People ask me what is actually inside a fixed-price MGT quote. Not the deliverables, the dollars. Where does the $1,350 for a Small build go? Why is the same scope $4,800 at an agency? What does the 20 percent margin look like in actual operating costs at the end of the month? Two pages on this site answer that already. Pricing Honesty shows the per-tier breakdown with the cost-of-goods bar. Financials publishes the live waterfall from the books. This post connects them, so you can read the price you pay against the cost it takes me to operate.
The short version: I quote four tiers, every tier carries a 20 percent margin, the rest is dev time, design, and infrastructure. The margin pays for the months without revenue, the tooling stack, and the tax reserve. None of it pays for an org chart, because there is no org chart.
Why fixed beats hourly at MGT-tier scope
The whole industry argument about fixed versus hourly is real but not the most interesting part of this. The interesting part is that it only works if the scope you are pricing fits a specific shape. MGT projects ship in 1 to 12 weeks, run by one engineer, against a one-page scope document. That shape is what makes the price predictable. A 12-month enterprise migration with shifting requirements cannot be quoted fixed. A title-company portal, a Discord marketplace, a real-estate recruiting site, those are quotable in an hour because the surface area is bounded and the work is recognizable from prior builds.
Hourly billing makes sense when nobody knows what is being built yet. At the scopes I work in, that uncertainty is not a real feature of the project. It is a sales process designed to defer scope conversations until after the meter is running. The deeper cut on this is in the original fixed pricing post. This one is about the math behind the number once it is set.
How the per-tier price gets calibrated
The four tiers on the pricing-honesty page are not arbitrary. They are calibrated against shipped projects.
- Micro ($199 to $699): 8 hours, single-page or simple flow, 1 to 3 features. Examples: a Stripe-checkout pricing page, a lead-capture quiz, a marketing site with a contact form.
- Small ($699 to $2,000): 22 hours, web app with auth and a few pages, 4 to 8 features. Examples: a Discord bot with admin panel, a small SaaS with subscriptions, an internal tool for a 5 to 10 person team.
- Medium ($2,000 to $5,000): 55 hours, full platform with payments and integrations. Examples: the eXp Black Lion course platform, a multi-role SaaS with Stripe plus Slack plus email, a custom CRM with webhook automation.
- Large ($5,000 to $12,000): 140 hours, multi-role platform with admin, reporting, and scale features. Examples: Regal Title, the Cardinal MLS-integrated real-estate platform, LaunchWise AI with RBAC and multi-tenancy.
The hour estimates come from invoices, not guesses. When a Medium build keeps coming in at 50 to 60 hours across the last eight projects in that bucket, 55 is the number you can quote against. If a project does not fit the shape, it does not get a tier price. It gets a custom scope, or it gets routed to the decide page for a no-code alternative.
The 20 percent margin, in plain English
Every tier carries a 20 percent margin. That is not a profit margin in the SaaS sense. It is the slice of each invoice that is not earmarked for direct cost of goods. It pays for the things the invoice cannot see: the months I do not close a project, the tooling stack that has to keep running between jobs, the tax reserve that gets pulled out before any of it becomes income.
On a $1,350 Small build, 20 percent is $270. On a $8,500 Large build, it is $1,700. Stack a year of those margins together and you get the number that runs the operation when the calendar is empty.
The operating cost structure
The financials page publishes the actual waterfall every month. April 2026 looked like this: $4,850 gross revenue, $400 in direct operating costs (infra, AI tooling, SaaS, hosting), $1,212 in tax reserve at 25 percent, $600 reinvested into product experiments, leaving the rest as take-home.
Direct operating costs are mostly fixed and small. Infra runs around $140 a month across Vercel, Neon, Cloudflare, and DigitalOcean. AI tooling is $128 a month, which is mostly the Claude Max plan that powers the parallel-agent workflow. SaaS line items (Stripe fees, Postmark, Sentry) come in under $100. Hosting and CDN add another $35 or so. Total cost of operating the studio is under $450 a month. That is what a 20 percent margin is paying off across active projects, plus the ability to take on a slow month without disrupting delivery.
The waterfall view of this is on /financials. It updates from the books on the 1st of each month. You should be able to read it next to a quote and see exactly what part of the dollars you are paying covers what part of the operation.
Where the rest of the dollars go
If margin is 20 percent, the other 80 percent is the work. On a Medium build, the typical split is 60 percent dev time, 12 percent design and copy, 8 percent infra (Vercel, Neon, Stripe processing, Railway, domain), and 20 percent margin. On a Large build, dev time creeps up to 62 percent because the scope leans further into engineering. On Micro, design and copy land at 20 percent because a single landing page is mostly visual surface.
The dev-time number includes everything that touches the codebase: scoping, scaffolding, building, testing, deploying, and the post-launch warranty window. It is not "hours at a desk." It is "everything from the discovery call through the production deploy." That is why the effective hourly rate that drops out of the pricing-honesty calculator looks low compared to agency hourlies. An agency at 3 to 4x the price is paying a project manager, a designer, a developer, and a QA tester to coordinate. I am paying for one engineer running parallel agents, which compresses the same scope into a single billable surface.
What scope creep looks like and how it is handled
Scope creep is the single biggest threat to a fixed-price model, and it shows up the same way every time. The client asks for "one small thing" during week two. By week four there are nine "small things" that, in aggregate, are a different product than the one quoted. Pretending this does not happen is how fixed-price agencies end up bleeding margin into rework.
The MGT process for this is mechanical. Every out-of-scope request goes onto a shared list during the build. Trivial adjustments (move a button, change a color, swap copy) get done without paperwork. Anything that adds engineering time gets a written estimate against the list. At each milestone the client and I review the list together and decide what makes it into the current phase with a price adjustment, and what gets deferred to a follow-on.
Most clients self-moderate when they can see the list growing. By the time "let us also add a chat feature" is item 11 of 14, the request reads as a phase 2 conversation, not a free addition. The 20 percent margin absorbs the trivial adjustments. The change-order process protects the margin against the additions that would otherwise eat it.
What is not included in the price
Transparency about exclusions matters as much as transparency about inclusions.
- Ongoing maintenance retainers. I build, deploy, and hand over. New work is a new scope, not a monthly bill on standby.
- Unlimited revisions. Three rounds of design feedback per page, named in the scope doc. After that, additional revisions are billed separately so a single homepage does not consume an entire backend.
- Third-party costs. Stripe processing fees, Twilio SMS, Resend email, premium API access. The infra line on the pricing breakdown covers what I provision and pass through. Usage-based costs after launch are the client's bill.
- Content production. I write copy and source imagery when scope says so. Writing the entire blog or producing studio-quality video is a separate engagement.
- Long-running enterprise scope. If the project will not fit in 12 weeks with one engineer, it does not fit my model. Either the scope shrinks, or the work routes elsewhere.
When fixed pricing does not fit
A few project shapes do not work as fixed-price MGT engagements, and saying so out loud is part of the deal. If you are not sure your scope is custom in the first place, run it through the no-code vs custom decision tree before getting a quote. The cheapest engagement is the one that turns out to be a Squarespace site and a Stripe checkout link.
Fixed pricing also struggles when the user research has not happened yet. If the client cannot describe the workflows because they do not yet know what the workflows are, the right answer is a paid discovery sprint at a fixed price for the discovery itself, not a fixed price for a product that has not been defined. And if the project genuinely needs three engineers and a project manager, an agency is the correct model and I will say so on the call.
Read the price against the books
The reason I publish both pages is that pricing trust is built by showing your work twice. Once at quote time, with a per-tier breakdown so the client sees the cost-of-goods structure of the dollars they are paying. Once at month-end, with a waterfall so the same client (and any prospect) can see what the operation actually consumes. The two views should agree, and they do.
If you are sizing a project: /pricing-honesty for the per-tier math, /financials for the operating waterfall, Regal Title and 2K Service Plug for shipped examples at Large and Medium tiers. If you are not sure your project fits this model at all, /decide is the honest first stop.