When Pretty Stops Selling
An Etsy seller spent four years selling watercolor clip art: peonies, eucalyptus, small gold frames that people dropped into wedding invitations. In some months the shop covered her rent. Last spring the orders simply stopped, and she concluded that AI had killed digital products.
The story comes from Miroslav Novohradsky, who writes for Design Nexus. His reading is narrower. AI, he argues, did not kill digital products. It killed the pretty ones. One caveat before going further: the seller is unnamed, no sales figures were published, and this is one person's account passed along by an author, not verified industry data. So this article does not turn the story into a rule. It separates the question the story raises from the rules we can actually check.
A laptop screen showing a grid of dozens of nearly identical watercolor flower thumbnails on a deskView original
When pretty was scarce, and when it wasn't
The author's argument is simple. For roughly a decade, making something look good took skill and hours. A planner with a nice serif font and a sage palette could find a buyer at a few dollars, because few people could paint a decent flower, and scarce things get paid for.
Once image generators could produce a passable flower in seconds, that assumption wobbled. In his telling, buyers then revealed what they had been paying for. He asks readers to recall the last file they bought and used: a budget spreadsheet, or a lesson plan an exhausted teacher grabbed at night. Chances are it was bought to end a nagging annoyance, with the design only making the buy button easier to click.
That is an observation, not a study. Still, it gives a product planner a usable split: is what I sell a good-looking output, or a tool that reduces my buyer's work? It is worth writing the two answers down separately.
What happened when she polished harder
According to the author, the seller did what many sellers do by instinct. She redid every listing photo, bought a pricier mockup bundle, and stayed up past two softening petals and polishing gold. He compares it to shouting at a competitor who owns a megaphone factory. Effort aimed at the wrong target, he writes, does not make you better. It makes you tired sooner.
A practical check follows from this. When sales drop, ask whether the hours you add go toward solving something for the buyer or toward sharpening a quality that has already become common. In the first case, listing text, usage steps, and answers to common questions change. In the second, resolution, color, and edge treatment change. Which one is right depends on the product, but without that distinction you can work all night and travel further in the same wrong direction.
A person pouring candle wax into glass jars in a garage workshop with an inventory sheet on the benchView original
The shape of products the author expects to last
He describes the survivors as things that sound boring at a dinner party. One example is an inventory tracker for people who pour candles in their garage. Another is a chore chart made by a mother of four who has fought the chore chart war and come back with notes. Neither would win a beauty contest. Each carries the trace of someone who lived with the problem long enough to get properly annoyed by it.
Taste still counts, he adds, but it means something different: knowing which details matter to whoever uses the thing. That is not an argument against using AI. It is closer to saying that the judgment about what to keep and what to remove in AI output has to come from understanding the buyer.
The ending was not a happy one
The seller eventually closed half her shop. She started selling what she knew cold, which was planning a small wedding on very little money, and her peonies became decoration on something useful. By the author's account her income is lower than in her best year but steadier, and buyers now come for her and what she knows.
He does not dress this up. Plenty of talented people will lose money they earned fairly, he writes, and there is no hidden niche or tag hack to put anyone back on top. That honesty makes the piece worth reading, and it also marks its limit. There are no income numbers, and "steadier" is the author's description. Nothing here supports the claim that turning to problem-solving products restores sales.
A handwritten wedding budget checklist next to a wedding invitation with a dried peonyView original
Read the Etsy rules alongside the story
Etsy's Creativity Standards, checked on October 5, 2026, add a practical boundary. Etsy allows sellers to use AI tools with their own original prompts to create work they sell, as long as the listing description discloses the use of AI. The same page lists a bundle, collection, scan, or PDF of someone else's work, and AI prompt bundles, among examples that do not count as designed by a seller.
Two things follow. AI-made digital products are not banned. And inside the permitted range, producing more and prettier files with AI is something every seller can do with the same tools, so it is hard to call it a differentiator. Following the rule is the starting line. Why a buyer chooses your shop has to be built on top of it. Policies change, so check the page again before you list.
Costs that come with the shift
Moving from pretty products to problem-solving ones has a price. The source gives no figures here, so this part is general reasoning.
Time changes shape. Polishing an image is finished once you stop. Understanding a buyer's problem keeps accumulating: answering how-to messages, handling a file that will not open, and fixing a missing field and re-uploading. Count that work as part of the price.
The audience narrows. A candle inventory sheet is essential to a candle maker and useless to everyone else. The listing becomes more accurate while the pool of buyers shrinks. Since you cannot know how many people need a product until you sell it, make one or two and watch before you build a catalog.
Fees and tool costs remain. Marketplace fees, payment fees, subscriptions, and the hours spent on production and support all come out before anything is profit. We cannot tell whether the author's "steadier" income was measured after such costs. To judge your own change, record a month by the same measure before you switch.
Where to look for the buyer's problem
If you imagine the annoyance from scratch, you end up with decorative sentences again. Better clues may already sit inside your shop: words that repeat in reviews, questions you receive in messages, and reasons for refunds or complaints. If people keep getting stuck at the same place, that spot is a candidate for your next product.
If you have no such records, it is better to say you do not know yet. With few reviews, a handful of comments cannot be generalized. You can also watch which questions keep recurring in communities around a field you know well. Either way, the aim is the same: do not invent a problem without evidence.
A small check you can run this week
The steps below are my suggestion after reading the case. They are not the author's method and they promise no result.
- Pick the slowest-selling product, not your favorite one.
- Delete the first sentence of its listing, especially if adjectives such as beautiful, elegant, or premium carried it.
- Write a new first line that says what hassle the buyer will be spared, such as not writing invitation wording from scratch or fitting a guest list on one page.
- If you cannot write that line, the product may not be solving a problem yet. Collect buyer questions before touching the design.
- Change nothing else for two weeks and record views and purchases. Changing several things at once hides what worked.
In a small shop, a two-week change can be chance. The purpose of the check is not an instant sales lift. It is to put into words which question your product answers today.
The question the author leaves
He ends with a question worth keeping: who does this help, and would they notice if it disappeared? If the answer hurts, he says, that is where the real work starts, and it is slower and a lot less pretty than anything you are selling right now.
Sources: Miroslav Novohradsky
