CIO Catalysts: Introducing Brenna Berman, Founder & CEO at City Tech Collaborative

Ardito:
That’s amazing. Can you share an example of a project at City Tech Collaborative that stands out to you and encompasses that feedback loop approach?

Berman:

I think my favorite one because so many Chicagoans can relate to it, is a project we did with partners like MasterCard, Microsoft, and Ideas42 to help the CTA smooth peaks in transit congestion on their northbound train lines on the night of Cubs games.

In general, the CTA is always trying to manage the peaks and valleys of its ridership for a couple of reasons, beginning with the comfort of your ride to the actual safety of that ride. There are times when the trains get so crowded that it's actually no longer safe, never mind comfortable, not to mention those peaks in travelers actually increase the wear and tear on their infrastructure. So there's infrastructure investment, and maintenance component to this too.

When there is a night Cubs Game, which starts at 7:05 PM, not only do you have all of those north side commuters traveling north on the red and brown lines between 5:30 and 6:30, but you have all of these game day folks too, who crowd those lines.

The goal was to use an incentive model and some analytics to smooth that peak without pushing people off of public transit and into other forms of transit. We used a number of incentive models and an opt-in text-based tool to influence people's decisions to get them to travel either earlier or later if they weren't going to the game, and we were able to achieve an 18% diversion which is really high for behavioral change.

Ardito:
For many other business leaders in similar positions, they've been storing data for the better part of two or three decades but they don't know what to do with it or how to best utilize it.

What advice would you give to IT leaders who are starting their data journey? Where do they dive in? How do they start?

 

Berman:

When I was the CIO for the City of Chicago, we were building a new data analytics team, the first municipal one in the country, and the city has an incredible treasure trove of data. As you said, it's decades old, not always well structured, and very dense.

I think there are really two ways to start. The first one, which I think is the more structured business-driven approach but frankly, less fun is for the organization’s business leaders to answer key questions like:

• What business challenges is your organization trying to solve?
• Where are the pain points in your organization?
• What data analytical models can you bring to bear to address those pain points or inform them?
• What's your business strategy?
• What are the barriers to getting to that strategy, and how can data and analytics help you?
• Where are you delivering or not delivering that analytics can help?

At the city, we began to introduce data-driven policy. We were able to pull together spatial data and operational analytics to bring insights to previously hard-to-understand questions. This helped the people who were writing the policy actually see where the problems really were.

One of the first areas we worked on in the city was around food inspections. The city of Chicago inspects upwards of 24,000 restaurants a year, with only about 30 inspectors.

What we were able to do with some really interesting data coming out of the city's 311 system, which is for non-emergency services, was build a model that would predict the restaurants that were most likely to fail their inspection. This way the inspectors could prioritize where they went and positively impact the health of residents by reducing instances of food poisoning across the city.

The other approach, which is less structured but can also lead to positive impact is to dive in and play with the data and see where you start to see patterns. As your data scientists start to find patterns, they may not know why those patterns are important, they're just going to be looking for strong correlations between different variables. They can then bring that data to the right people to draw conclusions.

It is a bit like those pictures of galaxies where you can see the density of stars in certain areas, and you may not know why that's there, but there's definitely a reason to go look at those places.

In the city, we had a number of data scientists constantly mining our data. The data scientists on my team once found a really strong correlation between a number of variables in the waste management and alley management space but weren’t sure why it mattered.

We brought the information to the Department of Streets and Sanitation, and the woman who had been leading rat-baiting, which is a major service in a city like Chicago, looked at that data and it aligned with the intuition she had developed over 20 years of working. This also indicated to her that there were areas of the city that they weren't serving.

Their day-to-day operational model, without the data, was never going to lead them to every priority area of need. With that data model, we built a pilot and a model with her that completely changed their service delivery model to reach more residents.

Ardito:
One of the topics that always comes up whenever we bring up data, is privacy and security. 

How do you balance looking deep into data without intruding on people? 

 

Berman:

Cities have an interesting challenge when it comes to opting in or out because if you're doing things in the public right of way, residents can’t opt out, unless you move, which is an unfair burden. So, how you develop a public-facing data privacy policy is unique.

I think sometimes there's this idea of, "Well, we can collect X, Y, Z pieces of data. So We should collect them all even if we don't know why we need them." Except for every piece of data you collect has a price associated with it, it has to be maintained, protected, and it has to have an appropriate privacy policy. So, it might not be worth it to your business to collect every piece of data.

1. First think strategically about the data you do need, and then plan accordingly to protect that data at the level that it needs to be protected.
2. Think about your data a little bit like a customer segmentation model. Define what it needs to be useful, safe, and secure because those create the balance between what it costs to have the data and the value of using it.
3. For data that doesn’t make sense to share with the public, determine how you manage that slightly lower hurdle around privacy and security, and then decide how you apply a much higher level of privacy and security when you're dealing with PII, HIPAA data, and things like that.

Ardito:
In your role as CIO of the city of Chicago, you were instrumental in creating that first open data policy and program nationwide. 

What does it take to build something like that? Walk us through, what does that feel like? 

Berman:

I was lucky to have a really fantastic Chief Data Officer, a partner in crime who is one of the best out there.

The first thing we focused on was putting together the right team then determining the best way to structure them.

You need the right skill sets, like any other core expertise in an IT department, which now seems obvious but this was 10 years ago and data science was still relatively new, and it was incredibly new to the public sector.

We were able to build an internal data management team and it was important that we did not create a separate data science team from our more traditional data management team that had our BI and database administration team members.

It didn't make sense to me to have them separated. The data scientists can't do anything without the data and the people actually managing the structure and security of the data to serve it up to applications or models need to understand the strategy of the data science team. You may also need to make adjustments to your data management approach. So we made them one team with one general mission to leverage data, to drive value around city services.

The second part that was important to us, because we were a public sector organization, was the partnership we created with the civic tech community in Chicago to help enable and upskill our team.

I think Chicago has the strongest, most vibrant, and most generous civic tech community in the country. I say that having worked in a number of other places and, I can't tell you the number of volunteers that worked on open data projects, the number of companies who shifted their model to give back and allow their data scientists to use their skills to solve problems in their city.

We had an ongoing set of pro bono partnerships with data science teams from across the city, more than 30 companies at any given point in time, and that was critical for us when building a new team because we needed that mentoring and that upskilling of our own talent.

Then the users came next. For us, and this I think is true for any city or government, transitioning to an open data policy is scary. And when it came to those issues of transparency, and we did release data that raised all sorts of questions, we worked with the press team to have a plan in place before the data was published, to help them understand how to respond and how to use what could be considered a negative situation and turn that into a positive starting point to improve specific pain points.

We didn't have that many technical challenges because the city’s IT operations were already centralized. Typically organizations are often dealing with traditional data silos with the data being all over the place, and the real value of analytics is being able to cut across those.

Luckily, we didn't have any of those constraints. All of the city's data management had been centralized into the IT Department and the work to centralize that data management across common platforms and our progress to the cloud, all of that was underway when I took over.

So I didn't have to go and have those difficult conversations with the departments to say, "Hey, there's value in this data set but you own it." There was business ownership, but the technical ownership resided entirely within the IT Department, which was a helpful building block.

Ardito:
While you were talking about building out your open data policy, you talked a lot about cross-functional collaboration. We hear this a lot when it comes to digital initiatives, it is going to take cross-functional collaboration. 

What does success look like when it comes to cross-functional collaboration and where can leaders start with that?

Berman:

I'm not surprised that I talked about cross-collaboration a lot, that's been a hallmark of City Tech because, especially in the urban space, but in other places as well, the problems we're trying to solve now with data and other advanced technologies, is rarely a problem entirely held by one person or one organization. You can't solve that problem yourself anymore.

I think cross-collaboration comes from a couple of places:

1. You do have to have leadership from the top. I know that sounds kind of trite. It's the answer to any executive question. But we had really strong leadership from the mayor.

    

2. As we developed more pilots early on and worked our way through departments to deliver data analytics models, we learned that you not only needed the leadership from the top of the department, which for us would be the commissioner. But there's usually a mid-level manager who owns the actual service that you're working on. Their buy-in is a critical goal for two reasons:
   
   - They're going to be the ones transforming this service, and if they're not onboarded, it is never going to happen.
   
   - They are the subject matter expert of their given city service. If you can find a champion and a partner in that person who owns the service, your chances of success go up.

    

3. The third thing that I think is critical to successful collaboration is understanding the level of readiness of your partner or their organization. I think it's true in any large project where that level of readiness is going to vary depending on which department you're talking to. We actually had to come up with a way to assess readiness because our biases of where you would find data-savvy partners did not bear out.
   
   I don't think it'll surprise anyone that the Health Department was a really good partner around data projects. They’re full of epidemiologists, which are just health-focused data scientists. They were ready to go. And in fact, we often had to catch up with them.
   
   You might be surprised to learn that the team at the Department of Streets and Sanitation, was incredibly tech-savvy, and very ready for data, partially because of some history in their service delivery where they had done a lot of technology implementation because they supply the city's most expensive services, trash pickup, snow removal, street cleaning. If you can shave anything off of those, you're saving a lot of tax dollars, so they were very data-driven already.