How to build a 6-figure dev agency

To build a 6-figure dev agency, you need one tight niche, gross margins above 50 percent, a retainer base that covers your fixed cost, and a bench you can flex without hiring. $100k a year in revenue is not a milestone; it is the bottom edge of viability for a real shop with one or two people on payroll.

This post is the honest version. Most "build a 6-figure agency" content is written by marketing coaches selling a course, not engineers who have run a P&L. The numbers below are what dev shops actually clear, the mistakes that actually break them, and the operating moves that actually compound.

What 6 figures actually means for a dev agency

$100k in collected revenue at a 50 percent gross margin is $50k of gross profit. Out of that comes your tooling (GitHub, Linear, Vercel, Cursor seats, accounting software), your benefits if you have any, your taxes, and any non-billable hour you spent selling, hiring, or sitting on a discovery call.

In practice, a solo dev clearing $100k revenue takes home $35k to $55k depending on tax setup and how much of the work was subcontracted. That is less than a mid-level engineering job almost anywhere.

The honest target for a real shop is $300k to $500k in year-two revenue with margin held above 50 percent. That is when payroll, partner draw, and a small marketing budget all fit without cannibalizing each other.

There is also a hard ceiling baked into the owner-operator math. At $150 per hour, 40 billable hours per week, 48 weeks per year, you cap at $288,000 of personal billings before tax. That assumes you sell zero hours, which is fiction. Realistic ceiling is closer to $200k to $250k for a working owner. To go past it, you have to stop trading hours.

Niching is the most-skipped lever in dev-shop scaling

Niching is the single highest-impact move and the one most owners refuse to make. They keep the door open to "anything Node or Python" because they are afraid of saying no.

A niche is a vertical plus a workload, not just a stack. "We build Shopify Plus migrations for DTC brands doing $5M to $50M" is a niche. "We do React" is a commodity.

The pricing power difference is enormous. Generalist React shops bill $80 to $120 per hour. A specialist who can name three previous identical migrations bills $200 to $400 per hour for the same code. The work is identical. The trust premium is real.

A few examples that work right now:

• HIPAA-compliant Rails apps for telehealth Series A startups
• Stripe + Supabase migrations for SaaS companies leaving Firebase
• Shopify Hydrogen rebuilds for fashion DTC
• Internal tools for private equity portfolio companies on Retool
• Klaviyo and Postscript implementations for ecom brands over $10M

If you cannot name the vertical, the workload, and the typical client revenue band in one sentence, you have not niched yet.

Positioning and the project vs retainer mix

The right mix at year two is roughly 60 percent retainer, 40 percent project. Retainers cover your salary and your team's. Projects fund growth, hiring, and the occasional bonus.

Pure-project shops live and die on pipeline. One slow quarter and payroll eats the cash buffer. Pure-retainer shops get stale; the work flattens, the senior people leave, and you cannot hire because nothing exciting is happening.

Productize at least one offer. A fixed-scope, fixed-price, fixed-timeline package that you have shipped at least 5 times. "$18,000 for a 4-week Shopify Plus migration including data import, theme port, and 30 days of post-launch support" is a productized offer. "We do migrations" is not.

Productized offers carry the highest margins because the SOPs are written, the estimates are calibrated, and the surprises are gone.

Pricing: hourly, fixed-bid, and value-based

Hourly billing punishes speed. The faster you ship, the less you make. It also rewards padding. Use it for discovery and for clients you do not yet trust.

Fixed-bid transfers all execution risk to you. If you mis-scope by 30 percent, you eat 30 percent. The defense is a written change-order policy and a discovery phase priced separately.

Value-based pricing means you charge a percentage of an outcome. "We will rebuild your checkout for $40k flat, and if conversion lifts more than 0.5 percent over 90 days, you owe a $20k success fee." This works when the outcome is measurable and you control enough of it to actually move the metric.

Pricing model	Best for	Margin profile	Risk owner	Scope-creep defense
Hourly	Discovery, ambiguous scope	30-45%	Client	Weekly cap
Fixed-bid	Productized, repeatable work	45-60%	Agency	Written change orders
Value-based	Revenue-attached outcomes	55-75%	Shared	Outcome metric in SoW
Retainer	Ongoing optimization	50-70%	Agency	Monthly hour ceiling

A healthy shop runs all four at once. Hourly for new clients and discovery, fixed-bid for productized work, value pricing for the few clients where you can measure dollars moved, retainer for ongoing relationships.

Margin math and the 50 percent floor

Gross margin is (revenue minus direct delivery cost) divided by revenue. If you bill a senior engineer at $200 per hour and pay them $100 per hour fully loaded (salary, taxes, benefits, equipment), that is 50 percent gross margin before any company overhead.

Anything under 50 percent and you do not have a business. You have a job that bills its clients. The traditional agency average is 10 to 20 percent, which is also why so many of them quietly close.

Productized retainer work can hit 65 to 75 percent margin once the SOPs are dialed in, the templates exist, and the same engineer can run three of them in parallel without dropping balls. That is the real reason productization matters: it is the only path to margins north of 60 percent without billing magic numbers.

If you are reviewing a project that prices below the 50 percent floor, walk away. It will not get better. If you find yourself needing to discount to win logos, your positioning is wrong, not your pricing. For broader context on what specialized engineers actually cost in 2026, the global benchmark data on freelance developer hourly rates by skill in 2026 is the cleanest comparison set we have seen.

Hiring your first three: subcontract first, hire later

The first hire is the most expensive mistake an agency owner can make. A senior engineer at $180k all-in (salary, payroll tax, benefits, equipment, software) costs roughly $3,400 per week with zero flex. If pipeline drops for 8 weeks, you eat $27,000 sitting on a bench.

The right sequence for the first three slots is: yourself, then one reliable subcontractor (paid per project or per week), then one PM or ops hire who actually frees your time, then your first FTE engineer. Most owners do this in the wrong order and hire the engineer first because it feels like growth. It is not; it is fixed cost.

Subcontracting and on-demand booking let you carry the work without carrying the payroll. Booked engineers cost variable, match your invoice cycle, and disappear when the project ends. A shop running on weekly-billed bench can take a $40k project, deliver it with a Senior at $1,500 per week for 6 weeks ($9,000 in delivery cost), and clear a 78 percent margin without ever touching the FTE budget.

This is where Cadence fits naturally for agencies. Every engineer on the platform is AI-native by default, vetted on Cursor, Claude Code, and Copilot fluency before they unlock bookings, which means a booked Mid or Senior ramps into your codebase faster than a cold contractor. Pricing is fixed: Junior at $500 per week, Mid at $1,000, Senior at $1,500, Lead at $2,000. There is a 48-hour free trial, weekly billing, and you can replace any week without notice. For agencies, the natural pattern is to book under your own brand at your agency markup. Your client never sees the underlying booking.

For a deeper read on how senior engineering compensation actually breaks down across stages, the CTO salary at startups in 2026 data is a useful sanity check on what you are competing against when you eventually do hire FTE.

Cashflow: the silent killer of 6-figure shops

Cashflow kills more dev agencies than bad delivery does. The pattern is brutal and predictable: you ship a $30k project on Net-30 terms, the client takes 45 days to actually cut the check, and meanwhile you have paid your engineer every Friday for 6 weeks. By the time the cash arrives, you have funded the project out of personal savings.

The defenses are mechanical, not clever:

• Net-0 deposits on every project. Minimum 30 percent up front, 50 percent on six-figure work.
• Milestone billing on anything longer than 4 weeks. Bill weekly or biweekly against deliverables.
• Net-15 on retainers, billed in advance. If a retainer client does not pay before the month starts, work stops.
• A cash buffer equal to 3 months of payroll before you make your first FTE hire. Not 1 month. Not 6 weeks. Three full months.

Weekly-billed bench solves part of this structurally. If your variable cost arrives weekly and your client pays on a milestone tied to that same week, you are never floating more than 7 days. Compare that to FTE payroll, which fires every two weeks regardless of whether the client paid.

Scope creep is the margin assassin

Every fixed-bid project needs a written change-order policy in the SoW from day one. The clause is one paragraph long: any work outside the listed scope requires a signed change order at the agreed hourly rate before it starts. No exceptions. No "we will figure it out later."

Track scope deltas weekly. The PM (or you) reviews everything that hit the project board that was not in the original scope. Bill it or kill it. Do not let it ride.

Lost margin on uncontrolled scope is typically 15 to 30 percent of project value. On a $50k project, that is $7,500 to $15,000 of gross profit you handed back to the client because you did not want to have an awkward conversation in week 3.

Walk away from clients who refuse to sign change orders. They are telling you exactly how the relationship will go.

The owner-operator trap and how to escape it

If you are still the senior engineer on every project, you are the ceiling. Revenue caps at whatever you can personally bill plus whatever your one subcontractor handles. That number is somewhere between $250k and $300k for almost everyone, and it requires 60-hour weeks to clear.

The escape hatch is not "hire more." It is replace yourself in delivery before you replace yourself in sales. Sales is the last thing the owner gives up because it is the only function where trust does not transfer easily.

Mechanically, this means: pick one productized offer, write the SOP, hand it to a Senior or Lead (booked or hired), watch where it breaks for two cycles, fix the SOP, then never touch that offer again. Repeat for the next offer. Within 12 months you can have 3 to 5 productized offers running without your hands on any of them.

The agencies that scale past $1M do this. The ones that do not, plateau and burn out by year three.

What to do this week

Concrete next moves, in order:

1. Write down your niche in one sentence: vertical, workload, client revenue band. If you cannot, you have homework before you do anything else.
2. List 10 dream-client logos that fit the niche. Reach out to 3 this week.
3. Set the gross margin floor at 50 percent. Audit your last 5 invoices. Anything below the floor gets repriced or fired at renewal.
4. Productize one offer. Fixed scope, fixed price, fixed timeline. Ship version one even if it is rough.
5. Open a bench account on a weekly-billed platform so the next spiky project does not force a panic hire. If you want to see how a Senior or Lead engineer actually performs on your codebase before committing, you can book your first engineer on Cadence and run the 48-hour free trial against a real ticket.

If you run an agency or do client work full-time, the Cadence partner program pays 10 percent recurring on every founder you refer for as long as they stay on the platform. Most agencies leave this revenue on the table because they think referral money is small. It is not, when it compounds across a client roster.

FAQ

How much revenue do I need to call a dev agency 6-figure?

$100k in collected revenue, not booked. Booked revenue is a vanity number. If your gross margin is below 50 percent, you are running a job, not a business, no matter what the top-line says.

Should I hire or book engineers first?

Subcontract or book on-demand for at least 6 months before your first FTE. You learn the role, you keep variable cost, and you avoid laying off a friend if pipeline drops. The first FTE should be ops or PM, not engineering. The pattern of weekly-billed booked talent is the same one solo developers and small shops use to take on senior work without committing to payroll, similar to the best home office setup for remote engineers becoming the default infrastructure for distributed teams.

What is a healthy gross margin for a dev shop?

50 percent is the floor. 60 percent is healthy. 65 to 75 percent is what productized retainer offers can hit once your SOPs and templates are dialed in. Traditional agencies average 10 to 20 percent, which is why most of them quietly close.

How do I price retainer work?

Anchor to outcomes plus a monthly hour ceiling. Charge for capacity and access, not effort. A 40-hour retainer at $200 blended is $8,000 per month. Cap usage at the ceiling, bill overflow at the same rate, and review the contract every 90 days.

How do I find clients for a development agency?

Pick one vertical and answer questions where its founders gather: Slack groups, subreddits, conference Discords, niche Twitter circles. Warm intros from existing clients close 4 to 6 times faster than cold outbound. Cold lists are for marketing agencies; dev agencies live on referrals and reputation in narrow communities.