grand prizes. Being in 9th place, Team Ai were in the running for the grand prizes. The quiz itself was as normal as any of the other weeks, but the cumulative scores from throughout the season didn’t matter for the playoffs. The top-scoring teams were the ones to take home the gold and silver, and I am proud to say that Ai came out in 2nd place!

One team received 1st place, and Team Ai was tied for 2nd place with another team. This meant a tie-breaker question given to one representative from each team, of which the first person to shout out the answer won. The question? Name the largest land-dwelling bird in the United States. Feel free to take a guess at that in the comments.

The 2nd place prizes included a certificate stating our placing, a $25 gift certificate to the restaurant/bar where the quizzes were held, and three free pitchers of beer. We also each received 2nd place medals. Congratulations to my fellow teammates: Benaz Hossain, Robert Gurdian, Josh Rusch, Isaiah Belle, and Richard Davy. Big thanks to the few substitutes we had fill in on a few nights: Katie Mericle, Marlen Bernardez, Mike LeDoux, Dave Knapp, and Anita Belle.

You are more than welcome to challenge Team Ai to a game of trivia. We will gladly oblige.

Magento is great, but it needs a good amount of hardware behind it.  Developing locally can get slow and cumbersome unless your environment is tweaked properly.  Here are a few tips for boosting Magento performance without impacting the rest of your development environment.  Please keep in mind that the memory allocations work well for my machine (dual core, 4 gigs of ram).

Database
Install innoDB.  Magento can use the in-memory buffer pool to cache table indexes and data.

Configure my.ini:
innodb_buffer_pool_size  = 64M
innodb_thread_concurrency = 4 (or 8 if you have dual core)
query_cache_size = 64M
query_cache_limit  = 2M

apache
enable mod_expires in httpd.conf

php
in php.ini enable:
realpath_cache_size = 16k
realpath_cache_ttl = 120

Install the eAccelerator binaries for php.  APC is a better solution but is less compatible with windows.  If you need to compile these, click here for instructions. Then configure it:
extension=eaccelerator.dll
eaccelerator.shm_size=64
eaccelerator.cache_dir=C:\PHP\tmp
eaccelerator.enable=1
eaccelerator.optimizer=1
eaccelerator.check_mtime=1
eaccelerator.shm_max=0

Install memcached.
add the following lines inside the config of epp/etc/local.xml
<cache>
    <backend>memcached</backend>
    <memcached>
        <servers>
            <server>
                <host><![CDATA[localhost]]></host>
                <port><![CDATA[11211]]></port>
                <persistent><![CDATA[1]]></persistent>
            </server>
        </servers>
        <compression><![CDATA[0]]></compression>
        <cache_dir><![CDATA[]]></cache_dir>
        <hashed_directory_level><![CDATA[]]></hashed_directory_level>
        <hashed_directory_umask><![CDATA[]]></hashed_directory_umask>
        <file_name_prefix><![CDATA[]]></file_name_prefix>
    </memcached>
</cache>


Admin Backend

• Keep the indexes up to date (System > index management)
• Compile Mage classes (System > tools > Complilation)
• Enable all cachine (System > Cache Management)

I called US Airways the night before a recent business trip to ask about a travel detail I couldn’t find online. (I’m not name-checking US Airways just to pick on them; it’s part of the story.) Their customer service is obviously outsourced to an overseas location–I had to call twice, and both representatives had trouble speaking clearly and understanding my question.

But this isn’t about offshoring, or customer service reps whose native tongue isn’t English, which doesn’t offend me. (I certainly couldn’t administer tech help in Hindi.) Rather, it’s about training.

Upon completing my second call, the US Airways CS rep said to me, “Can I help you with anything else today?”

“No, that’s it,” I replied.

“Thank you,” she continued, “for calling Use Airways.”

Use Airways. I headed to the airport the next morning still shaking my head about the woman who doesn’t know her employer’s name. Shortly after taking my seat on the plane, a flight attendant got on the PA system.

“All electronic devices must be turned off at this time,” he said. “If you do not turn them off and put them away, we will return to the gate and deplane you, and you will have to rebook on a later flight.” (Emphasis his.)

My seatmates chuckled at his earnestness, but I just thought about my phone call. In the span of a few hours, I encountered two different but striking examples of poor training and comprehension by consumer-facing employees.

My trips on US Airways have largely been pleasant and comfortable. But what is the brand impact of these employees’ mistakes? How many other people notice what I notice, and book their next flight on another carrier?

Airline flight attendants routinely say, “We know you have a choice.” What they–and their management team–need to say is, “We know you notice. And we’re trying our hardest.”

Tired: “the elevator pitch.” Also: escalator pitch, Reader’s Digest version, treatment, etc.

Wired: the Twitter pitch. Call it the SMS pitch if you prefer. It’s the new “25 words or less”–give me the summary in the length of a tweet.

Modern, timely, fun. Plus the first ten times you use it you’ll get to explain the term to your audience and lighten the mood of the room.

Welcome to Ai 2010!

The new alexanderinteractive.com website is a contemporary new home and showcase for Ai. With simple navigation, big images, proud data points and loads of case studies, Ai’s new site matches the standard of excellence we bring to our client projects.

This site has been a long time coming. As the cobbler’s children have no shoes, so too did we have a patchy old website. Dating to 2006, our company site had received one light refresh in four years, despite several attempts to do more.

We actually got a lot of compliments on our site—as befits our sensibilities, the old Ai was user-friendly, fairly intuitive and full of friendly plain-speak. But we needed something that matched the style and utility of our more modern work. Thus the new Ai, with its wide pages, new copy, and clean, inviting layouts.

Feel free to explore everything the new Ai site has to offer. We’re proud of our new home. C’mon in.

Alexander Interactive, having recently relaunched the Internet Retailer website, is pleased to be attending the Internet Retailer Conference & Exhibition 2010 in Chicago. The conference is shaping up to be a great one and we’re excited to be there.

You’ll see Ai on the expo floor in booth 356, where we’ll be giving live, on-the-spot critiques of attendees’ websites. We will be around the conference all week, and around everyone’s necks, too–look for Ai on the IRCE lanyards.

And, of course, visit the Ai-designed internetretailer.com for conference news and updates.