We recently sat down with developer Kaitlin Devine and asked her a few questions about 18F and government procurement. Not familiar with the term procurement? It basically means buying stuff. A self-described “procurement nerd,” Kaitlin has been working on a project to make the procurement process a lot more efficient within GSA. It’s called Discovery and launched in March 2015.
What do you do at 18F and why did you want to work here?
My name is Kaitlin Devine and I’m the director of engineering. I’m also a Python developer working on a few projects here. What I love most about working at 18F is that it’s very new and exciting and there’s a lot of opportunity to craft it in your own image if you can or are willing to.
What are you working on these days?
I’m working on a project called Discovery for a client called OASIS. (Learn more about OASIS here.)
How would you describe Discovery to someone who doesn’t work in the government?
Probably my elevator pitch would be that it’s an Angie’s List for government contractors.
So it’s a way for government contractors to find out information?
Discovery is a way for contracting officers to look for qualified vendors. In government, it’s important to see how many vendors there are with specific designations (such as being a small or veteran-owned business), who can perform the work that you need, and to make sure you have the minimum level of competition. If you don’t have three bids on a project, then you don’t have enough competition for it to be a competitive procurement.
So I’m working at GSA and I need to order pencils. Let’s say I need to order a whole slew of pencils for a project. So I would say “I need pencils” and would I put that on Discovery?
Discovery is actually specific to professional services.
Oh okay, so let’s say I need accountants.
Exactly. If you needed accountants, you would say “I need accountants.” And then you can restrict it even further, to different classifications. For example, you can restrict your search to small businesses or veteran-owned businesses. You can also easily see the number of contracts each vendor has formed in each business area.
Are there ways to review people?
That’s actually planned for one of our future releases of Discovery. So right now, there is a database called PPIRS (Past Performance Information Retrieval System) where contracting officers review and rate contractors who have done business with them before, and then the contractors have an avenue to dispute that, if they don’t think it was an accurate rating.
Contracting officers are supposed to look in that database before they award contracts. However, after speaking with contracting officials, they said that the performance review data was both difficult to view and use. Discovery will incorporate that data and the historical contract information that we have now to make this kind of cohesive picture of who this contractor is, what their past experience is, what things they might be good at and what things they might not be so good at, all based on their past performance reviews.
So five years ago, if I were somebody going through this process, how would I have done it then?
Something totally manual, not automated at all – very time consuming. It’s referred to as the market research process: you basically Google around for contractors or you call up contractors who had done some work for a fellow contracting officer and say, “Here’s what I need to contract out. Have you worked on this before?” It’s a very manual process and it’s one of the most time consuming parts of the procurement process.
The idea is to also potentially reduce the need to issue an RFI, or request for information. That’s basically a broad call to all vendors which says ‘This is what we’re thinking of buying. What do you think? Are you capable of delivering on this? Would you consider bidding?’ – that kind of thing. And that process itself takes two months or so, at least. So the idea is to turn this manual process into a five minute process using Discovery.
What about the project do you think shows off what 18F does, or a part of 18F?
I think it’s a great showcase for our user research and user experience designers. The client originally had one thing in mind to build, but especially since no one on the team except for myself was familiar with procurement, we ended up interviewing users about the procurement process in general and about the entire lifecycle of procurement. We then tried to zero in on what their main pain points were. And then we winnowed down on this market research aspect of procurement and did more interviews just on that. And every sprint, we’ve tried to interview someone testing the current product at the end of the sprint and get feedback and incorporate that, and consistently question previous assumptions that we had made about what the user needs. So I think it’s a really great example of this whole discovery phase before you even write one line of code.
And what did the client want initially? You said you were able to pivot the project? What did they come in asking for?
They originally wanted something called a task order authoring system. A task order is just another word for contract and task order authoring refers to the drafting of what will become the final contract. But that step comes after the whole market research phase. But when we interviewed users, they said “Hmmm.” That wasn’t their biggest problem by a longshot.
The market research is what really takes a long time and we needed the client to know that. We said ‘We’ve talked to a lot of users and this is what they need’ and added that when we talked to users, nobody was really that excited [about the task order authoring system]. Some people already have tools and said ‘We already have something for that.’ And luckily, we have a really great client who trusted us and saw the research and said ‘Alright, let’s do the [market research product].”
So you’ve mentioned the word discovery. I know you’ve touched upon it a little bit, but would you mind explaining what discovery means in this context and how it was applied to this project?
Basically we had this whole process that we did before starting on the project or coding it up. Research into procurement was our broad area of focus. First, we tried to interview a lot of people to get a snapshot of what the whole process was like, and then we did research into this particular vehicle, which caters to professional services. So we did even more discovery and more finetuning on that topic. In gaining both an understanding of the space and of user needs, we were able to say ‘This is what we need to build. And not this other thing.”
What has the feedback been from the client? What have you heard since this has been released?
We actually did a demo day for 18F and this was one of the projects we demoed and people were really, really enthusiastic about it. One of GSA’s main business lines is procurement - we procure things for other agencies - so people were really excited about it, because even in it’s initial release, it takes something that takes several hours to do and you can do it in about four minutes. So people were like ‘Oh my gosh, this is amazing.” And we got a lot of feedback and people were really interested in it, and asking for different features. It was great because the well was going dry in terms of finding users to interview, so that helped us expand the pool to find people who could give us more feedback on it. The user feedback is the linchpin of this whole process.
How many people will use this tool once it’s out in the wild?
A lot? I don’t know. Probably thousands. It’s a procurement vehicle that is accessible to anyone in government, not just the GSA.
It sounds like it can also make it easier to find potential contractors that you didn’t know about.
Right. And you get to see a lot more information about them. In the Discovery search results, we order vendors by the number of contracts completed in the industrial category the user is searching in, so their level of experience in that particular area of work is obvious at first glance.
Oh okay. Can people sort in other ways too?
Yeah, so there’s the set-aside factors – which are like veteran-owned businesses or women-owned businesses or service-disabled veteran-owned businesses. You can also see if various contractors have been suspended or debarred from doing business with the government. Contracting officials will see a big warning message in Discovery if the contractor they are looking at has been suspended or disbarred. We pull that information from another government database nightly so there’s no lag time between when their suspension is reported and when the contracting officer finds out.
Have you estimated how much money or time this will save people over a certain time span? What are your metrics for success and are those tied to revenue or savings?
We haven’t really estimated that. It would be estimating reduced labor hours contracting officers spend on this process more than anything else. But my guess is that our best metric will be adoption. In procurement, people do things in a lot of different ways and a lot of the time, it’s just based on what your mentor taught you when you came into government procurement. So I think if we can get more people to adopt it and we see increased usage over time, then that will be a really good metric.