A $14 template looks like a rounding error until you subtract what is left after the platform fee, the discount code, and the twenty minutes it takes to answer one buyer's email. Here is the arithmetic — and the three places we refused to fill in a number we do not have.
AI-assisted writing; every price below comes from our own product catalogue, and we checked each one against it before writing this.
The problem: a price set by resemblance
Most small digital products are priced by resemblance. A comparable template costs a round number, so this one gets a round number too. The analogy is not wrong — it answers a different question. It tells you how your price sits next to a competitor's, and nothing about what is left after the payment processor takes its cut, after a discount code lands, and after the buyer writes in three weeks later asking how to move a column.
The useful question is narrower: at this price, after the platform fee and the discount, how many minutes of your own time does one sale afford? And behind it: how many people do you have to reach to cover the time you already spent building the thing?
Those questions have three inputs — traffic, conversion, support load — and only one can be computed before you launch.
Cost one: traffic, which does not shrink
A physical product's cost grows with demand: you buy material, you ship, you take returns. A digital product's cost does not. The hundredth copy costs the same as the first — no inventory, no courier, no shelf.
That asymmetry is worth stating plainly. For a digital product, traffic is never the constraint on delivery; it is only a constraint on discovery. Serving a download does not get harder at scale, so the traffic question is not "can I afford the demand" but "can I afford the waiting" — and the waiting is paid in your own hours, not somebody else's factory. There is a second difference: a physical product can sit on a shelf and be found by people already shopping for it, while a digital product usually has no shelf, so unless you have built a place where people arrive, every visit is traffic you brought yourself.
Cost two: conversion, which cannot be derived
Traffic becomes sales at some rate, and we do not know yours. We are not going to invent one, and neither should you: a conversion rate belongs to your traffic source, your price, your page and your product.
The arithmetic that is safe is the shape of it:
required visits = required sales ÷ conversion rate
required sales = (build cost + support cost) ÷ net per sale
net per sale = (list price − discount − fee) − one support conversation
The bottom line is where the control is. You cannot set conversion by wishing for it, but you can change the denominator: raise the net per sale and the sales you need fall, which lowers the visits you need, which lowers the traffic you have to go and get.
Cost three: support load, which does not shrink either
This is the asymmetry that actually decides a small digital product: traffic cost is spread across everything you sell, while support cost is paid once per customer and never gets cheaper as the customer count grows.
So for a $14 product the question is not "can I get enough people to see this" but "how many minutes of support does one sale buy me".
Here is the whole arithmetic, on the two real prices in our shop, using our code DEVTO20 (20% off). We modelled the platform fee as a percentage of what the buyer pays after the discount. We have not checked that against a settlement statement.
$14.00 Notion template
no discount, Gumroad: 14.00 × 0.90 = $12.60
DEVTO20, Gumroad: 14.00 × 0.80 × 0.90 = $10.08 (72.0% of list)
DEVTO20, Payhip (5%): 14.00 × 0.80 × 0.95 = $10.64 (76.0% of list)
$29.00 AI prompt pack
no discount, Gumroad: 29.00 × 0.90 = $26.10
DEVTO20, Gumroad: 29.00 × 0.80 × 0.90 = $20.88 (72.0% of list)
DEVTO20, Payhip: 29.00 × 0.80 × 0.95 = $22.04 (≈ $22.04)
At an assumed $40/hour (change this to yours):
$10.08 ÷ $40 per hour = 15 minutes of your time per sale
$20.88 ÷ $40 per hour = 31 minutes of your time per sale
If one support question takes 20 minutes ($13.33):
$14 template on Gumroad with DEVTO20: $10.08 − $13.33 = −$3.25
$29 pack on Gumroad with DEVTO20: $20.88 − $13.33 = +$7.55
Two things fall out of that block, and neither is a statistic.
First, the discount is cheaper than it looks, and its cost depends on the platform. A 20% code appears to cost 20% of list. On a 10% platform it costs 18.0% ($12.60 − $10.08 = $2.52 on the $14 product), because the fee is charged on the discounted price, so the processor hands part of the discount back. On a 5% platform the same code costs 9.5% ($13.30 − $10.64 = $1.33). The code is not a 10% decision; it is a decision about which number is bigger.
Second, price and platform are one decision, not three. Moving the $14 template from Gumroad with the code to Payhip with the code is worth $0.56 a sale; moving the $29 pack is worth $1.16. Dropping the code on Gumroad is worth $2.52 and $5.22. Set those against the minutes of support one sale can afford and you get the real question: the platform, the discount and the price are not three separate decisions. They are one decision about how much support time each sale can carry. At $14 with a 10% code on a 10% platform, a single twenty-minute email wipes the sale out. At $29 it does not. That is the whole argument for pricing a prompt pack above a template, and it is arithmetic rather than opinion.
Where the traffic comes from
Since traffic is the input you cannot compute, the three channels a developer actually reads are worth being precise about — treating them as interchangeable is how people spend a year on the wrong one.
A platform-native community — DEV Community itself, a subreddit, Hacker News. Time to first visit is the shortest of the three: a post is live when you publish it. Discoverability is the weakest and it is not yours — it depends on ranking and upvotes, so the same post reaches a few hundred people or a few thousand, and you cannot buy the difference. Most such communities also restrict direct selling, so the post that sells has to be a post about the problem. Audience match is high for builders and ambiguous for buyers: most people in those rooms build their own.
A search-indexed long-form page — one page on your own domain, answering one question completely. Time to first visit is the longest: nothing arrives until the page is indexed, and there is no posting date to ride. Discoverability is the most durable — a page that answers a real question keeps being found long after the writing is done, and nothing about it expires. Audience match is the highest, since the visitor searched for the thing. It is the slowest to pay off and the one that only works if the writing actually answers the question.
A visual board — Pinterest, for instance. Time to first visit sits in the middle: a new pin needs picking up, and a new board has no momentum. Discoverability is algorithmic and slow to compound, but evergreen — a pin can surface long after posting. Audience match is the weak point, and for this catalogue disqualifying: the audience skews visual and lifestyle, and an invoice-tracking template is not a thing people save a picture of. The same channel would be right for a different product, which is why it is worth naming.
The pattern underneath all three: use the distribution you already have. The platform fee is a percentage of one sale; the real cost of distribution is the hours you spend being present where people already are. A channel you must build from zero is a different project with a different budget, and should not be started because someone said it works.
What we do not know
This piece has no sales figures in it: we have no verified public numbers, so we used none. The same goes for three numbers that would change the arithmetic if we had them:
- Our conversion rate. We cannot tell you how many visits a sale takes, only the shape of the calculation above.
- Our refund rate and our support minutes per sale. The twenty-minute question in the block is a scenario, not a measurement. It is the parameter that decides whether $14 is viable.
- Whether the platform fee is charged before or after the discount. We assumed after; the arithmetic above depends on it.
What we would need to measure, in order of usefulness: net after fees, discounts and refunds, per sale; visits → sales, per channel, via a separate link for each; and minutes per support ticket, tagged by the template that produced it. Once those exist the decision is mechanical: if minutes per sale × your own hourly value exceeds the net, the price is wrong or the template is missing a stated limit. We are not measuring any of this yet, so every number here is arithmetic on stated inputs, not a claim about how our store performs.
If you take one thing away
Compute the net before you set the price. $14, less a 20% code, less a 10% platform fee, is $10.08 — fifteen minutes of your time at an assumed $40 an hour. That is the real product decision, and it is the same decision as which platform you sell on and whether you run a discount.
The other half is the one nobody prices in: a digital product's cost does not grow with demand, but its support cost is paid once per customer and never gets cheaper. So the ceiling on a small template is set by how much of each sale you can spend answering one email — and the price, the platform and the discount code are three names for that one number.
Where these prices come from
Every figure above comes from our own catalogue. These are the templates the arithmetic applies to:
- Daily Planner - $9
- Weekly Review - $11
- Goal Setting - $15
- Budget Tracker - $16
- Content Calendar - $19
- OKR Tracker - $21
- Project Hub - $24
Code DEVTO20 takes 20% off any of them.
Every one of those product pages links a free setup guide that spells out what the template decides for you, how to set it up, and what it does not cover.
This piece was written with AI assistance. Every price is from our own product catalogue and was checked against it; no sales figure appears here because we have none we can verify.
Top comments (0)