“Over the spring and summer of 2014, New York City put in place a full-day universal pre-kindergarten (UPK) program. The blistering pace, enormous scale, and administrative complexity of this rollout were all striking: a program that did not exist when funding for it was finalized in March 2014 had put 53,250 four-year-olds in more than 1700 new full-day programs by the first day of school in September. This report provides a detailed account of the launch. It includes an extensive discussion of the city’s use of data science techniques; the city was able to combine and analyze databases in such a way that outreach teams could contact households that were likely to include four-year-olds and help interest parents sign up, all with a sharp eye for the privacy of New Yorkers. The launch as a whole combined the energy of a micro-targeted political campaign with service-oriented, street-level energy, and the lessons New York City learned in the course of this work should be useful to other cities and states.

1. Leadership is central. Given the administrative thickets that had to be hacked through in order for UPK to get off the ground in New York City on a large scale, the launch would likely have failed absent Mayor de Blasio’s frequent involvement.

2. Data systems may not be ready, but that is not a reason not to launch. New York City decided that the key interface it needed was the campaign-management function used by its outreach teams. Other data systems—for permitting, teacher recruitment, applications, and enrollment, among other elements of the rollout—were revised as needed, an approach that allowed the city to make agile, problem-solving decisions on the fly. Had the city waited until it had a perfect, bells-and-whistles system in place it likely would have missed the first day of school and might have ended up with a system ill-suited to its needs.

3. Having adequate in-house coding and data analytics resources is essential. New York is a leader in this area, but even in New York City data science may be insufficiently standardized, respected, and persistent in individual departments. Government needs more people who can write code and work with data as well as manipulate Excel spreadsheets, at all levels.

4. Mixing data science with grassroots campaign/organizing techniques and a commitment to affirmative government assistance can be a winning combination. Merely making UPK services widely available would not have worked in 2014. The de Blasio administration had to find the right households in an organized way and then help both parents and new program sites travel the bureaucratic path towards the first day of school. Hands-on, individualized assistance made the difference.”