Web 2.0 Expo: Akin Arikan on Measuring Web 2.0

It's 10:41, and we are back in room 2002 waiting for Akin Arikan, of Unica Corporation, to start his talk, "Top 5 Do's and Don'ts for Measuring Web 2.0." 

(It's the last day of the conference, and I am very sad to go.  The people here are great, the sessions have been awesome, the companies are extremely cool, and I love being back with people who consider jeans 'business casual' as long as you are wearing shoes. Oh, and the weather has been a steady 55-60 degrees with sun and a few clouds.)

What can you use analytics for?

The obvious answers are usability and conversion...but there's more!

How about:

• ENGAGEMENT!
• Market insight (social intelligence).  Car makers have been doing this for years through their car configuration tools. Think they're just trying to get you into the dealership?  No way! They can see what accessories are clustered together and use this info to design special packages and brochures.
• Relationship marketing. This includes using the firehose of data--all that jumbled up information--to create actionable profiles, and to be extremely relevant.  (This sounds like the hard part to me.) This is where the marketer becomes more like an effective salesperson: by LISTENING.

(wow..he goes fast!)

Case Study: Product review and participation site

The fully-feature, Web 2.0 camera shopping site is loaded with interactive shopping and review features:

• Review the camera
• Rotate a 3d camera
• Click hotspots for details on camera features
• Read reviews
• Create reviews
• Reply to reviews
• Ask a question
• Tell us how important a feature is to you
• If a feature is important, we can ask you specifically how you rate that feature, rather than just the product as a whole

Start with a goal

Right at the beginning, we need to think about measurement.  What are we trying to do, and what are the indicators that we can measure to show us whether we can been successful?  Web 2.0 is harder to figure out than Web 1.0.  Two key changes: don't think of measurement in terms of pageviews, and don't use your server log files.  These traditional approaches are inadequate for capturing the firehose of information that are available to you with Web 2.0 site features.

Some things have not changed

I'll say it again: identify your goals, to derive your key performance indicators.  If you don't , you just get a lot of data in your face and you don't know what to do with it.  It will be harder to sort things out and capture the data you need if you wait until later.

In this case, there are three main goals:

• Drive traffic. Key indicators for this include unique visitors and engagement indicators like session length, comments, and interaction with product viewing features like zoom and rotate.
• Drive revenue.  Obviously, you'll want to see if various site areas and features result in an increase in your revenue and conversion rates.
• Build the brand.

How do you get past page views?

It's great to say don't look at pageviews, but how do you capture richer data?  To do this, you need to 'tag' the various interaction and decision points on your site. For Flash this could be an ActionScript tag, in AJAX, DHTML, and blogs this would be JavaScript (such as Urchin), and in postings on other's blogs you can insert a pixel tag to ping back to you.

I'm not a Flash person, so I won't go into that.  However, I've been using the Urchin script from Google Analytics, and have found it to be very robust. By default, it reports the usual visitors, visits, and pageviews. However, there is a small piece of code that you can put in the onclick event of a link to record outgoing links, and GA lets you set conversion goals with dollar values attached to them.  I have not attempted this yet, but I suspect that the same little piece of code could be applied to other events, such as the onfocus of a form field or other events within your own JavaScript.

With the pixel tag, this simply mean putting a single pixel, clear gif image somewhere within the body of what you are posting. So if you guest author on someone's site, you can insert this single-pixel image in your post, with a file path that goes back to your server. Save a unique pixel for each site that you guest author on, and now you can see how many people view your article by tracking the hits to that pixel. Neat, eh?

What do you do with the firehose of data?

Start with measuring the contribution of all you RIA's.  Unica's NetInsights includes, among other things, a tool for monitoring the contribution of Web 2.0 applications toward your site goals.  This looks very cool, and if I could afford it I would love to get my hands on it! (I have no idea how much it costs, but they do have a free trial available.

A bar chart shows the breakdown among applications, and adds segmentation.  In this case, the segments are "Point and Shoot" and "Professional."  Very nice...it shows volume and dollar amounts by segments, giving a precise picture of which channels are actually making money, going way beyond pageviews.

Now we see a pie cart of camera review features. This report shows how many visitors, viewing time, resulting orders and dollar amounts for each of the product review features (zoom, rotate, reviews, etc.).  Looking at this, you can justify employee allocations to the various features.

There's also an intuitive path analysis, with a branching chart. This runs either from the starting point, or the end point.  Each point can be a page or a monitored action.  You can see what brought someone to an action (such as zooming), and where they went (did they click a button?).  If you see interruptions in your desired path, this shows you where to focus your usability efforts.

Measuring Community, Commerce and Engagement

1. Run a report that shows your reach, by segment.  Maybe by onsite behaviors...Creators, Readers, Critics.
2. Run Conversion rates by segments. This can show you which behaviors are associated with higher conversions.  This shows you which behavior groups you want to move people into, for the appropriate engement level that spurs purchase.
3. Behavioral funnel. You can set your funnel to focus on segment movements, rather than strictly sales conversions, and watch your progress in meeting you segmentation goals.

"An Insight Machine"

Custom reports identifying preferences, such as how many people rate a specific feature as important.  This tells you what your customers want, and creates an opportunity to provide information to your suppliers.  How cool would that be?  Recoup some of your inventory costs by generating valuable reports on consumer preferences that you can sell to your suppliers...who in turn can create more salable products that make you more money.  Can you say win-win-win?

Relationship Marketing

In a warehouse-style analytics system, you can drilldown into your funnel and see who composes each level.  In this case, we look at how people rated an LCD display, and where they sit in the funnel.  In this case, many people rate LCDs as important, many rate them as low, and then just 1 buys it. With this report, you can get back to those who are not satisfied with the display, and attempt to improve the situation.  The report exposes motivations behind buying (and product return) decisions.

Automate Communication

By setting decisions trees--based on onsite behaviors, communication permissions, and expressed opinions--you can optimize your communications. Is an automated email appropriate, or does this one take a  live phone call?

Online/offline interactions

If your online customer returns the camera to the store, how do you capture this?  Does your left hand talk to your right hand? 

Reearch by JupiterResearch shows that the rate of offline sales following online interaction is GROWING.  It's increasingly important to integrate and monitor your offline and online systems and marketing.  This is an ongoing challenge, and has been since the inception of e-commerce.  Its tough. 

You can try:

• Display and retrieve customer codes.  "VIP" ID's are one example of this.
• Correlate your trends.  Much like monitoring traditional advertising, you can apply the same methods to check for in-store trends that correlate with your online marketing efforts, like an email shot.
• Display unique 800 numbers. Customers who call the number got it from the web.  At least one company is very aggressive with this (Clickpath?) ad assigns a different number to referrers from each keyword or action, giving more precise connections.
• Tracking codes. Use special barcodes on printed coupons or reservation confirmations to indicate how the sale or download originated.

These help, but stil don't capture the person zooming on the camera and then buying in the store.  So what else?  There are campaign management solutions (including Unica's) that can "reconnect the dots," such as associating an offer code, loyalty card, or account number with an internet cookie; the interaction data associated with online actions can create a straight line from online interaction to in-store interaction.

There are many, many challenges in this. How do you account for in-store browsing?  How do you know if the information they give you, such as their Hotmail account, is anything of value.  (How many of us have a 'junk' email that we use for all of our shopping registration accounts?)

Wrap-up

Wow...that session was fast and furious, and packed with ideas for approaching today's measurement challenges.  As a Google Analytics user, I can see a couple new tricks I can try out on the sites I manage.

What are your tricks?  How do you approach site performance measurement in the Web 2.0 world?