Second-hand paragliding gear is an unusually hard thing to sell online. The items are expensive, the buyers are spread across continents, and the community is small enough that one bad story travels a long way. The classic marketplace answer — "here's a listing page, now go arrange it between yourselves" — fails precisely at the moment that matters: the moment someone has to send money first.
So the interesting part of AdrenalineShop was never the catalogue. It was answering one question: who holds the money while two strangers decide whether to trust each other?
The problem, stated honestly
Strip away the software and a used-gear sale between strangers is a standoff:
- The buyer does not want to wire money to a stranger abroad and hope a parcel arrives.
- The seller does not want to post an expensive wing and hope payment follows.
- Neither can realistically sue the other across a border for the value involved.
Every trust mechanism you can bolt on — ratings, badges, verified profiles — is a proxy. Useful, but a determined scammer with a clean profile still ends the day holding both the gear and the money. The only structural fix is to take the money out of the middle.
The money flow — which is the whole design
Here is the sequence the entire platform exists to enforce. Read it as the product specification, because that is what it is:
The consequence is worth stating plainly, because it is the sentence the whole business rests on: a scammer with a perfect story still never gets paid. They can fake a profile, fake photos, fake a history. They cannot fake the buyer confirming that real gear arrived, and that confirmation is the only thing that opens the gate.
What that forced us to build
An escrow flow sounds like one feature. It is not. Holding other people's money puts requirements on nearly every part of the system:
- Listings and seller accounts — because you cannot pay someone you cannot identify.
- Payment handling with webhooks — the payment provider tells the system a payment completed, and that message must be processed reliably.
- An invoice and payout trail — every movement needs a record, because "where is my money" is a question that must always have an answer.
- Dispute handling — sometimes the gear genuinely is not as described, and a human has to decide.
- Admin tooling — someone has to see the whole picture and act on it, quickly.
Hosting costs almost nothing, so the platform is not fighting a monthly bill before it makes a sale.
The failure mode that actually bites
Here is the thing experience teaches you and tutorials do not. In a money system, the dangerous failure is rarely money moving wrongly. Money moving is guarded, logged and usually requires a human click.
The dangerous failure is not seeing a signal in time. A webhook that silently fails. A notification email that stops sending. A payment that completes while the system quietly does nothing about it. Nothing is stolen — but nobody knows something needs doing, and the gap grows.
Which produces a rule we now apply to everything we build:
Why it is built out of boring parts
No framework. Plain HTML, CSS and JavaScript, with serverless functions for the parts that need a backend and a managed database behind them.
That is a deliberate choice, and the reasons are commercial rather than aesthetic. It loads fast on a phone on mobile data, which is where most buyers actually are. Hosting costs almost nothing, so the platform is not fighting a monthly bill before it makes a sale. There are no plugins to update themselves overnight and break checkout while you sleep. And every page is real crawlable HTML, so search engines can index the catalogue — which matters enormously when your buyers are searching for a specific wing model at two in the morning.
What we would tell anyone building something similar
- Design the money flow first. Everything else — the pages, the design, the categories — is downstream of who holds the funds and what releases them.
- Make the guarantee explainable in one sentence. If a nervous buyer cannot repeat your protection back to you, it will not reassure them.
- Instrument the quiet paths. Anywhere the system does something without a person watching, assume it will fail silently one day and build the alarm now.
- Keep a manual backstop. Automation should be the fast path, not the only path. Someone must always be able to look at the real state and fix it by hand.
AdrenalineShop has been trading since 2017. It is the system every other project we take on gets measured against — and the reason we are comfortable saying we build things that handle real money, rather than things that look like they could.