In the vigorous competition of the free market, online platforms have made innovation into an operational discipline. For these firms, being data-driven isn’t a catchphrase, it’s a foundational attitude. The performance of every service, every feature, every display, every transaction, every interaction is automatically tested for objectively superior performance against alternatives.
As a consequence, online experiences raise consumer expectations for superior customer service every day. The President’s Management Agenda calls on Federal Agencies to meet those expectations. As a former Federal Chief Data Officer (CDO), I am confident that we can meet those expectations by transplanting commercial sector best practices for AI and data engineering into Federal Agencies. In this pair of blog articles, I’ll share some of those best practices.
“…the Federal Government must center its services around those who use them—delivering simple, secure, effective, equitable, and responsive solutions for all who the Government serves.”
Commercial sector firms have long practiced the art and science of personalization. They learn your wants, needs and preferences. They build flexibility into their offerings to they can be personalized to you. Consider your Netflix¹ experience. Consider the recommendations Amazon.com and Grub Hub offer you. Consider the ads that pop up in Google Maps when you’re on the road to a destination. These aren’t just personal. They’re also proactive. Google Maps knows what you like and where you are. They push restaurant and gas station suggestions to you without waiting for you to pull or ask for them.
How do commercial sector firm accomplish personalized push? The short answer is recommender engines. Recommender engines use a variety of Artificial Intelligence techniques to identify individuals with similar wants, needs and interests. Recommender engines then do the people who bought X also bought Y thing we’re all familiar with. But the Netflix example is more subtle. Netflix knows that people like you also like a video you haven’t seen yet so they place it prominently on the screen for you. (Figure 1)
Figure 1 – recommender engines
So what does personalized push have to do with the government? Plenty! A recent study² revealed that 87% of citizens expect the government to proactively engage them. Let’s consider a few examples:
- New Zealand citizens don’t file their own taxes. The New Zealand tax authority, the Inland Revenue, generates tax statements and sends them to citizens.
- Singapore’s government knows when a citizen’s passport is going to expire, and automatically sends new passports to the citizen.
- Taiwan’s fourth e-government strategy includes a commitment to proactively and seamlessly delivering just-in-time services to citizens shaped around their individual needs, preferences, circumstance, and location.
Without the personalized push, citizens have to be aware of all of the potential things government might do for them and know which Agency to approach, know what services to ask for, and then know how and where to find the systems they can use to submit requests for those services³. This is usually a frustrating experience. (Figure 2)
Figure 2 – citizen’s journey
Now let’s bring recommender engines into government. Here’s how they’d work. Identify people with similar wants needs and interests based on demographics – AND – based on the services they requested from the government. Look for the gaps where a citizen might benefit from service but hasn’t requested it yet. Proactively offer that government service to the citizen. (Figure 3) This is a personalized push in government.
Figure 3 – recommender engines in government
AI solutions, like recommender engines, give us a tool for centering Federal Government services around those who use them, delivering simple, secure, effective, equitable, and responsive solutions for all who the Government serves. These commercial sector approaches enable us to improve how we design, build, and manage Government service delivery for key life experiences that cut across Federal agencies.
Recommender engines are also useful for improving employee productivity in back-of-house operations. Every large organization on this planet has a complex portfolio of IT systems. Employees switch between different screens on average 566 time per day4. That context-switching has a cost. When people shift away from a task they’re working on it takes an average of 25 minutes and 26 seconds to get back on the original task . That’s lost productivity. Government service is no exception to this pattern. For example, the DoD has “roughly ten thousand operational systems”.5 . Recommender engines can offset this context-switching by keeping track of what an employee is doing before the interruption then placing them back into that task after the interruption. Or preventing the interruption in the first place.
There’s a lot to like about recommender engines!
In the next article in this series, we’ll “take a peek under the hood” and see how recommender engines work.