Recruiting for the Mission

Every government software leader we talk to says a version of the same thing: we can't get the people. The people we want rarely apply. If by some chance they do, the process loses almost all of them. And for the very few we do manage to get, they are gone in twelve to eighteen months.

It reads like a labor market problem but it isn't one. The government programs that broke through–and they do exist–hired from the same market, at the same pay scales, under the same rules as everybody else.

They treated recruiting as a marketing and sales function, not an HR function. The people you want already have jobs. They aren't browsing USAJOBS. They will never see your announcement, and if they did, nothing on it would move them.

This paper is about doing that job. Give them something worth joining. Go direct. Leverage network effects. Recruit value stream engineers before technical ICs. Better yet: train the good people you already have first. Then hack the bureaucracy to actually hire your gaps.

And the part that gets skipped… You can't recruit without a story. You can't tell a credible story without something shipped. You can't ship without a small team, and you can't hire that team until somebody in the building knows what good looks like. This is why the first people you should develop are ones you already have. The order matters.

We have a bias for action, so every section ends with a "go do."

Diagnosis

You have money and authority. You have a mission that matters. And yet you still can't get the right people in the door. The ones you do get often aren't the ones you wanted, and your best ones leave inside a year. Three failure modes account for most of it.

Failure mode one: the wrong brand

Nobody dreams of joining the [Agency] Office of Modernization. The name on your announcement is doing recruiting work whether you meant it to or not, and right now it's working against you.

Failure mode two: the wrong roles

The instinct is to hire technical ICs. But technical ICs don't know the mission or the bureaucracy. They’re extremely proficient with technology and creating outputs. But you need outcomes, in production. That requires a deep understanding of the mission value stream and hacking the bureaucracy to enable delivery. Start with value stream engineers. Consider training your existing staff. Fix the IT value stream first. Then define mission outcomes. Then hire technical ICs. You will end up needing less of them and retaining them longer.

Failure mode three: the wrong funnel

USAJOBS only sorts the people who already found you; it does not go looking. The people you want are not looking for jobs there. They are also gainfully employed, and won't fight through a self-assessment questionnaire for a job title they don't recognize. You have to do outbound and create inbound.

These three compound

A weak brand and a passive funnel give you a shallow pool. A shallow pool gives you marginal hires. Marginal hires and missing roles give you slow delivery. Slow delivery burns out the few strong people you did land, they leave, and next year's recruiting is harder.

Go do

Before changing anything, baseline four numbers:

Candidates per opening — and how many were actually qualified.

On average, we receive over 200 applications per job posting, and about 5% of those are qualified. It's equally important to note that only about 30% of our hires come from those applicants. Roughly 40% come through our referral network, and the remaining 30% are candidates we identify through cold outreach, many of whom we've been building relationships with for months before the position ever existed.

Offer acceptance rate — and the stated reason for every declination.

A good offer acceptance rate is between 70 and 90%, depending on the level of the position. Most technical or senior roles should land between 85 and 90%. Rise8's offer acceptance rate is over 94% on a rolling 12-month average. That is because every touchpoint with an A Player is treated as a high-value conversation. We talk constantly about our Mission, prioritize candidate experience, and stay transparent about our decision-making process. You can't expect to attract A Players if you give them a B-quality interview experience.

Time from announcement to start date — measured to the first day of work, not to the tentative offer.

This is where reactive recruiting gets punished. You don't lose a hire at the offer stage, you lose it months earlier, when nobody was building a relationship with the person you'd eventually need. From the moment an opportunity is identified to the candidate's first day, you should target 45 days, and that number only works if you’re never starting from zero. The candidates who fill that window are people you've been talking to for months, sometimes longer, well before the role existed. That's the whole idea behind proactive recruiting: the pipeline isn't built in response to an opening, it's already there when the opening happens. Because of this process, Rise8 beats the 45-day target, filling new delivery roles in under 30 days of the opportunity being announced. We do this by maintaining a constant fully vetted candidate pipeline of 120% relative to the open positions we have projected over the next two months.

Twelve-month retention — for everyone hired into the program in the last three years.

The best way to ensure retention is to hire A Players who are aligned to your culture and believe in your mission, including your leadership team. McKinsey's research on why people quit found the top three reasons were feeling undervalued by the organization, feeling undervalued by their manager, and lacking a sense of belonging at work, each cited by roughly half of the employees surveyed. Leadership and culture aren't soft factors here, they're the two biggest levers you actually have.

These take an afternoon to pull and they will become the scoreboard for everything that follows. You cannot hack a process you have not measured.

Give them something worth joining

You can't reliably recruit talent to a government agency. Sorry. You just can’t. But you can recruit to a named organization inside it. Sub-branding is a lever that is actually yours to pull. Your agency brand is uninspiring and messaging is slow. You can't fix it and you shouldn't try. A sub-brand is yours to define and as fast as you want.

When I was doing this inside USAF, "Air Force Life Cycle Management Center, AOC Weapon System Modernization" wasn't going to attract the talent we needed. But “Kessel Run” enlisted a revolution. Same mission. Same authorities. Same pay scale. The difference was a name, a logo, a story, a workplace, and leaders willing to go direct.

Tell the story at level three. Every organization's story has three levels, and they're all true at the same time. According to Lulu Cheng Meservey at Rostra, the three levels are:

• what you're literally doing. "We're building a rocket."
• what it adds up to. "We're bringing American spaceflight back from the private sector."
• the dent in the universe if it works. "We're making humanity a multiplanetary species."

Ask a Boeing engineer what they do and you get "some space capsule." Ask a SpaceX engineer and you get "we're making life multiplanetary." It’s the same job but very different stories.

This matters because the level you use determines the talent you get and how hard they work. The people you actually want—the ones who will take a pay cut, fight bureaucracy for two years, and ship amazing software for your mission—only show up for level three.

The good news is that you have a huge advantage. Startup founders have to create this. You already have it. Your work sits on top of national security, veteran care, health care, public health, etc… all level three stories.

But when someone asks what your program does, they get level one: we're modernizing a case management system. You're sitting on the best recruiting asset in the world and describing it like a line item. Write all three levels down. Then never say level one out loud again unless someone asks.

The brand has to sit on top of something real. Don't rename the program office. Pick one mission pain that's actually bleeding, solve it fast in a way that works on the others too, and brand that. A name with no shipped work behind it is a logo on a PowerPoint, and people figure that out quickly.

Take a side, and pick the side carefully. Kessel Run's rebel alliance edginess was on purpose. It was a beacon for all the rebels we were looking for. It also made a certain kind of person say "that's not for me," and that was the point.

The trick is where you draw the line. Draw it so it pulls other employees in instead of cutting through them. No us versus them. Some version of: we believe this mission deserves software that works, against people who are fine with how it's been. Nobody in government should choose the latter, and the ones that do aren’t there for the mission. Say the thing your future hires already believe and nobody with a title will say out loud: that government software is broken and it doesn't have to be. You'll have their attention before you've described a single job.

Promise something verifiable. Not "we're transforming procurement." Code in users' hands early and often (lead time and deployment frequency). Software that improves the mission value stream (mission metrics, stability, reliability, security) and that users love (NPS). Keep those brand promises and recruiting, among other things, becomes easy.

Pick one line and repeat it. Lulu points out that Steve Jobs gave what many consider the best commencement speech of the century and people remember one line from it: “Stay hungry, stay foolish”. You get one line too. Pick it, change the stories and examples around it constantly, and never change the line.

Go do

Charter the initiative on one page:

• The name. Short, memorable, not an acronym, not a description of the org chart.
• The three levels. Write all three. Test level three on someone who doesn't work for you and watch their face.
• The line. Who's on your side, who isn't, drawn so it doesn't split your own people.
• The phrase. The one line you'll repeat until you're sick of hearing yourself say it.
• The first win. What ships in 90 days and who's using it.
• The brand promise. Three things a new hire can hold you to.

Some swag won’t hurt, either ;-)

Go direct: you are the recruiter

When you’re just getting started the only thing to join is you and your story. Congratulations, you’re a cult leader of one. That's uncomfortable if you came up through the government, where the job is to stay heads down doing the work and let public affairs handle the talking. You can’t outsource you and you shouldn’t outsource your story.

The message loses steam at every handoff. You tell your PAO. The PAO writes it up. It goes through review. Someone “smooths the edges”. What comes out the other end is technically accurate and reads like nobody wrote it. The conviction was lost. Go direct.

You don't need charisma. This works better if you're a natural performer, but it doesn't require one. Two things aren't optional. The idea has to be coherent enough that you can say it clearly in one breath. And you have to have conviction. Which means the first person you have to convince is you. If you can't say why this matters to you—not to the department, to you—nothing spreads.

Message, then medium, then messenger. In that order.

Message is where the leverage is, and where almost nobody spends their time. If the messaging sucks, nothing will save you. Your message doesn't need to be profound. It needs to name something your future hires already feel and can't put words to.

Medium is where those specific people already pay attention, which is not a govcon conference or the Washington Post. If you’re looking for the best software engineers, designers, and product managers, distribute your message where they are most dense.

Messenger isn’t always you… different claims need different mouths.

• Vision and intent: you, first person, and only you. Nobody else can say it as well.
• "This is a great place to work": your ICs, never you or your leads.
• "The software actually works": the user. Bring the analyst who got their Saturdays back and let them say it in their own words.
• "This program is real": the shipped thing itself, in public. Not a quote about it.

Watch your ship-to-talk ratio. Elon posts a hundred times a day and nobody doubts the rockets fly. But a leader who posts constantly with nothing shipped loses credibility immediately. Resist the urge to “fake it ‘till you make it”, it backfires every time. Ship the wins first. Then yap about the wins. Rule: never talk about the future in the present tense.

Attention is fuel, not the finish line. If something lands and the only place to send people is a USAJOBS page, you burned the one shot you had. Build the destination before the moment. A real page that says who you are and what you're doing, an open announcement or a way for someone to raise their hand, and a human being who answers within 48 hours. Be ready to catch the demand.

Now the part you're going to flinch at. You're a federal employee. You have a PAO, an ethics office, and a chain of command, and none of that is going away. Going direct doesn't mean going around them; it is being the one who speaks, in your own words. Two things that make this work:

• Get agreement from ethics and counsel before you start, in writing. Don’t ask. Tell them what you are going to do and why it’s ethical and legal.
• Recruit your PAO onto the team instead of treating them as a gate. Same play we run on contracting and compliance. Bring them in on day one, make the right thing the easy thing, and let them see the mission impact. The promotions that come later don’t hurt, either.

Be yourself, louder. Don’t try to be vanilla to avoid the hate, or someone else to avoid the work. Turn up the volume on who you already are. Over time your audience sorts itself into the people who actually align with you.

Lulu puts the job description more bluntly: you are starting a cult, and you signed up to be the one at the head of it. You set the vision and you take the arrows aimed at everyone else. The good news is it stops being all on you faster than you think. Eventually the first believers recruit the next ones for you.

Go do

• Work the chain backward. What do I want (hire six VSEs) → what must people do (apply) → who specifically (name the forty people or the exact pocket they live in) → what must they believe to do it. That last one is the hard one, and it's where your actual content comes from.
• Build the destination. Page live, application path open, one named human answering within two days.
• Ship, then speak. First post in your own name within 30 days of the first thing reaching a real user. Not before.
• Book two meetings this month. Ethics and counsel for your guardrails. Your PAO to bring them into the team.

The network effect

Once you've got something worth joining and you're the one talking about it, you stop being the only recruiter. It compounds in two directions, one of which you might not expect.

Your people recruit for you

Your first ten believers bring the next ten. A senior engineer trusts a friend's read on a job over any post you'll ever write.

Two things get in the way, and both are your fault.

• Your team doesn’t know what you’re hiring for. Make the list visible and the PDs specific.
• You overscript it. Let them pitch it. They know their network. You don’t. PA definitely doesn’t.

Contractors are hires, too

Be careful who you hire. B players attract C Players and this is true of contractors as well. The good news is that when you do the first parts right…

You get more bidders. A vanilla program office attracts traditional govcons. Yours will too, but it will also attract non-traditionals.

You get their A-team. They want to put their best foot forward. Also, their best people have some say in where they go and want to work on programs where they'll ship something that matters

They market you for free. Contractors have marketing budgets, conference booths, and channels you don't have. When they win work with you, they publicize it. When they ship outcomes, they publicize it. Other contractors see it, and so does talent. Your brand rides along for free. Make sure you choose one you want to ride along with.

Two things keep it compounding

Be seen. You can't build a network from your desk. Start by going to the two or three conferences where the people you want actually are (e.g. not the federal trade shows). Send your engineers, not just you. Your engineer in a hallway saying "yeah, we deploy weekly to a million users and save lives" is worth as much as you giving a great keynote. See Col Oti’s talk at SpringOne Conference. We had engineers from big tech ask if they could join (we weren’t ready to capture the demand, unfortunately). The following year, we were more prepared!

Then host, which is even higher leverage. Work with an influencer or major player in the space. You pick the guest list, you own the room, they get them there. Forty of the right people beats a booth at any trade show.

Be worth showing up to. None of this survives contact with a bad first impression. A candidate figures out what kind of place you are in the first ten minutes. A six-year-old laptop, no admin rights, GitHub blocked, six weeks to get an account, everything through a ticket. They've already decided, and no amount of level three story fixes it. Contractors too.

This fix is cheap, and it gets deferred every time, because nobody gets promoted for it. Fix it anyway, then show it off. Photos of the space, the boards on the wall, the deploy screen. A re-engineered onboarding value stream. People need to see it's real.

Go do

• Tell your team what you're hiring for. Visible list. Clear PDs. Let them loose.
• Send yourself and two people to a technical conference each quarter with the goal of recruiting. Fund it now, before it becomes the thing that gets cut.
• Book one hosted event within six months. Work with an influencer in the space that can get the people you want in the room. This can’t be another boring industry day.
• Walk your own onboarding. Write down every roadblock a new person hits before they become fully mission capable. Start fixing it.

Hire value stream engineers before technical ICs

The instinct in a program office trying to fix govtech is to hire technical ICs. It feels like the right move, but it produces more problems than outcomes.

Own the mission value stream. A contractor can bring you delivery capacity all day long. No contractor can bring you your own understanding of your mission, and no contractor has an incentive to tell you the thing you asked for was the wrong thing or to choose the harder, right thing.

The capability you're missing is people who can map your value stream, find the constraint, define outcomes, and hold everyone to real mission impact. FAANG ICs are not it.

Hiring technical ICs first can cost you. When technical ICs sit on top of the work, they start clashing with the contractor who is the one on the hook when delivery fails. Over time the government slides into the integrator role and the contractor becomes staff augmentation with no real accountability. You've taken on all the risk and given away all the leverage.

What a value stream engineer actually does. A VSE knows the “shop floor”. They walk it. They know the value stream start to finish, with real mission metrics, not IT metrics, including the four handoffs nobody documented and the spreadsheet somebody maintains at home. They find the constraints where the work actually piles up. They figure out the change in user or system behavior needed to break the constraint. Then they hold the delivery team, government or contractor, to that change instead of a product specification.

Ask your team what they shipped last quarter. The one who tells you about a change in system or user behavior that resulted in an improvement to the mission value stream… that’s your first VSE candidate. Get them some formal training to fill in any skill gaps and put them on your first problem.

Know the difference between an output and an outcome. An output is what got built. An outcome is what people (or machines) do with it. An impact is what that's worth to the mission.

Picture an analyst who spends her morning calling six offices one at a time to coordinate a request. The output is an app that sends those requests automatically. The outcome is that she stops making the calls and works a notification queue instead. The impact is that the whole process now takes two hours instead of two days.

Outputs only matter when they cause outcomes, and outcomes only happen in production. Everything before that is potential. Technical ICs out of FAANG are very good at shipping outputs. Few of them have ever had to produce outcomes in a mission value stream with no P&L and a mountain of bureaucracy. You need outcome-producers first.

You do need real technical depth. Just less of it than you think, and pointed somewhere different. Two or three highly talented technical ICs per value stream, and their job is not building; it is guiding the value stream portfolio. The product lead makes sure everything is aligned to the big picture you’ve laid out. The engineering lead uses domain driven design to guide the architecture to meet nonfunctional requirements and avoid rework and tech debt. The UX lead to focus on the holistic user journey. They can also help shape acquisition strategy and evaluate contractors pre- and post-contract. One potential exception: a small in-house team to explore problem areas you aren't ready to put on contract.

Write scorecards for each position. A scorecard should work backwards from the mission impact they will be assigned (which should be a 12-24 month goal). It should be an impact map: the list of outcomes, outputs, and activities that will be expected of them to achieve the impact. This helps you interview. But it’s also a magnet for high performers who want to be measured this way, and reinforces the brand.

Go do

• Look for VSE candidates who already work for you. Train as needed. Hire as a last resort.
• Write scorecards for your first five hires. Activities>Outputs>Outcomes>Impact(s) over 12 to 24 months. If you can't write the impact and outcomes, you don't yet know why you're hiring the position.
• Ask every candidate the same question: tell me why you got paid. They should give you a scorecard. Ask them how they did it. The ones who can walk you through the exact resources > activities > outputs > outcomes from zero to impact are the ones you want. If they hacked some bureaucracy along the way, move them to the top of the stack.

Train before you hire

You can't hire for a skill nobody in your building has ever practiced. And as we’ve already explained, most of the technical ICs out there don’t have the domain expertise or bureaucracy hacking experience you desperately need.

Think about what happens when you post a Staff SWE job today. Who reviews the résumés? Who interviews? Who reviews their Github? Who sits on the panel? You’re in a Catch-22. You need a Staff SWE, but you need a Staff SWE to hire one. Most gov orgs select on what they can judge. Keywords. Certifications. Confidence. You hire the best talker in the pile and find out eighteen months later.

You already have the people. In the early 1980s, GM's Fremont plant was the worst car factory in America. Workers drank on the line. Some of them welded Coke bottles inside door panels so the car would rattle and annoy whoever bought it. GM shut it down and fired everyone.

Two years later in a joint-venture with GM called NUMMI, Toyota reopened the same plant, and insisted on rehiring most of the same people GM had written off. GM thought they were crazy. Within months that workforce was building the highest-quality cars in the country. Toyota didn't lecture them about Japanese culture and philosophy. They taught them a different way to work. When people saw the results of working in a different way, it changed their values and attitudes which in turn changed the culture. The problem was never the people. It was the system. The remedy was changing behavior to change thinking.

Traditional training doesn't work. Send 10 people to a two-day agile course and you get ten people with the same jobs and a new vocabulary. Nothing changes. Send 10 people to ship real code to real users in real production environments alongside experts and you get behavior change, upskilling, and culture change. Consider a contract that includes both delivery and enablement, where your employees can pair with a best in class delivery team on real delivery. Paired product management. Paired design. Paired programming. A GS-13 who spends a quarter watching real users struggle with the thing they wrote requirements for does not go back to writing requirements the old way. Nobody has to convince them. I would consider the Kessel Run contract with Pivotal Labs, where we got this paired enablement, to be the biggest single reason we succeeded in delivery and in talent growth and retention. It was very similar to the NUMMI story. But they fell short in that while they were some of the best software delivery teams in the world and focused on enablement, they did not have the government expertise we needed. Contractors that have both modern software savvy and government expertise and are experienced in paired enablement are exceedingly rare. I know of only one besides Rise8. If you read this and you know one, reach out and let us know who they are!

Pick the right first cohort. Five to eight people. You already know who they are. Innovators and early adopters. Those are your people. Protect them, give them air cover, and clear their other duties. Half-time on the new team and half-time on their old job means full-time on their old job.

This pays in four ways. You get the wins you need to recruit. You now know what good looks like so you can interview better. You have somewhere to put an external hire where they'll be productive in week one instead of month six. Existing employees don’t feel threatened with replacement.

Be honest that some won't make it. Not everyone wants this. Some people liked the old job, and a few will actively work against the new one. Decide in advance what happens to the people who opt out, so you have a plan instead of a mess.

Go do

• Name five to eight people this week. You already have the list in your head.
• Clear their plates. Full-time on the first team. No exceptions.
• Pair them, don't train them. Choose a contractor that pairs daily as a practice, not one that just says yes to it. Every government person pairs 1:1 with the contractor full time shipping real outcomes to prod. No observers. Hands-on. Six months at least.
• Take them all to see the user. Whatever else slips, this one doesn't. It's the thing that flips people.
• Put two of them on your hiring panel at the end of the first quarter.

Hack the hiring process

Now that you have demand, you have to hack the hiring bureaucracy. This isn’t about rule breaking, or rule reform. The rule you're blaming is almost never the thing stopping you. In my experience two barriers do the real damage. The first is that you're using the wrong rule and don't know a better one exists. The second is that the rule is fine and the process wrapped around it is broken. Understand the system and use it. Map the value stream and engineer a faster process. That’s bureaucracy hacking.

Eric Schmidt made the same point to the House Armed Services Committee. He told them the flexibility to move fast already existed and the real problem was that too few people knew they were there or bothered to use them. He credited DIU's speed to its organizational design and the autonomy of its leaders more than the authorities themselves. The authority is rarely the bottleneck. Repeat that to yourself on the way into this fight.

Barrier one: you're standing at the wrong door. When we started hiring software engineers at Kessel Run, we did the obvious thing and went to the engineering functional leadership who managed all engineering labor categories. We got direct hire authority and thought we had solved the problem. It came with an entire bureaucratic apparatus that ran all the way up the command structure. Every piece of it was added for a “good reason”, I’m sure, but in sum led to bad outcomes.

Take our first interview panel. The senior engineering decision-makers all got seats on it, the questions had been handed to the candidate in advance, and we sat in a conference room on an audio-only line. They couldn't see our faces and we couldn't see theirs. The lead said, "Hello, thanks for joining. You've got 30 minutes." Then the candidate talked, uninterrupted, for 30 minutes. No follow-ups, because a follow-up might introduce bias. That was the entire process. Every constraint on it had been put there by a well-meaning person solving a real problem, and the sum of all those solutions was a system that could not tell a great engineer from a talking wall.

We also hit a hard requirement in that category: an engineering degree. This was right about when we were realizing that some of the best coders we'd ever met didn't have one.

So we went looking for another door, and there was one. A general IT labor category (GS-2210). Same institution, different rules: no degree requirement, and none of the panel theater. We found the person who owned it, told them plainly what we were trying to do, and worked through it with them. Nothing we did was outside the rules. We just stopped accepting that the first category we tried was the only one that existed.

The lesson isn't "labor categories." Yours will be different, and by the time you read this the specifics will have moved. The lesson is that the wall you're staring at almost always has a door in it you haven't looked for. The fastest way to find it is to go talk, like a human, to the person who owns the rule.

And notice what the old rules were actually for. The degree requirement, the reassignment logic, all of it was correct for what the traditional Air Force needed: people it could move into any role across a 20-year career. It was wrong for us, because we were hiring people for a two- or three-year tour to do one specific, hard thing. We didn't beat the requirement; we matched the authority to the actual need. That's the move, and it's almost always available.

While you're in there, fix the assessment, not just the authority. That audio-only monologue wasn't a rule problem. It was an assessment-design problem. The fix is boring and it works: subject-matter experts write the evaluation and run a structured screening instead of an HR keyword match or a panel built to survive an appeal. That single change does more to stop good people from being screened out than any authority you'll pick.

Barrier two: the rule is fine and the value stream is broken. Here's the part that surprised me. Once we'd solved the above, hiring was still painfully slow, and now there was no rule to blame. A candidate's folder had to crawl through a dozen offices scattered across the country: the people who approve billets, the people who manage manpower, HR, security, on and on. In a lot of cases the folder moved by hand. No single office was the villain. The work item just sat in queue after queue, and the queues were the whole delay.

That's a value stream problem, not a policy one.

So we practiced what we preached in our software practice: we mapped the value-stream and business processes to identify the constraint. The primary constraint was silly: the time spent in wait status between file handoffs and signatures. We flew the actual people who sat at those desks and passed around the folder into our WeWork office in downtown Boston on a regular cadence. We batched candidates. Then we physically walked each folder around a single conference table, person to person, until it was done. In plenty of cases we interviewed, hired, and swore someone in as a government civilian in a single day, for no reason other than that everyone who had to touch the paperwork was finally in the same room.

The Air Force took our playbook and ran it as a formal hiring event and put it on the record: more than 200 résumés in a month, 57 people interviewed on-site over two days, 28 offers made on the spot, and one candidate interviewed, onboarded, and sworn in the same day in under five hours. More than half the people hired had never worked in government. The personnel director's own explanation for why it worked was teaming: hiring managers and HR experts pre-screening together and sitting in the same room. That's the bureaucracy hack.

The people at those desks aren't obstacles; they're teammates you haven't recruited yet. Same play you run on the contracting officer and the PAO earlier in this paper: bring your HR people in on day one, put them in the room for one cohort, let them feel the speed and see the kind of talent that walks in the door- then ask them to help you scale it. An adversarial back office is the single most common reason a hiring surge stalls. A back office that's on the team does the opposite: it becomes the reason the surge works. It's the NUMMI point again, this time pointed at HR: the system was the problem, not the people.

One rule about the rules: stay legal, stay familiar. We didn't do a single thing that wasn't allowed. We creatively navigated the system, and we made every unfamiliar move look like something the bureaucracy already recognized, so our request routed down a path the system trusted instead of tripping an immune response. This can’t be the Wild West. You also can’t hack a system you don’t understand. Learn the laws, policies, and rules better than the wonks. It’s easier than ever with the help of AI. As one of our mentors Dan Ward says, “Ignorance of the [rules] is a greater barrier to innovation than the [rules] themselves.”

And a word on the authorities everyone wants a list of. They exist… direct-hire, term and temporary appointments, expert and consultant hires, IPA agreements to borrow people from labs and universities, and more. Use them. In the example above we used direct hire authority funded with our program dollars. But don't get pulled into an argument about which authority as if that were the hard part. It usually isn't. Pick one that fits, verify what actually applies to your agency today rather than trusting a list in a paper, and spend your real energy on the two barriers above.

Go do

• Map your hiring value stream end to end. Treat the candidate's folder as the work item and walk its whole path. Every place it waits is a queue, and the queues- not the rules - are your delay. Find them before you touch anything else.
• Go talk to the person who owns the rule. Before you accept "policy won't allow it," read the policies. Learn them cold. Find the way through or around the issue. Then talk to the human who owns that step and show them the better way, and how it makes life better for them, too. Most walls have a door nobody bothered to knock on.
• Match the authority to the actual need. You're hiring for a tour to do one hard thing, not for a 20-year career of reassignments. Say that out loud, and pick the pathway built for it.
• Fix the assessment. SME-written, structured, same screen for everyone, real follow-ups. Kill the audio-only monologue and every other version of it.
• Batch and co-locate. Bring the humans who touch the folder into one room, hire in cohorts, and walk the folder around the table. Target interview-to-onboard in days, not months. It's been done in hours.
• Recruit HR onto the team. Same move as the contracting officer and the PAO. Pair with them on one cohort, let them feel the speed, then hand them the win and ask them to scale it. Bring them into that level 3 story, and hand them a swag bag as part of the team.

Keep them by keeping the promise

If you’ve done everything up to this point, keeping them shouldn’t be an issue. Here are three things to pay particular attention to:

They can see the line from their work to the mission. Not a press release. A chain they can trace from what they did last week to something that changed for a real person and the mission. Same output-outcome-impact chain as before, pointed inward.

Their work isn't wasted. When somebody names a piece of friction, it actually gets removed. Every obstacle you leave standing after it's been raised is a small daily argument for leaving.

They're getting better. Learning time that isn't stolen by bureaucracy or emergency. Pairing with somebody sharper. A picture of who they're becoming. People who stop growing start looking.

Some antipatterns:

Delivery pain and attrition are the same loop. Deployments hurt. Rework piles up. Bureaucracy eats the day. People burn out and leave. The ones left behind absorb more. Delivery gets worse, which makes the pain worse. That's the loop from the diagnosis, and now you know how to reverse it.

The worst thing you can do is tolerate a bad teammate. Nothing drives strong people out faster than being forced to carry someone. Not to mention, we’re only as good as the worst behavior we allow.

Where the money comes from:

You'll say you can't afford the training time or the tooling, and the order is what solves that. Ship a real win. Let the user tell your leadership what changed, and make leadership the hero because they funded it. Resources follow. Put them straight back into your people.

You can't lead with employee experience because you don't have the resources yet, which means the first team eats dirt for a while. Be honest with them about that, be honest that you know it, and then actually fix it when the resources arrive.

Go do

• Talk about the value stream internally. Show employees how their activities and outputs trace to user outcomes and mission impact.
• Kill one piece of friction a month, publicly, and say who raised it. Nothing signals "this place is different" faster.
• Give everyone a real learning block on the calendar, defended like a mission event. Ideal: get a contractor who will pair on real delivery.
• Deal with your worst performer this quarter. Coach or move them, but don't leave them sitting there. Your team already knows.
• Market every win up the chain with the user telling it. That's your funding line for improving the employee experience that’s going to unleash the next series of outcomes.

What to do, in what order

Everything in this paper is sequenced and the order matters. Here it is on a calendar.

Days 0–90: Set the conditions and get one win

The job of this phase is one shipped thing and a small group of your own people who shipped it.

• (If Possible) Get a pairing partner on contract. This gates everything else and has the longest lead time. Not a vendor who says yes to pairing. One that pairs daily as a matter of practice. We do this and orders on our SDO IDIQ can be awarded in as little as two weeks.
• Charter the initiative. Brand identity, three level story, the line, the phrase, the first desired impact, the brand promise. One page.
• Name your first cohort. Five to eight internal people. Innovators and early adopters. Plates cleared, full-time, in writing.
• Find your VSEs in the building. Ask people what they shipped last quarter. The ones who answer with a change in user or system behavior are your candidates. Train the gaps.
• Pick the pain. Map the mission value stream, find a constraint you can ship against in 90 days, define the desired impact and outcomes.
• Start fixing the IT value stream, too. Laptops, admin rights, source control, a path to production. You can't pair on real delivery into an environment nobody can deploy to.
• Map (and hack) the hiring value stream. Not "pick an authority." Walk it end to end, find where the work piles up, and baseline the four numbers.
• Say nothing publicly yet. Never talk about the future in the present tense.

You've failed this phase if at day 90 you have a brand, a plan, and nothing in production.

Days 90–180: Turn the win into a story, and the story into a pipeline

• Make the user the hero. Bring the actual user to leadership and let them say what changed. Not your slide about what changed.
• Go direct. Get out there in your own voice within 30 days of the win. Level three. Your words.
• Build the destination first. Live page, an open way to raise a hand, one named human who answers in 48 hours.
• Keep pairing. The cohort is not done at 90 days. Six months minimum, hands on keys, shipping to prod.
• Start the quarterly conference calendar. You and two engineers, technical conferences only, recruiting as the stated goal. Be ready to catch the demand this time.
• Book the hosted event. Find the influencer who can fill the room. Nothing that resembles an industry day.
• Write scorecards before you post anything. Impact map first: activities → outputs → outcomes → impact over 12 to 24 months.
• Hire only the gaps. Internal first, trained second, external last. Use members of the cohort to interview.
• Fix your onboarding. Walk it yourself and clear the roadblocks before the first outside hire shows up.

You've failed this phase if you got attention and had nowhere to send it.

Days 180–365: Make it routine

• Cohort one starts teaching. They pair the next group, sit on panels, and tell the story better than you do.
• Second and third teams on adjacent problems, staffed from cohort two.
• Now hire technical ICs. The IT value stream works and the outcomes are defined, so they'll ship instead of leave. Two or three per value stream, pointed at portfolio guidance, architecture, and the user journey.
• Hiring becomes a process, not a favor. Standing panel, continuous announcement, and whatever your re-engineered hiring value stream turned up.
• First FFP outcome contract. Just one, to prove it can be done and to teach your contracting officer what it looks like.
• Publishing becomes a rhythm. One thing a month. A shipped outcome, a person's story, a technical writeup.
• Kill one piece of friction a month, publicly, and name who raised it.
• Measure both scoreboards. Applicants per opening, offer acceptance, days to start, twelve-month retention. And the brand promise: lead time, deployment frequency, MTTR, change fail rate and most importantly: mission metrics and NPS. Post them where the team can see them, where contractors will see them, and where future hires will see them.

You've failed this phase if hiring still depends on you personally pushing every action through.

The one thing to protect

If everything else slips, protect the first shipped win and the first cohort of people who did it.

Everything downstream is powered by those two things. The brand is credible because of the win. The recruiting works because the story is true. The next tranche of money shows up because you had something to show. The people you develop become the panel that hires the next group.

Lose the win and you have a logo. Lose the people and you have a project.

Now go do it.