Improve With Wisdom, Not Hype · Website & Content
The basic truth
Most website redesigns start with a feeling: it looks dated.
The feeling is real, and it’s almost always coming from the wrong person. You are the only one who has seen your website more than three times. You look at it weekly, notice everything you’d change, and are bored of it long before a customer has finished forming a first impression. Your buyer sees it once, for four minutes, while comparing you to two other companies — and dated is nowhere near the top of what they’re evaluating.
So the project gets scoped around how it should look, and the question of what’s actually broken never gets asked. A year later the site is handsomer and performs the same, because appearance was never the diagnosis. It was just the symptom anyone could see without looking.
What to do
Find out where people stall before deciding what to build. The evidence is already sitting in systems you own and mostly costs nothing but attention:
- Your internal site search. What people type into your own search box is a literal list of what they came for and couldn’t find. It is the most direct feedback a website produces and the most routinely ignored.
- Where sessions end. Not bounce rate as a number, but which page people are on when they give up, and what they were trying to do there.
- The paths that lead to real inquiries — and how many of them go through pages nobody thought were important.
- Recordings and heatmaps, used to answer a specific question rather than browsed for entertainment.
- Your sales team’s most repeated sentence. If they explain the same thing on every call, the site isn’t explaining it.
Be subtractive before you’re additive. Good strategy usually removes before it adds — pages nobody reads, navigation items that split attention, three overlapping explanations of the same product. Removing costs nothing to test and is reversible. Most sites would improve more from a careful deletion pass than from a rebuild.
Change one thing where you have a hypothesis. “People can’t tell which model fits their material, so we’ll add an application-first path and watch whether quote requests mention the application” is a test. “Let’s refresh the homepage” is a project. The first produces knowledge whether it works or fails; the second produces a homepage.
Test against the buyer, not the room. In a website review, the strongest opinion usually belongs to whoever is paying — who is also the least representative user in the building. Ask what evidence would settle it. If none exists, that’s the next thing to go find. “I don’t like it” is a legitimate reason to look closer and a terrible reason to rebuild.
The common trap
Treating the redesign as the remedy for everything.
It’s appealing for an honest reason: a redesign is a project. It has a budget, a start, an end, and a visible result you can show people. Diagnosis is a habit — unglamorous, ongoing, and it never produces a launch party. Organizations reliably choose the project, and then attribute whatever happens next to it.
The deeper cost is that the rebuild throws away everything the old site had taught you. Three years of accumulated evidence — which pages actually preceded inquiries, what people searched for internally, where they consistently stalled, which explanation finally worked — is mostly encoded in the thing you just deleted. The new site starts from zero, and because it’s new, nobody expects it to be judged for a while. The learning resets on a schedule.
The related trap: redesigning around whatever assets happen to exist. Someone has product photos, so it becomes a visual site. Someone wrote a whitepaper, so there’s a resource centre. The strategy gets assembled backwards out of available material rather than forwards from a diagnosis — and it looks like planning the entire time.
The deeper judgment
Improvement compounds only if you stop resetting it.
This is the real argument against the redesign cycle, and it isn’t about cost. A site corrected continuously accumulates knowledge: every change tested against what was already known, every result narrowing the next question. A site rebuilt every three years discards that and starts again from taste — which is why some companies have been through four websites and understand their buyer no better than they did after the first. Four redesigns is not four times the learning. It can easily be none.
Diagnosis is also where honesty about your own work lives. The uncomfortable finding — that the page you were proud of is where people leave, that the resource nobody championed is on the path to half your inquiries — only surfaces if you look, and looking is optional. That’s what separates improving with wisdom from improving with activity: one of them can tell you that you were wrong, and the other one never will.
And this is where the row closes back on itself. A website is the instrument you use to find out whether you understood your buyer at all — the stalls, the searches, the paths are the answer to the question Understand the Human × Website & Content opened. Skip the reading, and you’ll rebuild the same assumption in better fonts and call it progress.
Underneath it, the same line as everywhere: do not remove expertise from the moments where trust is being formed. A diagnosis is expertise. Handing that judgment to whoever is loudest in the review, or to a tool that reports what changed without knowing what matters, is the substitution this whole framework exists to argue against.
Where this thread goes
Diagnosis tells you what’s broken. Fixing it well is the rest of the column.
- Structure it around their questions — the assumption you’re testing — Understand the Human · Website & Content
- Remove the confusion, keep the friction that has a job — Connect the Experience · Website & Content
- What you count trains what you get, one channel over — Improve With Wisdom, Not Hype · Search & Advertising