First Comes Clarity, Then Come the Tools & Data


As I wrote in my last post, effecting real change can be difficult and, from a practical standpoint, change comes only when you do something different. I promised to share some things my team is “doing different” in 2013.

Let me first say that, as usual, All Analytics community members jumped in with great comments on that last post, several of which focused on tools and measures. I think it’s only natural for analysts to think first about data and tools, because these are core to what we do and we know what they’re capable of producing. But for my team, as we strive to bring a new level of analytics to our stakeholders, tools and data are literally the last things on which we’ll be working. That’s because in my own experience and in observing others, I’ve seen too many analytics initiatives fail because they started with the data.

I’ll repeat it until I’m blue in the face: Analytics is about questions. I don’t care how well you think you know your business, if you don’t first start with your decision makers and really get inside their heads, you’re going to come up with solutions that miss the mark. Everything pivots on having clarity on the questions that need to be answered. So until my team has that clarity, we’re not moving forward on tools and data. Here’s how we’re going to get there:

  1. Clarity on decisions. If the North Star of our analytics is to make better decisions, then we need to start by understanding exactly what those decisions are. So we’re going to start with spending time with our stakeholders and mapping out their decision making processes. This includes understanding the specific decisions, how often they make them, who else is involved, and how important each decision is.
  2. Clarity on questions. Once we have clarity on the decisions, we will then turn our attention to the questions. For each stakeholder and each decision, we’re going to identify the specific questions they need answered in order to make the decision. Although some of the questions will be data-specific -- such as “How much budget do I have to spend against this activity this month?” -- some will be more open-ended, such as “What other opportunities am I unaware of but should be exploring?”
  3. Clarity on answers. Now that we’ve gained clarity on the questions, we can start to think about data. But instead of focusing on systems and datasets and variables, our goal in this step is to focus on the metrics that we would need in order to answer the questions. In this case, “metric” can be abstract -- i.e. we may not know precisely where the data is to calculate this metric. We’re still not quite ready to talk about datasets.
  4. Clarity on reports. With our metrics identified, we turn to the presentation of the metrics. Here is where we’ll build mock-up reports and dashboards, defining how to display the metrics in a concise and actionable format. They should answer the known questions and also stimulate new questions.
  5. Clarity on data and tools. Now that we are clear on how we will deliver the answers to the questions needed to answer our stakeholders’ questions, we can finally turn our attention to the data and the tools needed to produce the reports. It’s likely that our initial deployment won't be able to produce all the required data or answer all of the questions, so this step includes a roadmap for how we will eventually satisfy all of our requirements.

This framework will ensure that we remain focused on our North Star and don’t spend our time putting together solutions that we think are helpful, only to find out at deployment time that we missed something. This is not an easy task, but a key success factor to this approach is how we engage our stakeholders.

How do you ensure your analytics efforts don't miss the mark? Share your advice below.

David Schmitt, Analytics Director, Hospitality Industry

David Schmitt, an analytics director in the hospitality industry, leverages his unique blend of analytical, technical, and communication skills to deliver key insights that help senior executives make better decisions. He has demonstrated success in leading formal and informal cross-functional teams to develop robust enterprise-level datasets and apply advanced analytics that turn data into practicable insights. His strong presentation and writing skills enable him to turn technical analysis into knowledge that audiences find engaging, relevant, and sometimes even fun. Currently a director at a global hotel company, Schmitt previously held a variety of positions in the high-tech industry, including several startups and Apple Computer. He holds a Bachelor's degree in electrical engineering from Georgia Tech and an MBA from Northwestern's Kellogg Graduate School of Management.

First Comes Clarity, Then Come the Tools & Data

My analytics team will "do different" in 2013 by first understanding the questions that need answering -- and then turning to the tools and the data.

How Do You 'Do'?

You may need to "do different" to affect the change needed for improving business decision making in 2013.


Re: first step
  • 2/8/2013 11:21:08 AM
NO RATINGS

Communicate and collaborate? Easier said than done, obviously, as this has been a mantra for years ...

Re: first step
  • 2/8/2013 7:26:44 AM
NO RATINGS

I agree Cordell - in many corporations there is still a disconnect between the business and IT. How can they better align goals?

Re: first step
  • 2/7/2013 5:05:24 PM
NO RATINGS

And one that's desparately needed, and long overdue. I think self-service has its place but the deeper a company gets into analytics the less self-service has a place, I think.

Re: first step
  • 2/7/2013 10:47:00 AM
NO RATINGS

There's plenty of blame to go around and you've hit on a key problem - neither spends much time thinking about analytics.  Now before everyone gets excited let me explain.  IT doens't spend a lot of time thinking about capabilities until their asked to do analytic work that requires something new.  Business doesn't think a lot about IT that could bring them new capability (they've got a business to run).  There's an ongoing problem of the business analytics thinking they can never get what they need into the IT queue.  This is why we're seeing so many self service IT products in the marketplace.  We need to see a melding of IT and business and not think of them as two seperate entities.  That's a different mindset.

Re: first step
  • 2/7/2013 8:18:16 AM
NO RATINGS

Interesting perspective on that mindset Cordell. Where you see that mindset would you say it emanates from IT or the business, or both, mostly?

Re: first step
  • 2/7/2013 1:47:36 AM
NO RATINGS

No you are correct. That's the other choice. Its just that few organizations have that capability. And many don't have the right mindset. They tend to think big means slow because that's their experience.

Re: first step
  • 2/7/2013 12:25:11 AM
NO RATINGS

But Cordell, isn't the goal/advantage of high-end, high-powered analytics (I'm thinking of in-memory or in-database analytics, for example) that you can cycle through iterations quickly? This is one of the advantages that SAS (this site's sponsor), for example, talks about with its Visual Analytics offering, I believe -- sitting down with the business, selecting variables, running the analytics, delivering the visual, and then selecting different variables and repeat. All super fast. Or aren't we thinking about "iteration cycle" in the same way? 

Re: first step
  • 2/6/2013 5:14:17 PM
NO RATINGS

I agree David that this is about reducing risk - specifically operational risk.  You have two choices, you can either achieve clarity, beginning with the end in mind so to speak, or you can work in environment where making mistakes has a reduced cost.  For those that grew up with Agile development it's harder to step back and focus on design.  They'd rather build, see if it worked, build again.  That's fine if the iterations cycles are cheap and quick and impact of errors isn't great.  Unfortunately I don't find too many places like that - especially with large datasets and high powered analytics.  Iterations are long and decisions made on faulty assumptions are costly.

A nod to community
  • 2/6/2013 3:39:23 PM
NO RATINGS

Hi David, thanks for pointing out the discussion your last post generated among All Analytics community members! That's what we're all about as a community site -- so here's to everybody reading, keep up the commenting! We love to hear from you.  

Re: first step
  • 2/6/2013 3:28:07 PM
NO RATINGS

Phil, I think certainty is perfection in another form. And as I've long stated, Perfection is evil! 

I don't think certainty is the goal. In a prior post, I wrote that all decisions are gut-based. Every decision involves some amount of gray area. Analytics helps by reducing the amount of gray, but I don't think it ever completely eliminates it.

But you do want to minimize risk. My next post will have some ideas about how to engage your decision makers to help reduce the risk that what you produce for them will be valuable to them. Stay tuned!

Page 1 / 2   >   >>
INFORMATION RESOURCES
ANALYTICS IN ACTION
CARTERTOONS
VIEW ALL +
QUICK POLL
VIEW ALL +