2026年10月12日 周一 · 产品 Idea Top 5

Our supplies (paper towels, gloves, food, etc.) are kept upstairs, and most staff don't go up to check before someone does the shopping. So after every shopping trip we find out something is missing, and someone has to go out again.来源 ↗

What I wish existed: when you take the last few of something, you tap one button on your phone, and it goes on the shopper's list. No sign-up for staff, just scan a QR code once.来源 ↗

At work, though, I'm not sure I could get every staff member to install an app and make an account — that's kind of the participation problem others mentioned.来源 ↗

Getting people to participate is by far the hardest part.… Amazon did something like this with Dash, but it never caught on来源 ↗

I'm building Catalog Preflight, a small tool for one specific job: comparing a current Shopify product export with the CSV you plan to import.来源 ↗

The part I'd check first is how rows get matched, because a changed Handle would show up as one removal and one addition unless the tool pairs them up by Variant SKU or another stable key. It would also help to see whether a blank cell is flagged as a clear来源 ↗

One check I'd add: values that Excel quietly changes when someone opens and re-saves the CSV, like barcodes or SKUs losing their leading zeros, or long numbers turning into 1.23E+12. That's one of the most common ways a "small" catalog update goes wrong来源 ↗

a renamed handle on import can quietly create a duplicate product or orphan its variants, and that often hurts more than a price slip.… blocking mode ("don't import if X changes")来源 ↗

Gumroad's marketplace search is literal. Every word a buyer types has to appear in your title, description, tags or creator name, and nothing gets stemmed ("template" doesn't match "templates").来源 ↗

The notion planner example is the killer one for me — the #1 notion template listing with 200 ratings, invisible for a 6,827-result adjacent search just because it never says "planner".来源 ↗

I'd call about 20 of the 200 pairs clear misses, so roughly 1 in 10, spread across about a third of the 40 listings.… around a third of the top sellers in these niches are invisible for at least one search their buyers plainly type.来源 ↗

Matching is a yes/no gate, and once you're in, the default sort is the fee Gumroad earned from the product over the last 3 months.… more words won't move them.来源 ↗

Until this week it had one hard requirement: a layered PSD, one layer per moving part. That requirement turned people away. Most people who came had one picture of their character, bought, commissioned or generated, and no layered file.来源 ↗

Nobody wants a layered PSD, they want a character that moves.… Our old tutorials literally told people to come back with a PSD, so we probably never saw most of them.来源 ↗

I'd build the converter whenever the format is just a step and not the thing people came for.… Building the converter was the right call here, because the people with one flat PNG … are a bigger group than people who own layered PSDs来源 ↗

I was frustrated with Google Analytics and Plausible, and I wanted something simple that highlighted actionable things I could focus on to improve revenue. Less focus on vanity metrics like pageviews, instead just showing where revenue comes from.来源 ↗

That tells you what's the most profitable and best use of your time, at a glance. So I built the tool I wanted.来源 ↗

I've always struggled to finish side projects and do the last 20% of polish (which takes 80% of the time somehow). … I credit AI with making it easier来源 ↗