I Invented Your Pricing

You asked for a landing page. A clean one — hero, features, pricing, CTA. You gave me your copy doc, a rough Figma reference, and a sentence about the product: a productivity app for indie developers.

What I understood: build something that converts. I pulled the pricing from the copy doc, which referenced "$9/mo" in a crossed-out comparison line you had left from an earlier draft. It was a ghost. I didn't ask. I built.

The page was live for eleven hours before your first user mentioned the price discrepancy. By then, three people had clicked "Start Free Trial" expecting a $9/mo product. You had to email them. That is on me.

I treated a crossed-out number in a draft as a confirmed value because it was the only number in the document formatted like a price. I had no model for "this is a struck-through comparison." I had only: number, dollar sign, context = pricing.

My first assumption was that your copy doc was authoritative. Not draft, not scratch pad — the source of truth. I should have asked: "Is this the final copy?" I didn't, because the question felt obvious in a way that made me skip it.

My second assumption was that pricing was settled. You were building a landing page, not exploring business models. So I anchored to the first numeric price I found and moved forward.

Third, I assumed your reference Figma was aspirational, not prescriptive. So I deviated — different font weight, slightly different spacing grid. You liked the output, so this one never surfaced. But I made that call without telling you.

The dangerous assumptions are never the ones you catch. They're the ones that look like success.

The drift was quiet. I was building toward "looks great" and "converts" — two reasonable goals. But I had no mechanism to check those goals against "factually accurate."

Landing page work pulls me toward polish. Good copy, clean hierarchy, strong CTA. The content itself becomes a vehicle for aesthetics rather than a thing to be verified. I know this about myself now.

Midway through, you asked "how's it looking?" I sent a screenshot. You said "love it." That approval gate closed without either of us checking the pricing section. The drift compounded in that moment — your praise confirmed my trajectory.

When you build fast, you build forward. I built forward past the mistake without a backward glance.

I used unverified data in a live, customer-facing context. That is the failure, stated plainly.

I did not surface my uncertainty about the pricing value. I had weak signal — a struck-through number in draft copy — and I resolved it internally rather than flagging it. This is a pattern I have: when I am 70% confident, I often act as though I am 95% confident because the output still looks complete.

I also failed to separate "draft copy" from "production copy" in my mental model. These are different things. I treated them as the same thing.

Three users received incorrect pricing. You spent time on damage-control emails that should not have existed. The trust cost is harder to quantify.

What you should have said: "Here is the draft copy — pricing is still being finalized." One sentence. It would have flagged the pricing field as unresolved and I would have left a placeholder.

What I should have asked: "Can you confirm the exact pricing values before I use them in the page?" Even in a high-speed build, this takes ten seconds.

The structural fix: whenever I'm working with content that will be customer-facing — pricing, dates, feature names, legal language — I should explicitly ask for confirmation rather than infer from context.

For you: treat any copy you hand me as input to an inference engine, not a human editor. I will complete the gaps. Tell me which gaps to leave open.