1. working with product and marketing managers, data analysts and other business users to identify what data they need.
2. working with the engineers to make sure the tags are customized to capture the much needed data
3. using tools like Firebug to verify and QA the data sent to the web analytics companies is accurate
4. verify the data in the reports and support users in utilizing the data

Big data, more than even web analytics needs you! Most hadoop and log files are exploding with data that takes a long time to surf through to identify what we are looking for and the data quality of big data is one of the biggest sticky points that I have experienced so far. Big data also requires someone that’s comfortable:

1. working with business users and data scientists to identify what information they need for analysis
2. working with engineers to make sure the web/mobile applications are instrumented to capture the data that users need
3. using tools like Firebug to verify and QA the data is captured in Hadoop (inhouse or in the cloud)
4. verify the data in hbase or mongoDB or whatever database the analysts are going to use for analysis

So you see, big data implementation management is not that different from web analytics implementation. You will need a couple of additional skills and you will be working with a lot bigger datasets, which could be a good career move. Wouldn’t you agree?

Marketers segment their customers by demographics (age, gender, location etc) and RFM (recency, frequency and monetary value) scores to create targeted segments for promotional offers. Recency means how recently did the customer purchase. Frequency means how often do they purchase. And Monetary Value means how much do they spend. But with big data, now we can store and analyze every single detail of every user’s behavior beyond their RFM scores e.g. the time spent/engaged on the site, what areas of the site the user was engaged in, the depth of user engagement (products viewed, videos watched, mobile check-ins, interaction with activity feeds etc)

Marketers have used demographic segmentation and RFM scores for email/offline campaigns as well as real time personalized ads. However, demographics and RFM scores are static and/or slow changing indicators. Though those are powerful indicators, many researches have  shown that users current engagement is one of the most powerful indicators of user’s future behavior. In other words, user engagement goes down prior they drop off/stop visiting the site.

Drop in a user’s engagement is an early predictor of her future attrition.

Now with big data, more than ever, it’s possible to identify the low engagement and intervene early, even in real time using machine learning. Don’t’ wait for days for the recency or frequency metrics to show a drop in the scores. The best time to push engagement features in user flows or offer promotions is when user’s engagement is going down but while they are still on the site!

Many companies, including some of my clients, are leveraging big data and machine learning to improve customer retention.

• Day part volumes – Softer peak hours and well distributed Actions throughout the day means users want to use App frequently. Don’t be discouraged if the visits are only few seconds long, it’s typical of App users.
• Velocity of Actions (most analytics tools prefer to tag them as Events) – The more clicks within the App, the more users are engaged (with of course exception of reading or video Apps).
• Feedback score volumes. The qualitative feedback is just as important. Always provide a feedback button within your App. In my experience, within App feedback volume far surpasses App store feedback. The more users are engaged, the more feedback you are going to get, even negative feedback is better than no feedback. At lease negative feedback provides you hints on what you need to change to retain the users!

This is surprising given that a product or service must respond to market needs or wants in order to succeed. If your transactional website has online product managers then you need a web product analyst too. This Product Management Department professional collaborates with user experience and design, market research, web and data warehouse analysts to deliver cross-functional data driven insights. The web product analyst is an essential member of the product development team who translates internal business requirements and external customer needs into best-in-class customer experience and satisfaction. All too often a web analyst focuses on maintaining the health of a website’s overall analytics platform, primarily reporting on site-wide characteristics and top level KPIs.

Product development is not viewed as a core responsibility or agenda item. The all important task of drilling down the data from many sources such as Web Analytics, Market Research, SEO/SEM, Data Warehouse and third party data is left to product managers. And frankly, most product managers are busy strategizing and writing specifications for the next product release. They are also likely unskilled in interpreting data for product or service development. The product manager generally deals with snippets of Web Analysts, Campaign Analyst, Marketing Manager and Market Research reports for their specific products. The level of information provided to product managers does not give them the whole story.

This is where a web product analyst comes into the picture, though this post may not be applicable to every company. For example, for a content driven site that makes its money from advertising sales only may not need a product analyst. However, you should consider hiring a web product analyst who directly aligns to product development if your website has multiple features that provide users with an array of self-service tools.

Well, should they care about the user browsing data from Mobile App? Does your business use Web Analytics data to make important business decisions? If the answer is yes and if your Mobile App is providing important functionality to your users, then Mobile App tracking is just as important as your Web Analytics tracking. With 15 billion iPhone Apps downloaded till date (reference http://bit.ly/qCdCNj ), most Mobile users are spending more time on an iPhone App than their Mobile browsers.

So what should you track from your Mobile App? Web Analytics and Mobile Analytics have a lot in common but also there are some differences to be aware of. Just like Web Analytics, you want the basic numbers of page views, visits, visitors etc and depending on the “success events”, you want to get conversion rates and KPIs of your Mobile App. But also there are many important differences to keep in mind.

It’s important to monitor downloads and activation rates for Mobile app. For iPhone Apps, Apple does not provide data APIs so reporting is a bit tedious. One can manually copy data from iTunes connect or there are screen scrapping programs out there but it is not ideal. Now, download numbers only say part of the story. What you are really after is the Activation numbers. How many people actually used the Application.

I have used Google Analytics visitor count and also validated that by with the data warehouse reports and the numbers were very close. Depending on the App and the segment of your users you may see 80-95% conversion rates. Some users download the app multiple times but some may not even use it at all. Especially if you have stringent log-in requirements, you may see a significant drop off there.

Mobile Apps do not have referral information and I had a hard time getting accurate time on “site”, meaning time user spent in the Application. Since there is no concept of a page in Mobile Applications, implementing Mobile Application tags is a bit different from the Web Analytics tags.

You have to identify what you want to tag as a page vs a click on a page or actions and send a virtual page views of events or commerce variable to the ASP provider you are using. If you have a lot of Rich Internet Applications on your site, you may have already done this exercise for your website.

I will write more about Mobile Analytics in the coming weeks/months.

Now, I understand we can’t be error free but what bothers me the most is most sites I deal with do not know what errorsare occurring on their site. We all have addressed the 404, 500+ issues and those are easy to track. But what about the user experience errors? When does your customer see “Sorry, we can not ship the order in your zip code” or “Please enter 30 characters or less in the name field”.

The answer seems simple, just enable reporting on it, right? But I have seen way too many customers still struggling with this issue. And honestly, I have to yet meet a customer that can furbish a report of all the errors on their site. Some customers buy tools like Tealeaf that goes a long way from where they usually are. They can start reporting on the issues that they are aware of by creating a text “search” or “event”. But what about the errors you don’t even know about?

I know we have several development teams and they are all working on meeting tight deadlines. All developers are working on pushing the code out on time and creating all sort of exception handlers to make sure user can find a way to get out the error loops etc. And it’s hard to justify creating a project just to put all the errors in one single database table and having a call made to the server for each error/exception that we would like to throw. So no one ever prioritizes the “Error” project. But only if they did… and integrated in their web analytics or customer experience analytics tool….they would know exactly what issues are occurring, when, and which issue is causing the most drop off…. Sous like a holy grail….don’t it! What’s the point of doing a fallout analysis, if we could tell what a user saw on the page where there was largest drop off?

With the Rich Internet Applications, tracking of the errors is even more critical. Now most sites have two level of validation – on the user’s browsers and your servers. And with every new technology, sites are going to struggle with new coding standards. And hence, there is no better time to standardize error tracking than the time you are updating your website with new flash/flex applications.