About Us | Login | Follow CITO Research:

Profile in Nerdage: Noah Broadwater, CTO, Sesame Workshop Part II: Portfolio Management

Nerdage Brain
Written by Dan Woods | November 08, 2012 | 0 comments

Noah Broadwater is the CTO of Sesame Workshop, the nonprofit behind the Sesame Street children’s media juggernaut. Dan Woods, CTO and Editor of CITO Research, sat down with Broadwater to discuss asset management and portfolio management from the perspective of the CTO, the effect and dictates of consumerization on the modern CTO, and his thoughts on demystifying the cloud.

In this segment, Broadwater discusses portfolio management.

DanWoods: What are your thoughts on portfolio management?

Noah Broadwater: Personally, I take the simplest solution: talking to people. I talk with staff, vice presidents, senior vice presidents, executives. If you want to know what’s broken in an organization, talk to the people who do the work. Get to know the staff in charge of a process. Understand their pain points because by the time it gets to someone at the CTO level, everyone’s already concluded that it’s either broken or it’s hunky-dory, but you never get any specifics.

Generally, I think the biggest task of gap analysis is understanding the gap between what’s happening at the ground level and what’s being described at a higher level. In that gap, you’ll find the nugget of truth.

You’ll also find that some people are never happy. They tell you what’s broken in the organization and what systems aren’t working, but they’re never going to be happy. “I want it to be as easy as an iPad.” “Well, I gave you an iPad.” “Oh, it’s still not easy enough.”

You can’t make that person happy. But at the same time, other people will give you great insights and ideas: “We do this one component really well. Why can’t we do the same thing for this other area?”

Woods: What are some of your best results from canvassing the staff?

Broadwater: This practice has led to some interesting data models. I have a DBA on my staff who is brilliant at data models, so we started creating what we call our Operational Data Store (ODS). It’s a data warehouse. He can look at any given system and say, “What are the components from that system that relate to these other systems, so that we can store them and then report off of them?”

One frustration is that everyone thinks that people out-of-house are smarter than people in-house. People will say, “Why don’t we hire Accenture?” And I say, “Why don’t we use the people who know the organization?”

The problem with consultants, especially when it comes to portfolio management, is they tell you about a great portfolio management tool. It’ll understand your systems, tell you what you’re missing, tell you how you do your business better, and tell you what components you don’t have. The problem is they don’t know our business.

They say, “We’ll do a discovery.” That’s fine, but you still don’t know our business. Every business is unique. We are one of the few non-profit media companies with a mixture of for-profit activities. We license toys, and that creates a really different world. So when a company comes in and tries to tell us, “These are the systems you need,” they almost always fall short.

Woods: When you describe data modeling, it’s really business modeling.

Broadwater: It is.

Woods: You have people who are creating operational data stores. They’re modeling the essential activities of the systems from which they extract data, so you can build applications and reporting capacity?

Broadwater: Correct. Our [data modeling] people are smart enough to know that we’ll need multiple dimensions and they plan to slice and dice that data in several different ways.

I also credit our finance team. We have brilliant people in finance who are very aware of business modeling and even data modeling. When we built our general ledger, our current CFO, Daryl Mintz, played a key role. He understood all the ways we would have to slice that data, so he designed budget codes to allow us to do that, even though, at the time, no systems could do it. We didn’t have Cognos, which is what we use for BI. We didn’t have a data mart; we didn’t have a data warehouse. But our CFO knew eventually we would be able to do something with it, so he built it smart from the beginning.

If you sit down with our VP of Project Finance, Maura Harway, and say, “I want to talk for five minutes about this data,” three hours later, you’re in an in-depth conversation, modeling out logic on something you might do in five years. She’s been involved in building logic structures for tech companies. Between the two of them, and many other really smart people in finance, our business modeling is very well done.

And then my team of technical advisors works to understand the business rationale and what the business wants us to do. We say, “Let’s build some data models around those objectives, build a data warehouse, and then let’s add applications and reporting and dashboards so that the business side can do business smarter.”

Woods: How does this relate to portfolio management?

Broadwater: Those areas tell us what we’re missing, the applications we’re missing, the segments we’re not reporting on correctly. They tell us which pieces of our portfolio we’re doing well, which pieces are underperforming, and which pieces we don’t know about. I think the biggest thing is the gap between what we know now that we didn’t know yesterday and between what we know now and what we still don’t know.

Woods: I see. Your plan sounds like this: If people are complaining about something, and you can’t address it through some systematic application of BI or application development, usually the reason you can’t address it that way is because you don’t have a good model of the activity. So the first object is to get a model of the activity that draws from all the sources, and then you can start describing it and tracking it. After that, you can start monitoring and analyzing, and eventually providing applications to automate it.

Broadwater: Absolutely. The game plan is, stay in touch with people about what the problems are, have a methodical approach for how we apply IT in general, and then follow that playbook.

Woods: You talked about identifying talent gaps. How does that work into your portfolio management?

Broadwater: This is something we’ve just started on. We have many people with multiple talents. Everyone at Sesame who comes from the outside will say, “Oh, so you mean you have someone who works in finance, who’s also a great singer, and could be used on the show?” The answer is “yes,” to a certain extent. My joke is that we don’t implement virtual desktops because every person in this company is a video editor.

Woods: But it sounds like that diversity of talent works for you.

Broadwater: It does. It’s almost like crowdsourcing within a company. When we do something internationally and have to proof something, instead of going out and hiring someone, we find a native speaker in the company. They may not work in the international department, but they’re already in the company.

Woods: How do you find those talents?

Broadwater: Think about Maura Harway. I know she’s got a talent for breaking down logic structures really well. Knowing that people have different talents, we’ve started to put data into a system about college degrees, college attended, and so on. My background, by the way, is anthropology. Anthropology and computers, who would have ever guessed? Then we can start cross-referencing degrees and talents and find out how to bridge some gaps through people within the organization.

Woods: But when you say, “We’ve got a gap here”—how do you identify what you need?

Broadwater: Groups just state it outright, saying “We want to hire for this project. We want a freelancer or a small contract staff, with these skills.”

Woods: There are two kinds of talent management in discussion here. One is talent management inside the company as a whole, and the other is the IT skills you want to develop. It seems like talent management was a place where you identified that your IT portfolio could grow and that you could help people do a better job of finding each other. but how do you look at the IT portfolio and say, “Look, here’s where I need more.”

Broadwater: For my staff, we keep an updated list of the skills sets of everyone on staff. Everyone on the staff puts in their skills, and we test certain aspects of those skills. If you say, “I can program in C+,” we’re going to check on that, because we’ve been burned before.

Woods: But in order to find out where your gaps are, you have to have an idea of what you want. You have to have an idea of the requirements. Where do you get that from?

Broadwater: We look at our expertise base. As a perfect example, we have expertise in C+, in Java, in PHP, Cobol, and ColdFusion. When we look at new projects, we make sure they fit into one of those areas. Every once in a while we get a worthwhile project that doesn’t play into our skill set.

For example, we have a .NET project going on right now; there’s a gap. Luckily for us, when we went back and pulled our staff, one of our developers said, “I have .NET experience, but it’s not that strong.” One of the big investments I make is putting aside money in my budget every single year so every single one of my staff gets training. And that training has to be, obviously, relevant to their job.

We’re focusing on .NET and we’ve got a guy who’s decent at it and interested in learning more, so he’s going to go and get more .NET training and he’ll be the project lead, and then we’ll bring a freelancer on to fill in any gaps. But he knows the organization, he knows our process, he knows our systems, and he can guide any freelancer, but he needs to know enough to be the project lead. So he’ll use his .NET training for us.

The implications of Automatically Created APIs.

CITO Research managing editor Deb Cameron pointed out a data governance problem described in this Forbes.com article, Box Partners With Roambi To Attack The BI Market.

Discover what mobile capabilities mean to a particular business.

Creation of VMs from a golden image can be automated, but the requirements are not easy to discover.