Watch: The Value of Data Visualization

Produced by Column Five Media, this is a great video that I hope to use someday to present and underscore why data visualization is important (though it should be self-evident).

Additionally, videos themselves are a form of visualization of analysis, tell a story, and provide a controlled, orchestrated presentation.  Creating such videos as these are something larger companies should consider as part of their reporting toolkit.  So often videos are used as internal branding, or vision inspirational, but using them to show and tell the story of data can make it extremely compelling.

Mining The Thought Cloud: Presidential Address

So I’ve been reading (and agreeing) with the use of search.twitter.com to see what is going on with the thoughts that people jot down on Twitter.  Particularly after hearing Obama’s first House Presidential Address.  And what better way to mash the dataz than syphon it through Wordle.net’s awesome word cloud creator (patched in via the feed function).  Below are some great gallery shots from just after the speech.

Skyrails: 3D Data Visualization System

I came across Skyrails through my regular Feedburner subscribe to FlowingData.  I love data visualizations that go outside the norm and break us free of bar, pie and line charts/graphs.  Also, if you want to display multiple data sets with complex and ever-changing information (GASP – like a live business!).

Skyrails is a completely programmable and customizeable visualization system.  And it looks like a visualization from the Movie Hackers (remember the worm-virus visual?). I’d love to hook this into an API tied to live stock market numbers and watch the ebb and flow of the economy.  And surf it to financial domination .  Someone buy this for their web analytics platform for visualizing website traffic through the site!

Hackers Movie:

ROI for Product Reviews

You just paid to have some expensive, or even free (but you paid programmers) to install product reviews on your website(s).  Great idea!  You know product reviews have value.  Duh.  But how to track a return on your investment?  Product reviews are a “conversion enhancer” and have lots of other benefits of increasing the effectiveness and customer retention of your site.  Trying to get an ROI out of it can be a lot like trying to get the flour out of cookies after you baked them.  Even if the cookies are delicious.

So, the question is: Is having product reviews getting you more cookies (yes, I’m going to keep running with this)?  I’ve tried to approach this several ways.  Of course, it often depends on what your needs are (or those requested from your supervisor).  Need to track this back to a P&L response?  List out the tangibles and intangibles?  Some other objective?  Hard, but not impossible.  Below are a number of ways to track an ROI, with some detail on how to really get started (not just a direction).

P&L Effect with Product Reviews
Method 1: Track customer acquisition.  Most product review services like Bazaarvoice or Power Reviews have some kind of SEO page that gets submitted to the search engines (even if just by Google’s SiteMap XML).  So, when a customer searches “Review for Product XYZ”, a page that is just dedicated to reviews of that product – your page, comes up.  Now, if you have analytics attached you can tell how many sessions that is nabbing you.  You should also be able to scrub that against other paid campaigns (ie – Omniture’s External Campaign tracking) to see if there is a net benefit: Revenue, Orders, or Leads.  Don’t discount “assists” in this, that this helped get a customer, even though they came back via search or some other campaign.

Method 2: Track Conversion Differences.  This is tricky and may require a complex analytics package like Omniture SiteCatalyst, Webtrends, etc.  I’ll explain it in Omniture terms, as it’s the system I know.  What you should be able to do is find out a list of products you know that have reviews: call this Test Group.  Next, you need the inverse (or called ‘none’ in Omniture reports) which are all other products that do not have reviews: Control Group.  Now, if you know you’ve done something recently to burst, increase or cause the number of reviews you’ve gotten to spike, this is an ideal measuring point.  If you don’t have this you could take the time before your product reviews were live.  So – you have your two groupings of products, and two time sections (before the “burst” and after – or before reviews and after).

In Omniture you can use SAINT and Classifications to group your Test group products.  The basis for this is to create a group that “eventually has a review” vs “those that never had a review”.  Now – trend them over time, to include a group of time (slice by month preferably).  You want at least 3 or more data points for each time period.

Now, if you’ve done all the classifying you can run your product classification report to see what your results are.  You will want to use a calculated metric to see what your conversion rate is.  I choose [Product Views / Orders].  Product Views, as you see the product review on the product page.  Orders because that’s the point!

What I did here was to take the data points of each line, for each time period: Before and After.  Take the average conversion rates of those time periods.  Is there a differences of those averages from the Before vs the After?  Was there growth or shrink?  If you had growth – You haz more cookies!  If less, maybe you had a lot of negative reviews?  Did you have no change?  Maybe the reviews weren’t being easily found or advertised.

If you do have a growth number (congrats!), then you can tie this back to a P&L effect.  The X% growth means that is growth on those products with reviews that you would not have had if you did not have product reviews.  Now, just do a conversion rate comparison to the After time period, on products that had reviews.  Reduce the conversion rate by your X%.  Apply your segment’s numbers for Average Order Value to both conversion rates.  Now you have Revenue with reviews and without reviews = P&L effect!  Apply that to the costs of your program and you have a ROI analysis.

Method 3: Marketing Emails
While you can “season” your entire website with product review ratings, don’t be afraid to use them as the prime message in a marketing effort.  People love opinion, discussion and feedback.  As retailers we can get very price focused because the customer’s appear to only react to price (free shipping, sales, coupon codes, etc). 

Try a marketing email with nothing more than products with ratings that include snippets of comments and short stories.  Direct links to other product review content can enrich it even more, if your review system supports pictures or video.  I’ve seen these emails do about the same sort of performance as an email with a good coupon code or sale.  This means margin you’d be spending on a coupon now goes right to the return of your program costs.  Of course, you usually need some reviews before doing this, but you don’t need too many to get started!

Method 4: Other Trackable Ideas
These are some ideas that I’d personally like to follow up on, but have not yet done the work.  Reduced Returns – If you have a good internal reporting system, you might be able to correlate a product having a review on it (based on the date it went live) against a change in return rates for those items vs those that never had a review.  Call Center, Reduced Calls – If you have the ability to track when a call center rep looks at a product (usually this is only when a customer asks them to), then correltate products that have reviews to those that do not have reviews.  I don’t think this is valid, as I’ve personally taken calls where a customer asks a question specifically about a review.  I mention it as I’ve seen it stated as a reason in other “best practices for ROI”.

These are the best methods I’ve been able to see for getting a good idea, and actionable means for getting an ROI into a presentation for your execs.  Maybe even a case study.

You KNOW that product reviews are a great value.  Your customers know.  Now prove it and show your company that your idea to implement product reviews brought the company $X in money and that they should give you a big raise and bonus.  Or maybe a share of the cookie jar .

Customer Analytics in Real Life

We use the information the customer gives us to help us better market to them, and hopefully improve their shopping experience. What would that be like in real life? Creepy comes to mind. Check out some of these videos. Some of them even bring specific service/product vendors to mind, that provide website solutions.

These are made by a company called 7 Billion People that make web analysis tools that give merchants natural language business observations about what your customers behavior and what they are interested in, so you can tune and test what you are saying to them.

“Foot Lotion” – Customer demographcs and behaviors. Ah ha! Are you back for foot cream? Wart remover? Please buy!!!

“Add to Cart” Did you want this product you saw last time? Rate me and I’ll show you more!

“Proceed to Checkout” Love the mention about MAP pricing (add item to cart to see price). Oh, and the “I’ll just add that to your cart…” *cough*GoDaddy*cough*

Interesting note: I’ve spent a lot of time talking with customers, as part of customer service to an internet-only retailer. You’d be surprised at how many of older customers expect an internet retailer to know and see what they are doing. After all – it’s your website, isn’t it? “Can’t you just pull up my shopping cart and see what I’m looking at?” 

Feedback from the Customer

There is nothing better than hearing from your customer: what they think, experienced, had trouble with, or got stuck on. Getting this information is like gold – it can be used soooo many ways. As analysts and especially those in Customer Intelligence – it’s creating actionable information from this feedback, that creates the core for what we want to present to execs, site managers, programming, etc.

Take this example:
“The site was very easy to navigate, the only suggestion I can make is to allow to search, other than by brand, but by color or style and narrow down further by another factor. I was unable to this, but could if searching by brand. Thanks!”

Fair critique, with positive feedback and a specific suggestion. Now, the customer may or may not have the best understand of website design, usability, function, logistics, etc – BUT there is more you can glean even the most basic statement. Other than the obvious – is there more here?

• Usability suggestions: Customers are looking for ways to shop and browse by colors and styles. Maybe, more than just drill-downs, you have other ways for customers to find what they are suggesting. Design, programming and site leads love this info.
• Natural Language: They used “color” – not finish, option, or some other term I’ve seen can be adopted by manufacturers or industries. Watch for other terms they might use about your products. Maybe your category names, product types, attributes or other basic terms you use might need to be brushed up. In our profession, we like to be accurate, but customers use what they use – in search and what catches their eye.
• Education/Age: Ok, this goes into speculation, but you can discern a person’s education from what language they use. Punctuation, grammar and structure of the feedback are other indicators. Some might also be derived from age as well, as older might use more formal speech patterns. IF THEY ARE USING ALL CAPITAL LETTERS, THEY MIGHT BE MY GRANDPARENTS.

Most already have a number of sources of customer feedback easily at hand. Some are a bit more difficult to catalog, index or get into a ready format to analyze and aggregate. If you take away our aggregation skillz, then where would we be!? The question is, how to get this gold without having to break your back to get it. Or your intern’s back. They are usually younger anyway…

Some great ways to get feedback that are often automated (after you have them installed):

Bizrate: Merchant Rating System. Bizrate is a company that changed it’s name a bit back to Shopzilla, but still maintain a website that has a decent system of offering post-purchase surveys to customers asking them to numerically rate the merchant and the products carried. I like them, as they are very automated and offer good aggregation and I get the comments emailed to me every day, what my customers are saying. I don’t like them from the standpoint they can get a bit officious or customer-tempting to get a customer to write the survey, and that might lead to a negative experience which might reflect on the merchant.

Product Ratings & Reviews: Having a customer write information and add content to the product page is very powerful. You can implement basic “write a review” functionality to your website, or have a vendor run the service which might include content moderation, hosting, additional widgets or review syndication. The two big vendors in this space are BazaarVoice and Power Reviews.

Call Centers: These can take a bit of effort or coordination, but if you can get support from Management to have your phone reps log the basic types of calls or take down specific feedback onto a place to be logged, this can be great feedback. If you have a phone system that can record and email phone calls (I love this), these can be great to review, though a bit more time consuming.

So! Lessons learned? There are some great things you can get from Customer Feedback, even beyond the blatant messaging of what they are outright telling you. Also, there are some great sources that can provide some tangible benefits and help you create a more accurate analysis, punctuate a presentation or make a outright statement to your management.