A playbook for data-driven innovation in cities

Photo by Henning Witzel on Unsplash

City governments have an abundance of data, but its real value could be lost in silos.

Big data and machine learning in housing could, for instance, help us predict which properties are likely to be blighted or which are at higher risk of catching fire. Then, resources could be deployed tactically to prevent rather than react when it’s too late.

Given the wide amount of services that local governments provide, there is a green field of potential innovations using data.

Analytics teams in cities, often headed by someone with a title such as Chief Data Officer, Chief Analytics Officer, or Chief Performance Officer, have been tasked with finding those opportunities and addressing some of the toughest urban problems by tapping into the disciplines of data science and operations research, as Craig Campbell and Stephen Goldsmith point out in Smarter New York City: How City Agencies Innovate.

But there are important questions to answer when applying analytics in local governments.

Over the past year, I had the pleasure of working with Amen Ra Mashariki, who spent time heading the Mayor’s Office of Data Analytics in New York City, and David Eaves, a lecturer of digital government at Harvard Kennedy School, to write a practical guide aimed at senior executives in city government and the practitioners that may end up running new analytics team, to highlight the principles, plays, frameworks and questions that they should consider.

The Analytics Playbook for Cities is meant to serve as a navigational tool for understanding the steps for an effective implementation and confronting trade-offs effectively.

The big challenges for analytics in cities

Below are some of the big questions that we decided to address:

• What is the best way to structure the team in relation to the larger organization? Should it be located under the Mayor’s Office or closer to IT?
• Who do you hire? What skills should you be looking for?
• How do you make the case for more funding?
• What is the right way to find and prioritize projects?
• How much time should an analytics team be spending on building the right digital infrastructure versus making progress on projects?
• How do you make sure that privacy of citizens is respected?
• What new dangers in the realm of cybersecurity should be considered?
• What should the city do to ensure that algorithms are not biased or that they do not perpetuate inequities?

Principles for an effective analytics team

Through our research, which led us to studying and speaking with some of the leading cities doing analytics in government, we found that there is no single strategy or set of steps that can be applicable everywhere. However, we were able to outline some general principles for effectiveness that we would share with aspiring chief data officers:

• Commit to better service delivery: It’s about improving something concrete that is meaningful to citizens.
• Be both the disruptor and the listener: You need to learn to simultaneously be an entrepreneur that challenges the organization, and also an empathic listener of both users and the employees who have been tackling this problem for some time.
• Start with the problem, not the solution: Resist the urge to apply the coolest technology or the latest framework.
• Ensure executive buy-in: You will need cover when you face resistance. And you will face resistance.
• Work in the open: Your work as an analytics team should be subject to scrutiny. Blog about your work, put your code online, invite feedback.
• Find your allies: No analytic solution will happen in the abstract. You will need to be skillful in working with other departments without an analytic background, collaborating with academia, and leveraging the press.
• Quickly deliver value, but think long term: Those quick wins will be important for cementing your position in the short-term, but know that long-term investments will create even greater analytics capacity.

We would love some feedback

Both the principles and the challenges, as well as the plays to address these are further explored in our playbook. We have been working on a Beta version, which you can read here: https://docs.google.com/document/d/1jI_9zbgK0F6lQh3xB2oHV5uFUqLzs4fXjeDfFqoDqnU/edit?usp=sharing

The document is open for comments, and we would appreciate any feedback.