Digital Products

Return on ad spend in the online-course category runs at about 1.27. That one number kills an entire strategy: at those economics you cannot buy your way to profit on a $40 product, and that is the category average across people doing it professionally, with budgets and testing. Anyone selling you an ads course for digital products is selling the thing that does not work.

The awkward truth about this route follows from that. It has the best margin of anything on this site, make it once, sell it repeatedly, essentially no cost per copy and it does nothing whatsoever to solve the problem you probably have.

Nobody stumbles across a digital product. Sales come from an audience you already have or traffic you already command, and with neither, the best product in your category sells nothing at all.

So the honest framing is this: digital products do not solve the audience problem. They multiply whatever audience you have. Multiply a small number and the answer is still small. What they do, once there is something to multiply, is convert attention into money at a rate no other route matches.

The arithmetic that decides whether this is worth a month#

There is no rate card here, because the number depends entirely on your own traffic. What can be given is the conversion maths, and it is unglamorous.

• A sales page for a course converts at roughly 1–3% on average and 5–10% for people who are good at this.
• An opt-in or landing page converts at about 3–5%, rising to 8–15% at the top end.
• Median order value in the e-learning and online course category is around $42.79.
• Return on ad spend in that category sits at about 1.27.

That last figure is the one from the top of this page, and it lands harder in context: everything above it is a respectable conversion rate, and then the traffic economics quietly rule out buying your way in. Sales come from audiences that already exist.

Aggregated e-commerce benchmarks, not a study of creator products. Sources: acceleroi on course conversion, Triple Whale benchmarks, Shopify on conversion rates.

Run your own numbers before you build anything#

Take an email list of 2,000 and a $40 product. A launch to that list at 2% is 40 sales and $1,600 gross. Take off platform fees and you keep somewhere between $1,300 and $1,450 depending on where you sell it, which is covered further down.

Now the part people miss. The second launch to the same list converts lower, because the people most inclined to buy already did. The third lower still. Steady-state revenue from a static list of 2,000 is a few hundred dollars a month, not $1,600.

Two conclusions follow. A month of building against a list of 2,000 is worth roughly a fortnight of client work so if you need money now this is the wrong route. And the variable that actually moves your income is list growth, not product quality, which is why building the audience is the prior problem and always will be.

Our income target calculator will run this for your own numbers, and the answer is usually larger than expected.

Why bother, then#

Because among people who reach a stable creator income, this is disproportionately how they got there. Around 45% of full-time creators have their own products or brand and earn near $100,000. Across the sector the pattern is consistent: digital products beat sponsorship on margin, subscriptions beat advertising on stability, and an email list beats any social platform on control.

There is also a structural argument that got stronger this year. Platform advertising pays you a share of someone else’s revenue at a rate they set and can change and every large platform changed its rules during 2026. A product you sell from your own page cannot be demonetised, cannot be caught by a policy update, and does not care what any recommendation algorithm did this week. In a year when YouTube doubled its entry thresholds and named three categories it will no longer pay for, that independence stopped being theoretical.

What to build first#

The instinct is to build a course. Resist it, at least at first.

Courses are the most work, generate the most support requests, carry the highest refund rate, and are the most likely to be abandoned half-finished. They are also the hardest thing to sell to an audience that has never bought anything from you, because the price point demands trust you have not yet established.

Roughly in order of effort-to-return for a first product:

A template or system. A spreadsheet, a Notion setup, a project structure, a contract pack. Days of work, not weeks. It solves a specific problem completely, the buyer knows within five minutes whether it works, and refund rates are low. This is the right first product for most people.

An asset pack. Presets, LUTs, sound effects, fonts, brushes, stock footage you already shot. If you have been making things for years you may already own the product; it needs organising and documenting rather than creating.

A written guide. Twenty to forty pages that answer one question properly. Cheap to make, easy to update, and it doubles as proof that you know the subject.

A toolkit. The template plus the guide plus the checklist, priced as one thing. Higher price, same work, and it reads as more complete.

A course. Once you have sold something smaller to the same audience and know they buy.

The test for any of them: can you name the problem it solves in one sentence, and is that sentence narrow? “Notion templates” is not a product. “A Notion system for freelance illustrators tracking client revisions and payment stages” is one, and it will outsell the general version despite addressing a fraction of the market, because the person with that exact problem recognises it immediately.

Validate it before you build it#

Three months of work on something nobody wanted is the standard failure in this route, and it is entirely avoidable. The fix is to sell it before it exists.

Write the sales page first. Describe the product, list what is in it, put a price on it, and put up a page that takes either pre-orders at a discount or names on a waitlist. Then tell your audience once, properly.

What that looks like in practice, as an actual message rather than a description of one:

I keep getting asked how I handle revision rounds with clients without it turning into unpaid work, so I’m building the thing I use: a Notion system with the client tracker, the revision log and the payment-stage template I’ve been refining for three years.

It’ll be $45. I’m building it over the next fortnight, and anyone who pre-orders now gets it for $29 and can tell me what to add: [link]

If nobody pre-orders I’ll take that as useful information and build something else.

That last line is not a joke, it is the point of the exercise. If ten people will not pay a discounted price for the description, a hundred will not pay full price for the finished thing, and you have saved yourself three weeks. Six or seven pre-orders is a green light. Zero is data, and it cost you an afternoon.

People who pre-order also tell you what to build, which is worth more than the money. The commonest thing they ask for is usually the thing you would have left out.

Pricing, where most of the money is lost#

The instinct is to price low because it feels safer and more generous. At a 2% conversion rate, a low price simply guarantees the revenue is trivial.

Work it through. A list of 2,000, converting at 2%, is 40 sales whatever you charge. At $9 that is $360 gross. At $45 it is $1,800. The conversion rate at $45 will be somewhat lower, say 1.5% so 30 sales and $1,350 and that is still nearly four times the revenue for the same work, the same list and the same launch.

Two further things nobody warns you about. Cheap products attract the most demanding buyers, reliably and disproportionately; a $9 customer will request more support than a $79 one. And a low price signals low value in a category where the buyer cannot inspect the goods before purchase, which makes the price itself part of the pitch.

Sensible starting points: $25–45 for a template or guide, $60–150 for a toolkit, $150+ for anything with your time attached. If you are agonising between two numbers, take the higher one, you can always discount and you can never quietly raise a price to existing customers without irritating them.

Where to sell it, and the tax thing nobody explains#

The platform decision matters less than people spend time on it, with one exception that matters a great deal and gets almost no coverage.

Selling a digital product to someone in another country can create a tax obligation in their country. EU and UK VAT on digital services is due where the customer is, not where you are, and there is no minimum threshold for cross-border digital sales into the EU. US state sales tax has its own patchwork of rules. Handled manually this is awful and it is the reason the merchant of record distinction exists.

When a platform acts as merchant of record, it is the legal seller. The customer buys from the platform, the platform collects and remits the tax, and you are paid a share. When it does not, you are the seller and the obligation is yours.

How to read that. Gumroad is the most expensive and has been since 2023, and what the 10% buys is that the tax question disappears. Lemon Squeezy is half the fee with the same protection. Payhip is the cheapest by a distance, its flat $99 a month works out cheapest above roughly $3,000 monthly revenue, and its free plan is hard to argue with under about $2,000 a month but you are the seller of record, and the compliance is your problem.

For most people starting out, the sensible order is: begin somewhere that acts as merchant of record, and only move to a cheaper non-MoR platform once the revenue justifies dealing with tax properly, which usually means paying an accountant to set it up. Saving 5% is not worth an unregistered VAT liability in six countries.

This is a description of how these platforms are set up, not tax advice. Thresholds and obligations depend on where you live and where you sell. Check with your own tax authority or an accountant before assuming any of it applies to you.

Sources: Letters by Burk, 2026 comparison, We Are Founders on real pricing, Getly fee comparison. Checked 7 September 2026. Platform fees change; verify before choosing.

The launch, as a sequence rather than an event#

A single announcement post is how most products fail. Not because one message is too few, but because the people who would buy did not see it. Assume half your list does not open any given email and most of your followers do not see any given post.

A workable sequence over about ten days, written out:

Email one, the problem. No product mentioned. Describe the situation the product solves, from your own experience, in a way that makes someone think “that is exactly what happens to me”. This is the most-read email of the sequence and it should be worth reading on its own.

Email two, two days later, the thing exists. What it is, what is in it, who it is for, what it costs, where to get it. Plainly. No countdown theatrics.

Email three, four days later, a specific objection. Pick the thing people actually hesitate over and address it directly. “This assumes you already work with clients, if you don’t yet, it won’t help you and you shouldn’t buy it.” Ruling people out sells more than including everyone, because it makes the inclusion credible.

Email four, day ten, closing the discount. If you offered a launch price, this is when it ends, and it must end. A deadline you extend teaches your audience that your deadlines are decorative.

The four subject lines and the shape of each, to fill in. Subject lines matter more than the bodies here, because three of these four emails only work if they get opened:

DAY 1  Subject: [the problem, as your reader would say it]
       No product. One story where you hit the problem
       yourself. End on the cost of it, not on a solution.

DAY 3  Subject: [plain name of the thing]
       What it is. What is in it. Who it is for. What it
       costs. Where to get it. Four sentences, one link.

DAY 7  Subject: [the objection, stated as a question]
       Name the hesitation and agree with it. Then say who
       should NOT buy: "if you [condition], this won't help
       you." One link.

DAY 10 Subject: [launch price ends tonight]
       Two sentences and the link. If you set a deadline,
       it ends today. Extending it teaches your list that
       your deadlines are decorative.

The third email is the one people skip and it usually outsells the second. Ruling readers out is what makes the inclusion credible, and it costs you the buyers who would have refunded anyway.

Then it goes on your site permanently and gets mentioned in passing, forever. Most of the lifetime revenue of a good digital product comes from the eighteen months after the launch, not the launch week.

After it exists#

Ask every buyer what nearly stopped them. One line in the delivery email: “Out of curiosity, what almost put you off buying this?” The answers are the most valuable marketing research available and they typically improve the next launch’s conversion more than anything you would do to the product.

Keep selling it. A surprising number of people build something good, link it once, feel awkward, and never mention it again. If the audience does not know the product exists, it does not exist. Mentioning it monthly is not aggressive; it is the minimum for a thing to be findable.

Handle refunds without argument. Give the money back, quickly, without asking why. The cost of a refund is trivial against the cost of a public complaint, and refund rates on well-described products are low. If yours are high, the sales page is overselling and that is the thing to fix.

Do not build the second product until the first sells reliably. Two mediocre products sell worse than one good one, and the effort goes into launching rather than compounding.

What AI changed here, in both directions#

Making the product got faster and cheaper. So did making something superficially similar, and the market has filled with AI-generated templates, ebooks and “systems” of poor quality. Buyers noticed.

Audience appetite for AI-generated creator content fell from 60% in 2023 to around 26% and mentions of “AI slop” rose ninefold in a year. In a category where the buyer cannot inspect the goods before paying, that suspicion attaches to anything that looks mass-produced, generic titles, stock-looking covers, contents pages that could describe anything.

That is a problem for generic products and an advantage for specific ones. What sells now is the part a model cannot produce: your particular experience of the problem, the decisions you made that turned out to be wrong, the numbers from your own work. A template built from three years of your own client disasters is not competing with a generated one, whatever they look like side by side in a marketplace listing.

So: use the tools to build faster if you like, and do not let the output look like it. The visible human specificity is not decoration, it is the reason someone chooses yours.

One practical note on the mechanics: checkout completion improved across e-commerce between 2024 and 2026, mostly through better payment handling and payment plans. If you last looked at your checkout a couple of years ago, it is worth looking again.

Who should not do this yet#

Anyone starting from no audience at all. Build the audience or sell your time first. Otherwise the product sits unsold and you draw the wrong conclusion, which is that the product was the problem.

Anyone hoping to buy traffic into profit. A category ROAS of 1.27 means the maths is tight for people who know exactly what they are doing and negative for everyone else.

Anyone who needs money this month. This is slower than client work by a wide margin and the first month is usually unpaid.

And anyone who wants to build it before talking to anybody. That is the failure mode, dressed up as craftsmanship.

The first month#

1. Week one. Write down the question your audience asks you most often. That is the product. If you cannot name one, you do not have enough contact with your audience yet, and that is the real problem to solve.
2. Describe it in one paragraph, decide what is in it, and price it. Higher than feels comfortable.
3. Put up a page taking pre-orders or a waitlist. Tell your audience once, properly, using something close to the message above.
4. Week two. If people pay, build it. If nobody does, ask the two or three who nearly did what stopped them, and change either the product or who it is aimed at.
5. Week three. Build and deliver. Choose a platform that acts as merchant of record unless you have already dealt with the tax question properly.
6. Ask every buyer what nearly stopped them. Fix the sales page with their answers.
7. Week four. Run the launch sequence to the rest of the list.
8. Every month after. Mention it. Same product, different angle. Only build the second one once the first sells without you thinking about it.

If the constraint turns out to be audience rather than product, which it usually does, owning your audience is the page to read next. Once money is arriving irregularly, money basics covers what to do with it.