No Accounts, No Excuses: What a Signup-Free App Taught Me About Friction

Anonylist started with a small domestic observation: getting 5 people to use a shared grocery list app requires 5 downloads, 5 accounts, and 5 people who care as much as you do. They never do. So I built the opposite: a list is a link. Open it, you're in. Send it to your family, they're in. Edits show up on everyone's phone as they happen. Nobody signs up for anything.

Building it taught me two lessons that apply directly to business software, plus one about infrastructure bills.

Every step you require is a tax someone won't pay

The uncomfortable math of adoption: each additional step (create an account, verify your email, download the app, set a password) loses real people. Consumer companies obsess over this. Businesses mostly don't, because employees have to use the internal tool… right?

Sort of. Employees can't refuse, but they can comply minimally. The portal that requires a VPN, two logins, and four clicks to log a field ticket gets tickets logged at the end of the week, from memory, wrong. The vendor who has to register for your procurement portal to answer one question answers by email instead, and now your records live in two places. Required-but-annoying software produces data that is late, sparse, and fictional. Friction doesn't block usage; it degrades truth.

In Anonylist, the deepest version of this: presence in a list is the permission. There's no role system, invitation flow, or admin screen; possession of the link is the access model, the same way handing someone a paper list is. Radical for enterprise software? Sure. But the question it forces is the right one for every internal tool: what is the minimum ceremony this action actually requires? It's usually less than what's built.

Low friction is an engineering result

Here's what "just a link" costs behind the curtain: live sync over WebSockets with optimistic updates, so your edit appears instantly and reconciles quietly. Reconnect-and-resnapshot when a phone sleeps. Row-level "Sarah is typing…" locks so two people don't fight over one line. An AI assistant that adds items from a spoken sentence or a photo of a paper list, with hard server-side cost caps so a viral link can't run up my bill.

None of that is visible. All of it is why the invisible feels effortless. When you evaluate software that feels simple, you're really evaluating everything underneath, which is why "can we see it work with two phones at once?" is a better demo request than any feature list.

The $6 postscript

Anonylist's entire production stack (server, database, backups, TLS) runs on a $6-a-month machine, with the database streamed continuously to off-site backup storage. Not because $6 matters, but because right-sizing does. The reflexive enterprise stack for an app like this would cost 100 times more and serve users identically.

The business translation: infrastructure should be sized to the problem, not the vendor's reference architecture. I've seen small companies pay SaaS-scale monthly bills for internal tools with 9 users. When I build for clients, the hosting bill is part of the design; you own the accounts, and the number on them should make sense for your actual load.

If your team's tools have a login problem, a "nobody fills it in until Friday" problem, or a hosting bill that smells wrong, that's a conversation I'm happy to have; the fix is usually subtraction.