Most engineering platforms bill monthly. Cadence bills weekly. The shift is small in form but huge in effect. After a year of running both models for different cohorts, the data is unambiguous: weekly billing produces tighter feedback loops, faster replacement decisions, and longer cumulative engagement durations, even though individual engagement length shortens.
This post is the data we collected, what changed in founder and engineer behavior, and why we think weekly billing is the right unit for AI-native software work.
Monthly billing is the default in software contracting. Toptal, Turing, Andela, dev agencies: all monthly. The reasoning is administrative simplicity. One invoice per month per engineer. Easier accounting. Predictable cash flow.
The cost of that simplicity is a 30-day feedback delay. With monthly billing, founders defer the decision to cancel for 25 days; the cancel-or-continue decision happens once per cycle. With weekly billing, the decision happens 4x more often. The frequency change rewires founder behavior.
Cadence bills weekly because the AI-native feedback loop runs weekly. Daily ratings, weekly retros, week-end replacement option. The billing cycle has to match the operating cycle; otherwise the operating cycle stretches to fit the billing cycle, and quality suffers.
The data from the first year of weekly-billed engagements:
| Metric | Monthly billing baseline | Weekly billing | Delta |
|---|---|---|---|
| Cancel-or-continue decisions per quarter | 3 | 12 | 4x |
| Average engagement length | ~3 months | 7.3 weeks | -45% |
| Engagement length × continue rate (cumulative) | baseline | +23% | +23% |
| Founder-reported friction to replace | "high" | "low" | qualitative |
| Re-booking same engineer rate | 31% | 58% | +27 pts |
Two things to note in this table:
Average engagement length decreased. Founders cancel earlier when it isn't working. With monthly billing, the typical pattern was "wait and see" through the next billing cycle; bad engagements stretched to 3 months. With weekly billing, founders pull the plug at week 2 or 3 if it isn't working.
Cumulative engagement length, weighted by continue rate, increased by 23%. This is the counter-intuitive result: shorter individual engagements but longer total relationships, because the founders who would have churned out frustrated at the end of a bad 3-month engagement instead replaced fast and stayed on the platform.
The re-booking-same-engineer rate jumped 27 points. Founders who had a good experience with an engineer at week 4 happily re-booked them for the next project. With monthly billing, founders treated each engagement as terminal; with weekly billing, they treated engineers as a roster they could call back.
Engineers experience weekly billing as accountability with humanity. They know the founder is choosing them every week. Negative ratings hurt, but the same week's positive ratings rebuild the relationship. The feedback loop is short enough that bad weeks don't kill engagements; only patterns do.
Three behavioral shifts on the engineer side:
Better communication. Engineers send more written status updates, more screenshots, more before-after comparisons. They know the founder is making a decision every Friday; the burden of proof for "I shipped this week" is on the engineer, and they meet it.
Faster scope clarification. Ambiguous requirements get clarified within 24 hours, not the typical 5-7 days. Engineers who let scope drift used to coast; with weekly billing, drift gets caught at the next Friday.
Less politics, more shipping. When the billing cycle is 7 days, there's no value in playing internal politics. The signal that matters is shipped code. Engineers who try to manage perception instead of shipping get caught quickly.
For engineers, weekly billing is cash flow improvement that's hard to overstate. Friday payouts for the prior week's work. Never more than 7 days between work and money. Net 30 invoicing kills freelancer cash flow; weekly billing solves that.
A typical Cadence engineer at $1,500/week gets 4 payouts of $1,200 ($1,500 minus 20% platform fee) per month. The 1-week hold on the first payout (anti-fraud) is the only friction. After week 2, the engineer is on a steady weekly cadence.
For comparison: a senior US contractor on a typical agency arrangement invoices monthly, waits 30-45 days for payment, and absorbs 8-12 weeks of cash float on the first engagement. The implied opportunity cost is meaningful for working engineers.
The biggest behavioral shift weekly billing produces is around replacement. Replacement is now a Friday afternoon thought rather than a major decision.
With monthly billing, the founder thinks: "Engineer A shipped okay this month. Replacing means I have to wait 30 days for the new engineer's monthly cycle to start, plus onboarding. Total cost of replacement: ~6 weeks. I'll just wait it out."
With weekly billing, the founder thinks: "Engineer A had a bad week. Engineer B is in the matched pool. If I replace this Friday, the new engineer onboards Monday, ships by Wednesday, and I've lost one week. I'll replace."
The asymmetry shows in the data: 12% of bookings end weekly via replacement, with no founder-regret signal in post-cancel surveys. The founders who replace report higher long-term satisfaction than the ones who didn't.
This connects to the 48-hour trial in shape: both are about taking decision frictions to near-zero. The trial reduces the cost of the wrong-pick decision (release at hour 48 = no charge). Weekly billing reduces the cost of the wrong-keep decision (replace at week-end = one week's spend).
A direct comparison: 12-week project, mid-tier engineer.
| Billing model | If it works | If it doesn't (replaced at week 6) | Friction to replace |
|---|---|---|---|
| Monthly contract ($4,000/mo) | $12,000 | $8,000 + new contract setup | ~30 days delay |
| Cadence weekly ($1,000/wk) | $12,000 | $6,000 + new engineer Monday | 0 days delay |
Identical happy-path cost. Meaningfully different unhappy-path cost. The weekly-billing model saves 25% of project cost when a replacement is needed and saves the full 30-day delay either way.
For founders running 4-12 week scopes, the unhappy-path savings dominate. A 12% replacement rate over a year of bookings translates to ~$2,000 of savings per booking that needs replacement, plus 30 days of project time that doesn't get lost.
Some founders ask for monthly billing as a default. The reasons are usually:
For procurement workflows, the consolidation option works fine. For accounting habit, weekly is just as easy once it's set up. Loss aversion is the real argument against weekly, and we think the cost-clarity it forces is a net positive.
Honest framing on when monthly billing wins:
For most projects under 12 months, weekly billing is the right choice. The data is consistent across cohorts.
If you're running a marketplace where the unit of work is under 12 weeks, weekly billing is probably right. Test it. The shorter cycle changes founder behavior in ways that improve the quality of the marketplace.
The platforms that have stuck with monthly billing (Toptal, Turing, Andela) have a structural disadvantage in the AI-native era. AI-native engineers ship 3-5x faster on shippable scope; monthly billing's 30-day cycle is too slow to capture the difference. By the time a monthly-billed founder evaluates fit, an AI-native engineer has already shipped 3 to 5 features. The cycle time mismatches the work cadence.
We expect to see more weekly-billed competitors emerge. The structural argument is strong; the only thing holding the model back is operational complexity, and that's a one-time engineering cost.
See it work for your project. Book on Cadence, pick one of the matched engineers for a 48-hour trial, and you'll experience the weekly billing cadence from week 1. If it isn't working, you replace. Weekly cost transparency, no notice period, no contracts.
After the 48-hour free trial, your card on file is charged weekly for the engineer's tier (junior $500, mid $1,000, senior $1,500, lead $2,000). The cycle starts on the first day of paid engagement and continues until you cancel or replace.
Enterprise accounts can configure consolidated monthly invoices on top of weekly billing. The underlying cycle is still weekly (so your replacement and cancellation rights are weekly); the billing rollup is monthly for AP convenience.
Cancellation takes effect at the end of the current week. We bill the partial week pro-rata. So if you cancel on Wednesday and the week ends Friday, you're billed for ~3.5 days at the prorated rate.
Tighter feedback loops. Cancel-or-continue decisions happen 4x more often, which means founders pull the plug on bad engagements 4x faster. Replacement friction drops to near-zero. The data shows shorter individual engagements but 23% longer cumulative engagement when measured across cohorts.
Generally positive. The cash-flow improvement is significant; Friday payouts for the prior week's work mean engineers never wait more than 7 days for money. The accountability is real; engineers know the founder is choosing them every week. The good engineers prefer the structure.
The cycle is weekly but the engagement can be long. Many Cadence engagements run 12+ weeks; some go past 6 months. The weekly billing doesn't shorten engagements artificially; it just makes the cancel-or-continue decision available every week. Founders who want long engagements have them; founders who want short engagements aren't trapped in a long contract.