Hourly vs weekly vs monthly billing for engineers
Hourly vs weekly vs monthly billing for engineers
Hourly billing (Toptal, Upwork) makes engineers richer the slower they work. Weekly billing (Cadence) caps spend and lets you swap any week. Monthly retainers (most agencies) lock you in 30+ days with opaque utilization. For most founders shipping a product, weekly is the right shape. Hourly wins for one-off fixes under 10 hours. Monthly only makes sense if you genuinely need the same team for a quarter.
The billing cadence you pick decides three things you will live with for the next 6 months: what the engineer is incentivized to do, how predictable your burn is, and how fast you can stop. Most founders pick on price-per-hour and regret it when invoices arrive 40% over estimate.
The three shapes, briefly
| Model | Who uses it | Typical platforms | Commitment |
|---|---|---|---|
| Hourly | Toptal, Upwork, Arc | $50-$150/hr | None to 40hr minimum |
| Weekly | Cadence | $500-$2,000/wk flat | One week at a time |
| Monthly retainer | Most dev agencies, Andela | $8k-$30k/mo | 1-3 month minimum |
Each model is a system of incentives, not just a pricing line item. The price is the smallest part of the decision.
Hourly billing: the predictability tax
Hourly is the default for marketplaces like Toptal, Upwork, and Arc because it feels precise. You pay for what you use. The engineer logs hours, you approve, you get an invoice.
The problem is the incentive. Every minute the engineer works slower, they earn more. Every refactor they avoid, every test they skip, every "let me research this" hour, all of it is revenue. The best hourly engineers fight this gravity. The median one does not.
We have seen Stripe webhook integrations quoted at 8 hours bill at 23. We have seen a single Supabase row-level-security policy take a freelancer 11 hours when our internal benchmark is 90 minutes. The honest reasons vary (unfamiliar codebase, overestimated complexity, learning on the job), but the founder absorbs all of it.
Where hourly genuinely wins
- Defined scope under 10 hours. Fix a bug, write a script, set up a single integration. The work is small enough that overruns cap at maybe 2x.
- You can supervise tightly. If you are technical and review every commit, hourly forces a tight feedback loop.
- You need a specialist for one weird thing. A WebGL shader, an Erlang debugging session, a legacy COBOL touch-up. You will not need them again.
Toptal's median freelancer rate sits around $80-$120/hr for engineers. At 40 hours a week, that is $3,200-$4,800 weekly, with no cap. The "weekly" budget is whatever the engineer logs.
Weekly billing: predictable burn, flexible exit
Weekly billing flips the incentive. The engineer earns the same amount whether they finish your feature in 3 days or 5. Faster work means more time for the next sprint, not less revenue.
Cadence runs on this model: flat weekly rates at four tiers (Junior $500, Mid $1,000, Senior $1,500, Lead $2,000), no notice period, swap any week. The founder knows their monthly engineering line within $4 either direction. The engineer is incentivized to ship cleanly so they get rebooked next week.
Where weekly wins
- You are building something, not just patching it. Features, refactors, end-to-end shipping. Anything that spans days.
- Budget predictability matters. You raised $500k and need to know your runway to the dollar. A senior at $1,500/week is $6,000/month, full stop.
- You want optionality. Did the engineer underperform this week? Daily ratings inform an auto-replace flow. You are out next Monday, no severance, no negotiation.
The trade-off is honest: weekly billing assumes you have at least a week of work. For a 3-hour fix, you are overpaying. For comparable platforms in this category, see our Turing alternatives roundup, which covers the booking-vs-hiring split in detail.
Monthly retainers: lock-in dressed as commitment
Monthly retainers are how agencies and shops like Andela, Crowdbotics, and Toptal Enterprise prefer to sell. You commit 1 to 3 months minimum. You get a "dedicated" team. The pitch is continuity.
The reality is utilization opacity. A $20,000/month retainer for "two engineers" usually means two named engineers who may or may not actually be working on your project full-time. We have audited retainer engagements where the senior engineer billed 35 hours a month, the junior billed 60, and the agency pocketed $14,500 in margin against $5,500 of actual work.
The commitment shape also kills you when things go wrong. If the team underperforms in week 2, you still owe weeks 3 and 4. The exit conversation is awkward, sometimes legal.
Where monthly retainers actually win
- Multi-quarter projects with stable scope. A 6-month rebuild of your billing system. A year-long platform migration. You genuinely need the same team for a long time, and you can write a SOW that holds up.
- You want a managed account. Some founders prefer one throat to choke. The agency PM handles staffing, replacements, reporting. You pay for the management layer.
- The agency has IP you cannot get elsewhere. Specialized fintech compliance, regulated healthcare, defense work. Sometimes you are buying the firm, not the engineers.
Head-to-head comparison
| Factor | Hourly (Toptal, Upwork) | Weekly (Cadence) | Monthly retainer (agencies) |
|---|---|---|---|
| Typical cost | $50-$150/hr, uncapped | $500-$2,000/wk flat | $8k-$30k/mo, 1-3 mo min |
| Incentive | Slow work = more revenue | Ship clean to get rebooked | Maximize margin on bench time |
| Commitment | None to per-task | One week, swap anytime | 30-90 days minimum |
| Budget predictability | Low (invoices vary 30-100%) | High (flat fee) | Medium (fee fixed, scope creeps) |
| Speed to start | 2-7 days vetting | 48-hour trial | 2-4 weeks onboarding |
| Quality vetting | Variable by platform | Voice interview + AI-native baseline | Agency-vetted, opaque to you |
| Exit cost | Zero | Zero | Often a full month |
| Best for | Defined tasks under 10 hours | Active product development | Long-term scoped programs |
Worked example: 6 weeks of backend work
Let's price the same job three ways. Founder needs a backend engineer for 6 weeks to build a Stripe billing flow, a webhook ingestion service, and a Postgres schema migration. Budget: ~$10,000.
Toptal (hourly, $95/hr senior)
Estimate: 40hr/week × 6 weeks = 240 hours. At $95/hr, that is $22,800 if the estimate is right. Industry data on freelance overruns suggests a 1.3-1.6x multiplier on initial estimates. Realistic spend: $28,000-$36,000. Burn predictability: low. Exit: you can stop next week, but you have paid for whatever they logged.
Cadence (weekly, Senior $1,500/wk)
6 weeks × $1,500 = $9,000. Exact. The engineer is auto-matched in 2 minutes, you do a 48-hour free trial, daily ratings catch underperformance early. If week 3 they are not delivering, you swap and the new engineer starts Monday. Total exposure: one week.
Boutique agency (monthly retainer, $14k/mo "dedicated mid engineer")
You commit 2 months minimum. Total: $28,000. You get a named engineer, an account manager, weekly status calls. The engineer is allocated 80% to you on paper. In practice, utilization audits commonly land at 50-65%. Effective hourly rate when reverse-engineered: $130-$170/hr.
The weekly model wins this scenario by 3x on cost and on exit optionality. The hourly model wins if the actual job turns out to be 20 hours, not 240. The retainer model wins if you genuinely need 6 months and want the agency PM layer.
Decision matrix: pick by your real constraint
Three founder profiles, three different right answers.
Founder A: pre-seed, $300k raised, building MVP
Constraint: runway. Every dollar counts. Need to ship a working v1 in 8 weeks.
Pick weekly. Mid tier at $1,000/week for 8 weeks is $8,000, locked. Hourly risks blowing 25% of runway on a single engineer's overruns. Monthly retainer is a non-starter at agency rates. We cover the broader founder math in our Cadence ROI breakdown for founders (and in the worked example above, the weekly cost beat hourly by 3x for the same scope).
Founder B: seed, $2M raised, 6-month replatform
Constraint: continuity. Need the same 3 engineers across the whole project. Internal team will own it after handoff.
Pick monthly retainer (if you can find a good agency), or stack 3 weekly bookings on Cadence. This is the one case where monthly is defensible. The retainer gives you contractual continuity. The Cadence alternative is to book 3 engineers, rebook them weekly, and trust the daily-ratings system to keep them aligned. The agency is more predictable on staffing; Cadence is more predictable on cost and exit.
Founder C: Series A, post-PMF, scaling
Constraint: speed of execution. Need to ship 4 features in parallel across 2 quarters.
Pick weekly, with multiple engineers booked. You want optionality to scale up and down by feature. The flat weekly model means you can run 5 engineers for 3 weeks during a sprint and drop to 2 the next. Hourly burns too much budget on coordination overhead. Monthly retainers cannot scale week-over-week.
What about quality? Honest take
Hourly platforms vet hard at the top (Toptal famously claims a 3% acceptance rate). The engineers are real. The problem is the model, not the talent.
Monthly retainers vet the agency, not the engineer. You get whoever the agency assigns, which is often whoever is on the bench that month. Quality is uneven and you have little recourse.
Weekly platforms in 2026 vet for two things: code quality and AI-native fluency. Every engineer on Cadence is AI-native by default, vetted on Cursor, Claude Code, and Copilot fluency in a voice interview before they unlock bookings. The 48-hour free trial is the real quality check: if they cannot ship in 2 days, you walk for free.
If you want to see how this compares against a tool-fluency-vetting alternative, our Cursor vs JetBrains AI deep comparison covers the IDE side of what "AI-native" actually means in practice.
What to do
Pick by your real constraint, not the sticker price.
- If your project is under 10 hours of clearly-defined work: hire hourly on Toptal or Upwork. Just write the spec tightly and cap your authorization.
- If you are actively building product for 2-12 weeks: book weekly. The predictability and exit optionality compound. See how Cadence compares for founders and start a 48-hour free trial if it fits.
- If you have a multi-quarter scoped program with stable requirements: evaluate 2-3 agencies on monthly retainer. Demand a utilization clause in the contract.
The bigger meta-point: hourly billing made sense when measuring software work meant measuring time. With every engineer now AI-augmented, time is a worse proxy than ever for value delivered. A senior engineer with Claude Code shipping a feature in 4 hours is worth more than one logging 14 hours of careful artisanship. Weekly billing prices the outcome window, not the input clock.
If you are at the point of picking a billing model, the cheapest experiment is a 48-hour Cadence trial. Match in 2 minutes, work with the engineer for two days, pay nothing if it does not click. It is the lowest-risk way to test whether weekly billing fits your team before you commit a dollar.
FAQ
Is hourly billing always more expensive than weekly?
Not for short jobs. If the work is genuinely under 8-10 hours and tightly scoped, hourly at $80-$120/hr will beat a $500 weekly minimum. The break-even point for most product work sits around 6 hours of actual engineering time. Above that, weekly wins on both cost and predictability.
Can I switch from a monthly retainer to weekly billing mid-project?
Usually, yes, but you will pay out the remaining retainer commitment. Most agency contracts have a 30 or 60 day notice clause. Read it before you sign. The cleanest exits we see are founders who plan a 2-week overlap: keep the retainer running while a Cadence engineer ramps on the codebase, then drop the agency at the renewal date.
What happens if my Cadence engineer underperforms in week 1?
Daily ratings during the 48-hour trial flag misalignment early. If you rate the engineer below threshold, the platform auto-suggests a replacement and the trial extends to the new engineer. If you discover the issue in paid week 1, you swap for week 2, no penalty. The auto-replacement flow is the core reason weekly billing works as a model.
How do I prevent scope creep on a weekly engagement?
Write a tight booking spec. Cadence engineers self-select tier against the spec, so a vague brief gets a junior; a precise one gets a senior. Re-write the spec every Monday. Treat each week as its own micro-SOW. This is closer to how product teams work internally than to how agencies bill.
Does AI-native engineering change the math here?
Yes, decisively. Hourly billing assumes time tracks value. When Cursor and Claude Code compress a 2-day task into 4 hours, the hourly engineer earns less for shipping faster. The weekly model captures the AI productivity gain on the founder's side. This is the underlying reason weekly is gaining share against hourly in 2026.
Are there hidden fees on any of these models?
Hourly platforms often add 5-10% client fees on top of the engineer rate. Monthly retainers bury management fees (often 30-50% of the line item) in the bundled price. Weekly platforms vary: Cadence's pricing tiers are the full price, with engineers earning 80% of the rate and the platform taking the 20%. Always ask for the all-in number, not the "engineer rate".