Friday, September 25, 2009

Dear WebTrends, I need some REST!

REST logo
Early August 2009, it was quite difficult to not hear about WebTrends Analytics 9 launch. The new release generated a lot of buzz in the Web Analytics community – mainly about the new “revolutionary” user interface that “reinvented” Web Analytics (Cough! Cough!). Well, I had the opportunity to get my hands on the new baby and to test it. To be honest, what excited me the most was not the new UI (which is good I must admit) but the integration of REST URL’s, the latest WebTrends data extraction API.

What are REST URL’s?
REST acronym stands for Representational State Transfer – in case you want to know. It relates to a style of software architecture for distributed hypermedia systems such as the World Wide Web (thank you Wikipedia!). But let’s put aside the technical aspects. What is important is what you can do with REST URL’s: extract data from the Web analytics system in a easy and structured way!

WebTrends REST URLs generator
A REST URL will return a data set whether in XML, HTML or JSON format. One can not only retrieve report data but also profiles list, reports list or even report properties. WebTrends people are so kind that they even provide a REST URL’s generator that helps you to generate the URL’s you need.


The URL can be used in whatever application that support Web requests. First one that comes in mind is of course Excel - the most widely used spread-sheet application in the world. Just imagine that you can build self-refreshing dashboard & tables in a very easy way. You can find an example on WebTrends Developer Network site.

With a bit of programming expertise, you can use REST URL’s to feed other data systems. Or you can integrate data in a website like an Intranet site and display key web statistics to the whole company for example. Or develop a Flash desktop dashboard that retrieves & refreshes data automatically, giving direct access to key infos to your stakeholders in one double-click.

Well, you get the idea, REST URL’s offer a lot of possibilities to better communicate Web data!

Free the data! Communicate them!
Get data out into whatever format you want
With REST URL’s, Webtrends finally opened this 'closed box' that contains all your precious Web data. Analytics product is really good at what it is designed for: convert Web data logs into (plenty of) reports. But Webtrends Analytics like any other Web Analytics tools is NOT a communication tool. Period.

Data needs to be communicated to the people who need or use them in order to turn these into “intelligence”, “knowledge” or “insights” (pick your favourite one).

To quote Dennis R. Mortensen, “providing people a login and password to the analytics interface, is in most cases more harmful than good”. I can’t agree more with that! No one can expect business users, managers or executives to log in the tool, to start learning how to use it, to find their way in the enormous pile of reports, to figure out how to extract the data they need... And in the end, what they get is just data, still have to transform these into insights.

Web activities are often only a part of the whole picture. Being able to get your web data out of the 'closed box' allows you to integrate them into your reports/systems with other data sources.

Data must be freed from the analytics system so you can communicate them in most appropriate way according to your stakeholders. For that, openness is key. WebTrends REST URL’s are maybe not revolutionary (other tools also have very good API’s to extract data – just look at Google Analytics & Excellent analytics plug-in for Excel) but WebTrends certainly made a huge step in the right direction towards data openness. Finally!

I want it! I want it! Pleaaase!
Please, I am begging you!
Unfortunately for me, my company is a software customer, so no REST for us as it is currently only for On-demand customers. Boohoohoo (me crying)!

Such flexible API is terribly lacking in current WebTrends Software offering. Yes, the software version has the ODBC driver, which is good and useful. But it is showing limitations as it is not meant for retrieve huge amount of data. So when we generate differents (automated) dashboards on a monthly basis for more than 30 countries and two brands, I can tell that our server suffers!

Moreover, implementing good ODBC dashboards require some expert programming skills. In our case, our former Web Analytics agency had such expertise and did a very good job (thanks to you Guillaume! :-)). On top of that, the ODBC driver needs to be installed & configured on every client computer, format is not real SQL... All in all ODBC lacks the flexibility that REST URL’s offer.

So I am dying for the day when Analytics 9 will be released in Software version with, I hope, REST URL’s support. Please Mr. WebTrends make sure it will be there. It is so so important. Otherwise it would be a huge disappointment.

And don’t make us, software customers, wait to long, pleaaaase!

Related resources & references

Friday, September 11, 2009

Usability labs: a way to really understand your online customers!

Hi, I am a Web user. Called me stupidImagine the scene. You have what you think is a very simple. But when you look at the poor results you get from your favourite Web Analytics, it leaves you scratching your head and wondering: “How can our online customers failed to use it properly? Are our website users stupid?”

“I am a web user, call me stupid!”
Well, there are probably few of your users who may be stupid like in everywhere – but no, online users (i.e your customers) are not stupid! It is just that too often, we get it wrong!

Yes us, you know the super Web experts (web designers, online marketers, Web Analytics ninjas).Too often, we forgot one important fact: we are NOT our customers. We are not our Web users. The problem comes from the fact that our perspective is biased. We know all about our websites, their functionalities & tools. We designed these, we know their purpose, and how these are supposed to be used.

Our customers don’t. They have a radically different perspective and it is a very difficult exercise to put yourself in their shoes. For me, we should never forget a golden rule when trying to interpret the results of any online page or tool: "Never assume that other people surf they way we do"

“If users are not stupid then data must be wrong”
Let me share a personal experience. A while ago, we started a new major redesign of our website. In that context, we did some measurements to better understand how users navigated from our homepage to the product sections. One way was via a product overview page – that was supposed to help our customers choose a specific model and drive them to corresponding model section.

Marketing didn’t expect the results we got. The page was not performing well. The results were so surprising that first reaction was to question the data (“yeah – let’s blame the data”). But data were proven to be correct. So how was that possible? After all, the page was darn simple to use. At least that what we though. The answer came few months later…

Usability labs to the rescue!
Users under close observation during a usability labAs the new site design was in advanced stage, a usability lab was done to test the new layout. The great thing with usability labs is that not only users are filmed but they are also interviewed: they explain what they think about the page, how they understand it and what they expect from the page to do... It provides qualitative insights (hurray!).

The overview page was part of the content that remained unchanged – still it was tested and we finally got our answer. We actually saw it with our own eyes: users were definitely struggling with the page. While they entered the page with the right intent (i.e. get an overview of the full product range), they were puzzled by the way it was presented and how it behaved. They basically expected something else, they got confused and mostly frustrated in the end.
And what was a mystery for us suddenly became so obvious. Yes, they were absolutely right (customers are always right anyway :-))! How did we manage to not figure it before? Because we are not our customers.

Qualitative is a must-have because can it bring some lights on the numbers and the “why” that is missing. Without, one can spend weeks trying to figure out the reasons (and never find these).

D.I.Y. Usability labs
Usability labs are one way to get qualitative data. Sure, usability labs have some flaws and limitations but the great thing is that you can really see what people do with your site and understand what's in their mind.

“But a usability labs is hard to set-up, requires an expert company and damned expensive” you might say. It is true that usability labs are often used for major redesign or critical projects. These are often handled by expert companies that work for a price. Not something most of us can afford too often.

However for smaller projects, you don’t especially need the “big artillery”. So why not doing it yourself? It is not that hard to set-up your own usability lab. One laptop with a video camera & recording software, a moderator (ask your beloved web agency for help), a quiet place to do the tests and some incentives for the testers (like free Belgian chocolates or…beers! ;-)) and that’s it. It shouldn't cost you more than 1,000 or 2,000 EUR max. Of course, it may not have the rigorous & scientific approach that you get from an expert company nor the big powerpoint/100-page report but what you can learn can be very valuable.
The 5 things you need to run your own usability lab
Small details can turn a project into a failure. Doing light usability labs can help you uncover these and avoid painful results. Again, I am talking about experience here. For a recent project, we did a small usability lab with the help of our Web agency, all at a low cost. The feedback was not only very interesting but more important it helped us to discover a serious usability issue caused by a little detail. Project was a bit delayed in order to fix the issue but without this finding, we would have put live a project that would certainly have failed to achieve its goal.

Get qualitative insights!
All in all, the important points of this post are:
  • You are not your customer - never assume they surf the way you do.
  • Get a better understanding of your customer: complement your data with qualitative insights
  • Small usability labs can be done for a cheap price and still get you valuable insights that can make a real difference, from failure to success.
  • For bigger projects/major redesign, look for real professionals. It has a cost but often worth it.
And you, have you ever done usability labs? What do you think about these? Like or dislike? Don not hesitate to share your own experience and views on that topic.

Related resources:
Feedback Form