The first 90 seconds of a Cold Read

The Cold Read is the free audit I offer: send a URL, get three specific, honest observations back the same day. No deck, no pitch. I have done a lot of them, and the uncomfortable truth is that most of the verdict is formed in the first ninety seconds.

That is not laziness. It is the same ninety seconds your buyer spends. If I take twenty minutes to understand what you sell, the audit stops being representative of anything real.

So here is the sequence, in order. The order is the method — a site does not fail everywhere at once, it fails at a specific step, and everything downstream of that step is wasted effort.

0–10 seconds · The squint test

I open the page and immediately blur my eyes. Not metaphorically — I actually defocus, or shrink the browser to 25%. What survives is the visual hierarchy: one dominant shape, maybe two supporting ones.

Then I ask the only question that matters at this stage: can I say what this company sells and who it is for?

Roughly half of B2B sites fail here. Usually because the headline describes an ambition ("Unlock the future of work") rather than a category. You do not need to be clever in the H1. You need to be placeable. A visitor who cannot file you in a mental category cannot evaluate you at all.

10–30 seconds · What the first screen is spending its space on

Now I look at what is actually above the fold and I add it up. Almost always, some large share of it is doing no work: a cookie banner, a promotional bar, a navigation with eleven items, an illustration that could be swapped for a competitor's illustration without anyone noticing.

The first screen has one job: make the visitor want the second screen. That is it. Every element that does not serve that job is charging rent and paying nothing.

• Is there exactly one primary action, and is it visually the loudest thing?
• Does the subhead add information, or does it restate the headline in longer words?
• Is there any evidence — a number, a logo, a name — or is the first proof point 1,400 pixels down?

30–50 seconds · The proof gap

I scroll once and look for the distance between the first claim and the first piece of evidence. Call it the proof gap. When it is short, trust compounds. When it is long, every additional claim actively costs you credibility, because the reader has started keeping score.

An unbacked claim is not neutral. It is a small withdrawal from an account you have not funded yet.

The best proof is specific and inconvenient: a real number with a real timeframe, a named customer, a screenshot of an actual dashboard. The worst proof is a wall of grey logos with no context — it tells me you had a trial user once.

50–70 seconds · Where the next step went

I try to convert. Not seriously — I just look for the path. Three failure modes show up constantly:

1. The single-gate site. The only action anywhere is "Book a demo". Everybody who is not ready today is worth nothing to you, which is most of the market.
2. The seven-CTA site. Demo, trial, pricing, newsletter, whitepaper, webinar, chat. Every choice you offer is a chance to choose nothing.
3. The disappearing CTA. One button in the hero, then nothing for six screens. Intent does not peak where you built the button; it peaks right after the thing that convinced them.

What I want to see: one primary action repeated after each moment of persuasion, and one low-commitment alternative for the 90% who are not buying today.

70–90 seconds · The phone

Last, I open the same page on a phone, because that is where the decision usually happens. I am not checking whether it is "responsive" — everything is responsive now. I am checking whether it is readable: body text at 16px or below, line lengths that break awkwardly, tap targets under 44px, a sticky bar eating a third of the screen.

A site can be technically flawless on mobile and still be exhausting. Exhausting is a conversion problem. I wrote more about that in reading effort is a conversion metric.

What I deliberately do not look at

Not in the first ninety seconds, anyway: the stack, the page-speed score, the animation quality, whether the design is to my taste. All of those matter later and none of them matter if a visitor cannot place you, believe you, or find the next step.

Taste is the last conversation, not the first one. Most sites I audit do not have a taste problem. They have a sequencing problem.

Written by Luka, who runs Napa Studio and does every Cold Read personally.