May 19, 2012

Web Metrics Pitfalls and Incentives

Website managers operate in a data rich environment. If it happens on your web page, & you decide you care about it, you can track it. And if you can track it, you can set goals around it & hold people responsible for meeting those goals…. The marketing team is responsible for transfers. Repeat visits are the responsibility of the engineering team. John is responsible for new account sign ups. With appropriate instrumentation & tracking (sometimes more difficult than it sounds!!), an organization can evaluate whether individuals & groups are meeting their goals. Evaluation is step one towards compensation, and when compensation is on the line, you can be damn sure people are going to be focused on making. their. numbers.  If you are a boss, YOU better be damn sure you are giving your people the right numbers to focus on.  In large organizations esp., many people end up  managing to a single number.  And throughout large organizations, even when many metrics are consulted, the drive for simplicity & collective understanding (and hopefully the organizational focus that results) will often lead people to a single number. One metric to rule them all, if you will.

One obvious way to go wrong when using this model is by picking the wrong metric(s). Again, focusing on a single number magnifies this danger.  For example… driving transfers to your site is your goal. Single-minded pursuit of THAT number can easily result in driving unqualified, unproductive traffic to your site. All while you comfortably make your goal. Oops. In this case it would be best to balance w/ length of session and/or repeat visits. Admittedly, this introduces complexity.

The more insidious way of going wrong is gaming the system. Metrics are about measurement. And measurement requires a known process. And known processes are gameable. You could change your URL structure to get previously uncounted pages counted; you could split articles into multiple pages to spike page views; you could redirect established traffic funnels.

And let’s not forget about who’s at the wheel… a bunch of type-A achieveaholics.  And that’s life, but if you give a type-A achieveaholic resources or an established traffic funnel and a number to make… suddenly massive amounts of traffic are being redirected to achieve goals. And on the collateral damage side of the equation.. the users.  They end up getting jerked around so people can make their goals & get paid (worthy goals, no doubt!).

And here’s an additional downer, if an independent party is responsible for measurement & you get caught gaming their system, it could sully your relationship w/ them & look bad publicly. Internally, the consequences are potentially worse. You’re going along thinking you are doing fine, all the while you could be undermining yourself.

So what do you do?  Most people say to go simple.  But single-minded focus on achieving a number can blind you to the complexities of what is happening on your website.  Instead, I recommend countering the drive for simplicity with some supplementary complexity.  Single numbers (or small numbers of numbers) are powerful tools for clarity & focus.  This is especially valuable in a large organization where people can easily lose sight of their goals.  But there have to be additional numbers that people watch.  Ideally these would be metrics that can provide contrary evidence to success in the face of growth in the key metric.  To repeat the example from above, if driving trials is your goal, stay focused on that number, but also keep an eye on repeat visits, & length of session.  Ultimately, you want to direct your numbers towards your users.  Of course, in a cookie based tracking system, you can’t really identify the user all that well.  But that’s why the job is fun, or at least, they need someone to do it.

Originally Posted at: http://www.menggoh.com/63/web-analytics/web-metrics-pitfalls-and-incentives/

Analytics Mission Statement and Team Structure

Analytics Mission Statement: “Bring data to the masses and make data-driven decision making a reality”


How do you build an analytics team? Tell your manager you need money to measure social media! No seriously, I heard it’s easy to hire people, everyone on the “interweb” is “social media experts”. For some of us, we have multi-channels analytics need, SEM, Display Advertising, good ole fashion Direct Marketing and a whole bunch of other internal data we need to deal with. Building a team could be quite daunting.

Before I starts the usual blah blah on how to do hiring, I want to share a story.  In the not so distanced past, I had my career review(here at the mother ship we take it seriously), the memorable feedback I received stand in my way of promotion is leadership. “You have excellent management skill, Meng; but you need to improve you leadership” said my manager.

What is Leadership? For me, leadership is to have vision, to anticipate future growth and direction of the company, have the courage to take calculated risk and execute. So…do you have the vision, courage and patience to bring ass-kicking analytics practices, insights and in turn “mucho mullah $$$” to your company? Do you have the courage to take calculated risk? Be the Hokage of your clan, leads your ninjas to bring insights to your organization against the corporate hacks, bureaucrats and simple-minded marketers (I kid, I kid, I am a marketer at heart). Have you been wrong many times? If you haven’t, it’s time you try something new because you haven’t been “A/B testing” everyday and doing your “MVT”. In fact, My A and B both sucks many times!

Our team has very rigorous process of setting yearly commitment tied to precise deliverable. Having a analytics mission statement help guides me to focus on the right thing and not waste resources. I know mission statement is corny, I don’t care, I am writing it anyway. My team’s mission statement is “Bring data to the masses and make data-driven decision making a reality”. Here are a few sub-objectives to bring more clarity to that mission statement:

  1. Build a sustainable data infrastructure (must..resist..the urge..to knock..Google Analytics) to measure multi/cross-channels digital marketing ROI. I have two types of audiences: a. the big bosses, you know, the gazillions Directors, GMs, VPs around here. b. the marketers and agencies. For the big bosses, we deliver “BI” and KPI, for the marketers, we delivers data (see #2). We build platforms and tools to enable marketing operation, reporting and analytics, share best practices and improve Go-To-Market efficiency. For example, we build our own data warehouse and segmentation tools, our own marketing process management workflow tool and execute our direct marketing, all in-house because scale is the challenge with everything we do here. Our scale is enormous in comparison to other companies, we collect petabyte of data, from which I need only a few drops of marketing traffic, a strict qualitative ROI model would’ve #failed as our primary objective is still mainly to build brand. We have our internal awareness and perception tracking system to measure traditional media such as TV ads and brand improvement. We use comScore and Compete for research, out of those insights we then have to build targeting capabilities, it’s useless if we can act on these finding.
  2. Knows where my team fit in and build efficient organization structure. I am part of centralized Business Intelligence and Customer Intelligence team, we serve both product management and marketing and I am on the marketing side.  On the marketing side, each marketing department from different business groups have their own marketing analytics people, they are our partners. The reason behind this structure is ..well, it’s your money, if you wasted it buying $20 CPC keyword because your relevancy and quality score sucks, well it’s your fault. Simple as that.
  3. Review and improve our processes, conduct researches on various marketing campaigns and channels, share best practices across different business groups. Publish training and information so that marketers worldwide can efficiently leverage our capabilities.

Now I have a better idea on what type of resources I need to deliver my commitment. I have databases to maintain so I need a SQL expert. I have campaign sites to tag so I need an instrumentation consultant. I need to pull data regardless of what all the expert say, so I need reporting robot. Lastly I need “Analyst”, my ninjas, the one who dig into the data and find the golden nuggets. I am ready to build a team. As you can see, my needs are very different from yours, so my team would be very different from yours even though we are in the same analytics field of work.

If you have a mission statement for your team, I would love to hear it. If not, maybe it’s time to create one.

Originally Posted at: http://www.menggoh.com/72/management/analytics-mission-statement-and-team-structure/