I Made $2,890 in 60 Days Selling AI Prompts on PromptBase and Gumroad: The Complete Process Nobody Talks About

Sixty days of building, testing, and selling AI prompts across PromptBase and Gumroad. $2,890 total revenue ($2,159 net after fees). Tool spend: $40 per month. The three prompt categories that sold versus the four that did not. How pricing determines sales volume on these platforms. The testing process that separates prompts worth $15 from prompts worth $8. Every number, every platform, every mistake.

🔧 Tools mentioned in this article

ChatGPT

Used for testing ChatGPT-specific prompts before listing them — ensures prompts work as described

chatgpt.com

Visit

Midjourney

Used for creating and testing image generation prompts — the best-selling category on PromptBase

www.midjourney.com

Visit

PromptBase

Primary marketplace for selling AI prompts — takes 20% commission on each sale

promptbase.com

Visit

Gumroad

Secondary platform for prompt bundles — higher ticket sales ($15-49 bundles versus $4-9 individual prompts)

gumroad.com

Visit

Marcus Webb

June 19, 2026

What Sells and What Does Not on PromptBase

• TOP SELLING: Midjourney image prompts for commercial product photography — 94 sales, average $4 each, $376 gross
• TOP SELLING: ChatGPT prompts for professional email writing in specific industries — 87 sales, $3.50 average
• TOP SELLING: ChatGPT prompts for social media content in specific niches (real estate, fitness, food) — 76 sales
• DID NOT SELL: Generic ChatGPT writing prompts — 3 sales total across 4 listings
• DID NOT SELL: Claude-specific prompts — PromptBase audience primarily uses ChatGPT and Midjourney
• DID NOT SELL: Overly complex multi-step prompts — buyers want immediate, one-click results
• DID NOT SELL: Prompts without example outputs shown in the listing

The Prompt Testing Process Before Listing

# Prompt Quality Testing Process
# Run every prompt through this 5-test process before listing
# A prompt fails if it does not pass at least 4 of 5 tests

## Test 1: Consistency Test (run prompt 3 times with no changes)
The output quality must be consistently good across all 3 runs.
If results vary from excellent to mediocre: the prompt is not ready.
Fix: add more specific constraints until results are consistent.

## Test 2: Beginner Test
Ask someone with no AI experience to use the prompt.
Can they get a good result without knowing anything about AI prompting?
If no: simplify the prompt or add clearer instructions.

## Test 3: Edge Case Test
Run the prompt with unusual or difficult input values.
Does it still produce acceptable output?
Example: an email prompt for a dentist should also work
for an accountant without the user understanding prompting.

## Test 4: Value Test
Is the result genuinely better than what someone could ask ChatGPT/Midjourney
without this prompt?
If the difference is marginal: the prompt is not worth buying.
Buyers can feel this on first use and will leave negative reviews.

## Test 5: Example Output Test
Generate 3 example outputs using the prompt.
Would you personally be impressed by all 3 if you saw them
in a product listing?
If not: the prompt needs more work before listing.

## Prompts that passed all 5 tests:
18 of 34 prompts tested — 53% pass rate
Failing prompts were fixed and retested, not abandoned
Final listed prompts: 22 (4 were rebuilt after failing twice)

The Gumroad Bundle Strategy That Generated 57% of Revenue

Individual PromptBase sales max out at $5 to $9 per transaction. Gumroad bundles allow $15 to $49 pricing for collections of prompts. The same prompts that sold for $4 each on PromptBase were bundled into themed collections and sold for $29 to $39 on Gumroad. The audience is different: PromptBase buyers want a specific prompt for an immediate use. Gumroad buyers are often small business owners or freelancers investing in a toolkit they will use repeatedly.

# Gumroad Bundle Creation Strategy

## Bundle 1: 'Real Estate Agent AI Toolkit'
Contents: 15 prompts covering
- Property listing descriptions (5 prompts for different property types)
- Client email templates (4 prompts)
- Social media captions (3 prompts)
- Open house scripts (3 prompts)
Price: $29 (€26.70 / £22.90)
Sales: 22 copies = $638 gross

## Bundle 2: 'Fitness Coach Content Pack'
Contents: 20 prompts covering
- Workout plan explanations for clients (6 prompts)
- Nutrition advice posts (4 prompts)
- Instagram captions (5 prompts)
- Client check-in messages (5 prompts)
Price: $19 (€17.50 / £15)
Sales: 18 copies = $342 gross

## Bundle 3: 'Midjourney Product Photography Mega Pack'
Contents: 30 Midjourney prompts for
- White background product shots (10 prompts)
- Lifestyle product photography (10 prompts)
- E-commerce hero images (10 prompts)
All prompts include example outputs in the listing
Price: $39 (€35.90 / £30.80)
Sales: 15 copies = $585 gross

## What made these bundles sell vs generic prompt packs:
1. Specific profession or use case in the title
2. Example output images shown (crucial for Midjourney bundles)
3. Listed number of prompts explicitly
4. Gumroad preview showed 2-3 prompts to demonstrate quality
5. Priced at perceived value of time saved, not at number of prompts

Mistakes Made in 60 Days

• Mistake 1: Listing prompts without example outputs — listings with examples converted at 4.2% versus 0.8% without. Added examples retroactively to 8 listings and saw immediate sales improvement.
• Mistake 2: Pricing too low to test market — initial pricing at $2-3 per prompt. Raised to $4-5 and sales did not drop significantly. Left money on early sales.
• Mistake 3: Creating prompts for Claude on PromptBase — PromptBase audience is 80% ChatGPT users. Claude prompts had 91% fewer sales than equivalent ChatGPT prompts. Claude bundles moved to Gumroad where audience is more tool-agnostic.
• Mistake 4: Not building an email list from Gumroad buyers — 55 buyers with no contact mechanism. Now using Gumroad's follow-up email feature to notify buyers of new bundles.

Final Thoughts

Prompt selling is a real income stream with $2,890 in 60 days as documented proof. The ceiling is lower than digital product selling because prompt prices are constrained by what people will pay for something they could theoretically create themselves. The strategy that broke through that ceiling was niche bundling on Gumroad where the audience understands the value of a curated, tested collection for their specific profession. The $30 per month in tools against $2,159 net income makes this one of the highest-margin income streams documented across all experiments.