Use a consistent sales base
A percentage fee is incomplete without its denominator. One contract may charge on gross sales, another on platform-net revenue, and another only on sales attributed to automation. Subscriptions, tips, refunds and chargebacks may be treated differently. Ask for a worked invoice before comparing rates.
The calculator uses one user-entered eligible sales base for both models. Convert each real quote to that basis first. The defaults are illustrative assumptions and are not a representation of any provider’s full price or of typical creator earnings.
Include the work around the tool
Human chat cost can include paid hours, commission and operational overhead. An AI workflow may include a fixed software fee, a variable fee and human review. Excluding review time makes automation look cheaper without showing what it actually requires.
The model deliberately keeps those components visible. It excludes acquisition costs, platform fees, tax, refunds and costs you have not entered. It also assumes the same sales base and comparable scope for the arithmetic. Real service quality and actual sales can differ.
Interpret break-even carefully
The break-even point is where the two entered cost equations cross. It is not the point at which a provider becomes more profitable in practice. If the variable rates are equal, there is no single crossing point unless the fixed costs are also equal. A negative crossing point is outside the non-negative sales range.
Measure cost alongside quality, time spent supervising, incidents and retention. Run a limited trial and inspect the invoice. A lower fee can be attractive, but only if the workflow delivers the service you need under conditions you can sustain.
Normalise the quote before using the calculator
Suppose one quote charges on gross attributed sales and another on the amount after a platform deduction. Convert them to a common denominator before entering the commission fields. For example, under a hypothetical 20% deduction, a 5% charge on the remaining amount equals 4% of gross: 0.05 × 0.80. The deduction is an example, not a statement about your contract.
The calculator uses the same hourly cost for human chat and AI review. If your reviewer has a different rate, convert review cost into equivalent hours at the entered hourly rate or include the difference in the AI fixed-cost input. Label that adjustment in the downloaded record. This keeps the formula transparent rather than silently assuming two different wages.
The model is linear: fixed cost plus a rate multiplied by eligible sales. A tiered commission, minimum spend, credit bundle or volume discount needs separate scenarios around each threshold. Keep refunds, tax, acquisition cost and platform charges outside this comparison unless you explicitly add them; the output is an operating-cost comparison, not a profit statement.
| Line in the quote | Clarify this | Why it changes the comparison |
|---|---|---|
| A percentage of sales | Gross, platform-net or attributed sales? | The same rate can apply to different amounts. |
| A fixed monthly fee | Per workspace, creator or seat? | Account count can multiply the subscription. |
| Usage or credit charge | What uses a credit and what is included? | Overages may sit outside the displayed subscription. |
| Human review | Who reviews, for how many paid hours? | Automation can still require operating labour. |
| Refunds and adjustments | When are fees reversed or credited? | Cash billing may differ from dashboard attribution. |
ILLUSTRATIVE WORKED EXAMPLE
A $5,000 sales-base comparison, step by step
Illustrative inputs: 120 human hours at $15/hour with no commission; AI at 20% with no fixed fee; eight review hours at the same hourly rate.
- Human cost = 120 × $15 = $1,800 per month.
- AI cost = $5,000 × 20% + 8 × $15 = $1,120 per month. The entered AI model costs $680 less at this sales base.
- Equal cost occurs when $1,800 = $120 + 0.20 × sales. Solving gives $8,400 in eligible monthly sales. Above that point, these particular AI costs exceed the fixed human cost.
A cheaper model at one sales level can become more expensive at another. This example makes no claim about the sales or service quality either workflow would deliver.
Replace every assumption with your own quote and rerun low, expected and high sales-base scenarios.
Put it into practice.
A transparent cost comparison. It does not predict revenue or model equivalent service quality.
Calculate your costs ↗An AI chatter pricing worksheet for real quotes →
Related questions
Is the default 20% an endorsement of a pricing model?
No. It is an editable scenario input. Use the exact rate and eligible sales definition in your own quote.
Does this predict how much an AI chatter earns?
No. Sales are an input, not a prediction. The calculation compares cost equations at that input.
Why is there sometimes no break-even result?
Equal variable rates produce parallel cost lines. A crossing outside the non-negative range is also not a usable sales threshold.
Original practical guidance prepared for Onlytool. Worked cases are illustrative, not measured customer results. This publication does not claim independent vendor testing. Methodology and disclosure.