Since the Show HN last year: rebuilt production so we don't outsource printing anymore, added frame-by-frame preview before checkout, and shipped a lot more wedding-favor + paper-anniversary orders than I expected.

Weirdest lesson: people care more about seeing the exact frames that will print than about paper specs. The preview tool converted better than any copy change we tried.

https://videotoflip.com

The site consumes those values and can switch between completely different themes without changing the frontend code

- The first clip-path is a circle (top-left corner) - The second clip-path is a polygon, that acts like a ray (hardcoded, can be improved)

The original work by Iventions Events https://iventions.com/ uses JavaScript, but I found CSS-only approach more fun

Here's a demo and the codebase: https://github.com/Momciloo/fun-with-clip-path

Comments URL: https://news.ycombinator.com/item?id=46925811

Points: 141

# Comments: 38

I figured it might be useful to share what happened after the spike - specifically, the parts that didn't work, things we failed at, broke, or turned out very differently than we expected. The whole year felt like putting out one fire after another, to the point where we'd just start laughing.

Here are some of those fires:

- Vercel shut down our website A few hours after the HN post went viral, we hit the free plan limits and the site went down. It took me about 30 minutes to notice and upgrade to a $20/month plan. Fair enough.

- People are nasty Some visitors uploaded NSFW videos, which caused Cloudinary to block our account several times in the first 48 hours. We were on the free plan, so I upgraded ($90/month), after which moderation stopped being an issue. Flipbooks are personal, intimate gifts, so as long as everything is consensual, we don't mind.

- We spent ~$20k on paid channels. Almost all of it failed. Paid ads didn't work for our product (skill issue is possible), and I'm not a fan of pouring money into Meta's or Google's black-box algorithms. After about four months of testing, we stopped. The total return was under $1,000.

- Most "product" and "vendor" directories are borderline scams. We assumed flipbooks could work well as wedding favors.. so we listed on several wedding vendor marketplaces. This resulted in dozens of automated or suspicious inquiries per week, almost none of which led to real customers. We paid around $500/month for several months and got nothing measurable back.

- Social media wasn't the growth channel we expected. We saw well-produced videos flop and low-effort videos go viral. At some point, it stopped feeling like quality was strongly correlated with outcomes. We thought this might be the kind of product that sells itself on social media. It isn't.

- Bringing production fully in-house gave us control and new problems. Owning the process improved quality, but also meant printers breaking, technicians not showing up, and production halting at the worst moments. We eventually bought two additional printers so we could run multiple machines in parallel and keep one as a backup. Reliability came from redundancy, not optimization - which still hurts my IT brain.

- SEO worked, just not how we expected. Organic search (and even ChatGPT and Perplexity) has been one of the few consistently positive channels. Wedding related keywords didn't move the needle much, but "paper anniversary" did. We're not based in the US and didn't even know this was a thing. Now, roughly 90% of our orders are for paper anniversaries - and such a cool use case.

- Shipping and customs to the US When the US removed the minimum order value threshold for international shipments this August, some customers learned about customs duties only after ordering from us, when DHL contacted them for additional payment. We're still improving how we communicate this.

None of this killed the business, but most of it contradicted our assumptions and hurt our wallet. A lot of the year was spent paying for lessons rather than making the progress we expected. Strangely, after 15+ years in IT, dealing with physical problems like these still feels easier and more satisfying than debugging production systems.

That said, a few basic decisions helped the business survive:

- As order volume increased, we negotiated DHL shipping from $90+ USD for 4-day delivery down to $15. - Don't want to brag but... technically, we had zero production issues over the year: no downtime, no critical bugs. - We found a reliable paper vendor with consistent quality. - Our binding and cutting machines didn't break once, which feels like a small miracle.

Happy to answer questions or go deeper on any of this.

But yes, I agree even a simple flag (“tester”) on accounts would have gone a long way. At least then we could separate metrics and clean up the noise afterwards without burning so much time. Definitely something I’d do differently if we ever tried again.

I expected people to try breaking into user data, finding crazy privilege escalations, or discovering ways to dump entire databases. What actually happened was… different.

We got reports about: - Creative ways to inject JavaScript into HTML through weird embedding edge cases. - Strange file upload pitfalls we hadn't thought of. - Tons of tiny UX bugs filed as "security." - And the most time-consuming one: things that worked exactly as designed, but looked like bugs to outsiders.

Example: in our CMS, you can use an "entry pointer" to select related content. A blog editor with no permissions for “products” should still be able to select products to recommend, but not open/edit/delete them. To make that UX work, our API returns product titles even if you don’t technically have product access. This is by design. We had to spend hours explaining this again and again to bounty hunters convinced it was a data-leak.

What we hoped: our existing users would feel encouraged to report issues. What we got: bounty hunters who’d never heard of us, hammering the API.

Another surprise: the program didn’t really motivate our existing users to report bugs. Instead, we suddenly got a wave of people who literally searched Google for

inurl:security "reward" "cms" bug bounty bug bounty reward $100

and landed on us. No genuine conversations, just transactional hunting. For a few months, 30% of all site visitors came from that one Google query. Our Sentry lit up from random API fuzzing, exceptions everywhere, onboarding metrics tanked.

So yes, the bounty helped. We did patch dozens of real issues that would have been painful later. We ended up paying out ~$5k total, and several reports led to meaningful patches. So even with small rewards, we got serious contributions. But it also forced us to “fix” things that weren’t broken (just to stop getting the submissions about them haha), drained time explaining design decisions, and buried real user issues.

We eventually shut it down. For us right now, it’s more valuable to understand how new users onboard, where they get stuck, and what makes them stay.

So, if you’re considering creating a bug bounty, be ready for bounty hunters to outnumber your real users. If you want signal from your actual users, maybe wait until you can afford the noise.

Comments URL: https://news.ycombinator.com/item?id=45167961

Points: 2

# Comments: 2

As for the equipment, we currently use:

- 450VS/520E Electric Paper Cutter - BINDER K5 (Soft Binding Machine) - Ricoh MPC3003 (Color Laser Printer)

Thanks again for the insights!

Hi HN! I'm Momcilo Popov, co-founder of BCMS, a flexible, developer-friendly headless CMS used by developers and agencies to build highly customizable websites and applications. We’re looking for freelance YouTube content creators to help produce:

- Tutorials on using BCMS effectively (from setup to advanced features). - General front-end web development videos aimed at educating and inspiring developers.

Our ideal creators are skilled at making clear, engaging content with a focus on real-world scenarios. Familiarity with BCMS is a plus but not required.

If you love teaching, storytelling through code, and video production, let’s talk! Please reach out at video@thebcms.com with samples of your work or channel link.

Looking forward to hearing from you!