How to Price a Custom Request You've Never Done Before

Price a custom request you've never done before by treating the first order as paid research: block enough uninterrupted time to finish it once without rushing, check what comparable work sells for elsewhere, then add a buffer on top for the unknowns a first attempt always turns up. Quote that number, not a guess pulled from thin air.

A regular asks for something specific — a custom mashup, a family-tree report, a personalized meditation track — and it doesn't map cleanly onto anything already on the page. Naming a number on the spot is risky both ways: too low, and that first price becomes the number everyone expects; too high, and the fan drops the request. The fix isn't a better guess. It's a short, repeatable process for turning "I've never done this" into a number worth publishing.

What's the actual framework for pricing something new?

Pricing something new takes three inputs: a timed first attempt that establishes real hours, three or four comparable public listings that establish a market band, and a percentage buffer added on top to cover fees, tax, and the surprises a first attempt surfaces. Multiply hours by a target hourly rate, add the buffer, publish one number.

None of the three steps requires special tools. A time estimate can come from a single timed test run. Comparable rates come from listings already public on other creators' pages, marketplaces, or freelance platforms. The buffer is a fixed percentage applied after the other two numbers are set, not a feeling layered in at checkout. Plan on 15–30 minutes for the whole exercise — that range is FanBell's editorial planning target, not a measured benchmark — and run it before quoting rather than mid-conversation with the fan.

Why do first-time time estimates usually run short?

First-time time estimates run short because people predict their own completion times too optimistically, even when they know similar tasks have taken longer before. Psychologists call the gap between predicted and actual time the planning fallacy, which means a gut-feel estimate for brand-new work should be treated as a floor, not a target.

The effect was first demonstrated in a study of students estimating how long their own honors thesis projects would take. Only 30% of participants finished by the time they had personally predicted (Buehler, Griffin & Ross, 1994, APA PsycNet). In the same study, students predicted an average of 33.9 days and actually took an average of 55.5 days — a 64% overrun on their own forecast (Buehler, Griffin & Ross, 1994, APA PsycNet).

Practically, that means timing a first attempt rather than estimating from memory. If a custom mashup or a rewritten resume takes 90 minutes end to end the first time — false starts included — use 90 minutes as the baseline, not the 45 minutes it felt like it should take.

Where do you find comparable rates for something you've never sold?

Comparable rates come from adjacent, already-public listings rather than one competitor's number: other creators selling a similar deliverable, freelance-platform project pages for the closest matching skill, and published wage data for the underlying craft. The goal is a low-to-high band, not a single figure, because a novel request rarely has an exact match anywhere.

Published wage data gives the band a floor. The median hourly wage across all U.S. occupations was $24.51 in May 2025, according to the U.S. Bureau of Labor Statistics Occupational Employment and Wage Statistics program (BLS Table 1) — a useful reminder that a target rate below that number is below the median for any job at all, before self-employment costs. Craft-specific medians run far higher:

"The median hourly wage for musicians and singers was $47.80 in May 2025." — U.S. Bureau of Labor Statistics, Occupational Outlook Handbook: Musicians and Singers

Comparable seller listings are easy to find because the freelance base is large. A Pew Research Center survey found that 16% of U.S. adults had ever earned money through an online gig platform and 9% had done so in the previous 12 months (Pew Research Center, 2021). Look at three or four comparable listings rather than one: a single data point could be underpriced or a loss-leader, while three or four gives a workable band to anchor the time-based estimate against.

How big of a buffer should you add on top of your estimate?

A workable starting buffer is 25–50% on top of the time-based number, scaled toward 50% for a fully unfamiliar request and toward 25% for something close to work already priced before. The buffer is arithmetic, not padding: platform fee, card processing, and self-employment tax together take roughly a quarter of a US sale before income tax.

Here is the full calculation on a real quote. A custom mashup that takes 90 minutes on a timed first attempt, priced at a $60/hour target rate, starts at $90. Add a 35% buffer — the midpoint of the 25–50% range — and the quote is $121.50, rounded to $125 — the same rounding call choosing $19 vs. $20 as a listing price walks through in more depth. FanBell's 12% platform fee takes $15.00 of that $125 (pricing), and typical US online-card processing of 2.9% + $0.30 takes $3.93 (Stripe pricing), leaving $106.07. Self-employment tax of 15.3% applies to 92.35% of net self-employment earnings (IRS Topic no. 554), which takes $14.99 and leaves about $91.08 — roughly $60.72 an hour, almost exactly the $60 target. A 35% buffer, in other words, is what it costs to actually earn the rate you quoted.

The self-employment tax rate itself is fixed and published:

"The self-employment tax rate is 15.3%. The rate consists of two parts: 12.4% for social security (old-age, survivors, and disability insurance) and 2.9% for Medicare (hospital insurance)." — IRS, Self-Employment Tax (Social Security and Medicare Taxes)

Federal and state income tax come out after all of that, which is why the top of the 25–50% band is the safer default for genuinely new work. Custom-request income can also trigger quarterly filing: the IRS requires estimated tax payments from individuals who expect to owe at least $1,000 in tax for the year after withholding and refundable credits (IRS, Estimated taxes). The 15.3% rate and the $1,000 threshold are US federal rules only — creators outside the United States size the same buffer using their own country's self-employment tax rules and any VAT or GST owed on services.

Testing a price costs nothing on the platform side. FanBell is free to start with no monthly fee, and the 12% platform fee applies only when a fan actually pays, so publishing a new custom listing at a buffered price and revising it later carries no standing cost.

Should you quote a flat price or a range for unfamiliar work?

Quote a flat price. A FanBell Creator Service listing carries one set price and one turnaround the creator chooses, with delivery capped at 120 hours (how it works), so a range has nowhere to live at checkout. A range also hands the pricing decision back to the fan mid-conversation.

That said, "flat is always better" overstates the evidence. Across five experiments, Columbia Business School researchers Daniel Ames and Malia Mason found that bolstering range offers — a range whose low end matches the single number you would otherwise have asked for — produced better outcomes for offer-makers than equivalent point offers, without measurable harm to the relationship (Ames & Mason, 2015, Journal of Personality and Social Psychology). Those results come from back-and-forth negotiations, not from a published listing a fan buys in one click, so the advice below is format-specific rather than universal.

If a request turns out genuinely outside what was scoped — a three-track remix pitched to a one-track listing, say — the creator can decline and refund it instead of absorbing the extra, unpriced work. Payment happens before the work starts either way, which removes a separate pricing risk: a creator who has already been paid isn't finishing unfamiliar work on the hope that an invoice gets settled afterward. Related: should creators require a deposit upfront covers why upfront payment matters even more for made-to-order work.

What if the job runs longer than the estimate anyway?

Finish the current order at the quoted price, then raise the listing before accepting a second order at the same rate. The first sale of anything new is a live pricing test, and the honest response to a bad first estimate is fixing the number going forward — not renegotiating mid-project with a fan who already paid in good faith.

Log the real total time once the request is delivered, including parts that weren't visible when quoting: back-and-forth to clarify scope, a revision, file conversion. Compare that total against the original estimate. A 90-minute estimate that actually took 130 minutes overran by 44%, which a 35% buffer only partly absorbed — that is a signal to relist at a higher price, not to write off the shortfall quietly. If the fan who placed that first order is a regular who'll likely ask again, how to tell a regular customer you're raising your price covers how to introduce the new number without souring the relationship.

How do you turn a one-off price into a repeatable listing?

A one-off quote becomes a repeatable listing once the estimate has survived at least one real order at that price. At that point the ad-hoc number moves from "what I told one fan" to a published price, with scope, turnaround, and exclusions spelled out the same way any standing offer on the page would be.

How to price creator services without undervaluing your time covers the ongoing version: calculating an effective hourly rate across a whole service menu once there is more than one data point to work from. A single novel request only needs the three-step framework; a recurring category of requests earns a real, calculated rate.

Frequently asked questions

Do I have to honor my quote if the request takes far longer than I estimated?

Yes, for the order already placed at that price — the fan paid the quoted amount in good faith. Deliver it, log the real time it took, and raise the listed price before the next fan requests the same thing, rather than renegotiating mid-project.

Should I tell a fan I've never done this exact request before?

It's reasonable to be direct about it. Being upfront that a request is new to you, while still quoting a firm price and a realistic turnaround, sets accurate expectations without underselling the work.

What if I can't find any comparable pricing for a truly unusual request?

Fall back on the time-estimate-plus-buffer half of the framework: time a first attempt, then add a buffer closer to 50% than 25%, since the market-rate check that would normally narrow the buffer isn't available.

How much of a custom request price actually reaches me?

On a $125 US sale, FanBell's 12% platform fee takes $15.00 and typical US online-card processing of 2.9% + $0.30 takes $3.93 (Stripe pricing), leaving $106.07 before self-employment tax of 15.3% on 92.35% of net earnings (IRS Topic no. 554) and any income tax.

Can I change the price after quoting a fan on FanBell?

You can update a listed price for future orders at any time; an order a fan has already paid for is delivered at the price they paid. Treat a bad first estimate as a reason to adjust the listing going forward, not the order already in progress.

Does FanBell charge extra for pricing or publishing a new custom listing?

No. FanBell is free to start with no monthly fee, and the 12% platform fee applies only when a fan actually pays for the listing. There's no separate cost to test a price or change it.

Create your free FanBell page