Flexibility. If there’s one thing that everybody involved in government procurement — chief information officers, procurement officials, vendors and civic tech folk — appear to agree on, it’s that the future of government’s technology-buying should be more flexible.
The trend in that direction seems well underway too. Consider, implores longtime public-sector IT specialist Sherry Amos, the cloud.
“I’ve seen a maturity in the last five years, maturity not only of acceptance of cloud solutions but of [the] necessity of cloud solutions, not only for cost containment but because fundamentally it’s the only way to use certain solutions,” said Amos, the managing director of industry strategy for education and government at Workday.
And the cloud, even more than traditional government tech work, is a kind of symbol of flexibility. The whole software-as-a-service model is built around the concept of buying something available over the air instead of installed via disk; purchased through subscription as opposed to bought outright; updated weekly rather than released in a new version every year.
The cloud isn’t the only driver here. Government technology procurement is, in the timelines of the digital age, an ancient quagmire. The process differs from jurisdiction to jurisdiction, but generally goes something like this: Agency X has a tech need. It spends a really, really long time trying to write down its needs in a request for proposal. It requires vendors to take on as much risk as possible in the process. The vendor gets a massive check and a deadline, often a year or longer depending on the system, to deliver the goods.
Everybody has something bad to say about this approach. It takes too long. It costs too much. It favors the entrenched tech giants over innovative smaller players. It delivers products that users find difficult to navigate. It props up a paradigm of slow replacement cycles resulting in legacy systems that become more difficult to keep running over time. The list goes on.
So how are state and local governments working to create a better system?
18F was always, in its genetic matrix, meant to be disruptive. The “digital consultancy” set up within the U.S. General Services Administration in 2014, essentially co-opted talent and ideas from the private tech sector in an attempt to fix government procurement.
In California, it first took the form of a request for a new case management system from Child Welfare Services (CWS) — an important tool, since it’s the nexus through which the state checks in on tens of thousands of children suffering abuse and neglect each year. The agency took years crafting exactly the kind of RFP everybody complains about. Those familiar with the project say even the people who wrote the RFP didn’t think it would end with a good system.
Then, Code for America got involved. It took, more or less, two ideas from 18F’s way of doing things. First, it broke up the contract into pieces. Second, it retooled the vendor qualification process.
“The approach we wanted to use in the Child Welfare Services process was in embedded technology delivery, not just project and contract management,” said Dan Hon, who was at Code for America when the project got underway.
Breaking up the contract into modules opened the door for a system that could be built with continual evaluation. A big system delivered all at once may or may not have actually spent time in front of users. According to Hon, that was a central component of the CWS project.
“The state took these prototypes and put them in front of state employees and allowed them to complete a test,” he said.
Also key to the project’s success was the shortening of schedules. The team asked for vendors to rapidly develop prototypes that could be tested. That was the test for prequalification — if the prototypes performed well, it served as a foundation of trust. The state could put them in a pool with other vendors who could then pitch their bids for each modular RFP as they were released.
A funny thing happened. Some vendors thought the prototyping sprints were too fast, others thought they were too slow.
“There were some vendors who informally provided feedback saying two-week sprints would be difficult for them, and I would characterize those as traditional government software vendors. And the state saw interest from a new class of vendor they hadn’t seen before who said they would have interest in working against weekly sprints,” Hon said.
These concepts, though far from being canon among California state agencies, have already spread to at least two more projects. The Treasurer’s Office is looking for a new debt management system, and two departments charged with regulating newly legalized recreational marijuana are looking for the means to use the same licensing platform.
The Treasurer’s Office project in particular is similar to the CWS project, according to California Chief Information Officer Amy Tong.
“They were looking to replace a legacy system and instead of going for a single system integrator, they too have grouped their functionalities in smaller scope and then they sequenced those scopes based on the priority of the business,” Tong said. “So that kind of approach is similar to what Child Welfare Services is doing. The difference is instead of the state entity [becoming] an integrator managing these multiple smaller contracts … they hired another vendor that has the sole purpose of managing and integrating these various contractors.”
With the marijuana licensing project, she said, the two agencies are seeking one vendor to perform integration work instead of putting out separate bids for each.
For both that project and the treasury one, the emphasis is on speed and adaptability.
“Each contract has a much shorter timeframe from the get-go, which allows the state entities to have more flexibility to continue with the contractor if they perform well. As opposed to in the past, it’s a much longer contract, and if the contractor doesn’t perform well, we have to figure out a way to get out of the contract,” she said.
The result, according to Tong, is projects that are described in terms of months rather than years.
In Ohio, 18F’s ideas took the shape of a statewide push for data analytics tools. And the emphasis, according to Ohio CIO Stu Davis, was on revisiting the terms and conditions that make it tough for small vendors to compete for awards.
That included things like liability and insurance stipulations, as well as a requirement that vendors must have performed work for states similar in size to Ohio — of which there are about eight.
“We still have some of those [requirements] in there, but it’s pretty broad-brushed,” Davis said.
The general idea is to open up the process so smaller groups with a broader range of perspectives and ideas can pitch suggestions for how to solve problems. It also means involving procurement officials in the process earlier on, and opening up people’s minds about what kinds of solutions might work rather than asking for a specific solution.
“It costs money for vendors to submit these proposals. ... We need to do our due diligence and make sure our criteria is as tight as possible and we don’t disqualify them for crazy ideas,” he said.
Another possible new way of doing procurement is actually an old one: cooperative purchasing.
That’s when a group of government agencies, working through a central group such as the National Association of State Procurement Officers (NASPO), all jump in on the same RFP. One agency takes the lead, with others sitting in to ensure they agree with the terms, and then once the RFP is finalized the participating entities can all take advantage of the process. It’s economies of scale at the government procurement level.
It's just that now, it appears more governments are turning to the process. And the cooperative purchases may be getting bolder.
For example, there was the NASPO ValuePoint cooperative RFP for cloud services launched in September 2016. It was a big blanket, covering infrastructure as a service, platform as a service and software as a service — vendors were put into those categories, many of them in two or all three — and 35 states signed on. So far, there are 29 vendors in the pool.
In a way, it’s not far off from the prequalification idea underway with the California CWS project.
“It’s like a convenience store for the companies that have already been awarded these services,” said Tony Encinias, vice president of technology at ViON and the former CIO of Pennsylvania.
Amos and Encinias both said they think government is starting to use cooperative purchasing more. “The last couple of years, a number of states really started to understand that they can’t procure modern technology and cloud-based technologies the way they used to,” Amos said.
Steve Emanuel, public-sector adviser for Alliant Technologies and the former CIO of New Jersey, agrees. And he sees the change as a matter of necessity. Cloud purchasing doesn’t make sense under the terms of traditional procurement — take unlimited liability, for example. A company that hands off a disk to a customer can take liability. A customer that provides a service tied into other services, running off a third-party platform and storing data on an enormous network of servers is another matter entirely.
“How can you hold the vendor at the other end of the Internet accountable for [distributed denial of service] attacks? And not just to them, but to other people,” Emanuel said.
There are several other ideas for fixing procurement making the rounds in state and local government.
Another one from 18F that might have potential, especially at the local level, is micropurchasing. 18F has developed a program for putting out small programming requests — not very much work, and at low enough cost that it can be charged to a card. So small it doesn’t have to go out to bid.
At the local level, there’s already a mechanism in place that works something like that — at least, insofar as it avoids the need to go out to bid or authorize big payments. That would be pricing services so they come in under the RFP-triggering threshold.
That was the strategy of Binti, a startup that came out of San Francisco’s Startup in Residence program. The company developed software to help foster care agencies process people applying to house children. Then it priced the software so that most counties in California wouldn’t need to go out to bid to buy it. Within five months, 19 counties signed on as customers.
Startup in Residence is, itself, a way to find new solutions to problems and work faster to implement them. The program, which is spreading outside the San Francisco Bay Area, involves bringing in entrepreneurs to work closely with city officials to solve problems.
Somewhat akin to that idea is the Citymart model. Citymart, a startup aimed at changing procurement, seeks to center the process of buying technology around problems — putting the emphasis on what needs to happen, then getting creative in finding possible solutions.
“This is a kind of situation we find over and over again, that you actually realize that you’ve just been renewing and adding more and more stuff to a problem instead of really thinking, ‘Can this be done in a different way?’” said Sascha Haselmayer, CEO of the company.
Take, for example, a street lighting purchase. “It’s not that you don’t know the 30 vendors who can give you a street light, it’s that you don’t know the 30 alternatives to the street light,” Haselmayer said.
It’s far from certain what the future of procurement will look like. But if it’s generally moving in the direction of greater flexibility, there are some guesses for what that will mean for vendors.
For one — increased competition. The NASPO cloud contract has a big pool of vendors jockeying for business. The California vendor prequalification scheme calls for faster development on smaller projects, ultimately meaning more projects to work on.
And it means that the big companies, often accustomed to moving slower, will have to keep up with the little ones.
“You can never be too comfortable, as an incumbent, if you aren’t continuing to innovate on some level,” Amos said.
That will likely mean an increased emphasis on user-centered design and agile development processes.
“It is my understanding that during the [California CWS] evaluation process, some prototypes failed because users were not able to complete the tasks based on the design of those prototypes,” Hon said. “And when we come back to a high-level strategic vision for the quality and the goals of government technology, then when we say we need to get the job done, getting the job done also means the software actually allows end users to complete the tasks they have to complete.”
It also means technology development processes might be a little more fluid overall. That is, vendors might have to work with each other more, they might have to respond more to customers calling for help — even if it’s on part of the system another vendor developed.
“You have to know where the limits are and know that there are some gray areas and know that the companies that go that extra mile in the gray area are going to be the ones that are touted as more flexible — and the ones that more people want to do business with,” Emanuel said.
Most people involved with these ideas think that whether or not they spread — or even become the norm in future government tech procurement — relies on how successful they are. Still, change is hard.
“If [the California CWS] demonstrator is successful, it will stand as a good example and the intent, from my external point of view, is that it would serve as a brilliant case study for other states and other governments,” Hon said. “But ultimately [it’s] harder than saying, ‘We want to do things in that way’ because it requires not only organizational change management, but it requires rather hard work in looking at the procurement environment.”
There is some indication that they are already spreading — after all, people from Indiana and Oklahoma have reached out to Ohio for more information about its data analytics project, according to Davis. California’s modular, agile approach seems to be making its way from department to department. And some think procurement change is inevitable when it comes to new tech because some new tech is incompatible with the old processes.
And then there’s the sheer complexity of some of the problems government is trying to solve. Technology, especially with potential disruptors like machine learning and blockchain, offers government the potential to navigate such tangled forests as opioid addiction, infant mortality and homelessness — but only if the government can access it and use it effectively.
“We have to look at this completely differently from how we have in the past,” Davis said.