A while back, I had two client conversations, one day apart.
The first was with a marketing lead whose organization had just launched a beautiful new website. Genuinely beautiful — stronger brand, better photography, more emotional. Also slower, less usable, and worse for SEO than the site it replaced.
The second was with a comms director who told me, flat out: “I don’t even want people trying to look at our website.” She was routing traffic to third-party tools specifically to keep visitors away from her own site.
Two organizations, opposite ends of the redesign cycle, same root problem. And it starts long before anyone opens Figma.
The portfolio trap
One of the quietest reasons redesigns fail is how the agency gets chosen.
Most organizations pick a web agency the same way they’d pick a graphic design firm: flick through the portfolio and ask, “Do we like how this looks?”
Which makes sense. Right up until you realize a beautiful portfolio proves exactly one thing — the agency can make things beautiful.
Here’s the problem: a website isn’t a poster. It’s software that happens to look like a poster.
A poster has one job — look good and say the thing. A website has a dozen jobs running at once: performance budgets, semantic structure, accessibility, SEO, AI-search visibility, analytics, conversion paths. None of which show up in a portfolio thumbnail.
When I asked that first marketing lead what the website goals were, the answer was honest: “It was a more emotional and artful goal.” The brand had been refreshed, the design firm was a trusted long-term partner, and the website simply followed along as a visual update.
Did anyone set performance targets? “We didn’t probably set performance goals at the beginning… we’ll probably live to regret some of that.”
That’s not a bad client or a bad agency. That’s someone hired for the wrong job — and nobody noticing until launch.
Where redesigns actually go wrong
1. The assumption gap
Clients assume the agency will “handle the technical stuff.” The agency assumes the client mostly cares how it looks.
The most telling thing I heard on that call: “I assumed they would just take care of it without us having to talk about it.”
And honestly — why wouldn’t you assume that? You hire someone who says “I build websites” and you figure that includes how websites are supposed to work. As she put it: “I don’t even know to ask for that.”
But here’s the catch: your design team is trying to please you. They’ll do what they think will make you gasp when the homepage loads. They won’t necessarily check you and say, “you can have this — at the expense of X, Y, and Z.”
So nobody writes down speed targets, SEO requirements, accessibility standards, or AI readiness. The brief covers fonts and feelings. The software requirements are vibes.
Translation: what isn’t specified can’t be enforced.
2. The most expensive lie in web design
The same client said something I hear constantly: “As clients, we think it’s one lane or the other. I can have a beautiful site or a functional site.”
So teams choose beauty, because beauty has a name: “brand.” Then they’re surprised when the new site is slower, harder to use, and slipping out of search.
That’s not a trade-off you were forced into. That’s a governance gap nobody noticed. You can absolutely have both — but only if both are in the requirements before design starts.
And the stakes are higher than they used to be, because every website now serves two audiences: humans, who need clarity and a path forward, and machines — search engines, screen readers, and AI systems that need structure and meaning. People don’t browse the way they did; they search, and increasingly they get AI-generated answers. If your headings are out of order and your structure is a mess (I recently looked at a major campaign site where every single heading was an H1), the machines can’t make sense of you — and the humans never find you. A site that delights people but confuses machines disappears. A site that satisfies machines but confuses people doesn’t convert. You don’t get to pick one.
3. The organization changed. The site didn’t.
The second conversation was the other end of the cycle: the redesign that’s overdue.
New leadership, 18 months in. New strategy. New programs. New language. The website? Frozen in time — built exactly right for the people who were in the seats when it launched, and wrong for everyone who came after.
Her words: “You need to be real smart to figure out how to even use that website.”
The site had grown to nearly 300 pages of posts, programs, and page types. Layer upon layer of complexity for an organization whose actual offer, in her own assessment, “is not that complicated.” The new target? Under 100 pages. Maybe under 50.
That’s not a website. That’s a museum of previous strategies. And like most museums, nobody internally trusts it — which is when the workarounds begin.
4. Tool sprawl ate the website
Here’s what those workarounds look like: forms on one platform, events on another, intake routed to a single landing page “because I don’t even want them trying to look at the rest.”
One page — their top landing page, no less — had two different third-party forms on it, almost no content, and average visits of about two seconds. Unbranded forms shipping visitors off to who-knows-where. A page still collecting traffic through a form connected to a platform the organization had stopped using entirely.
Your visitors feel it, even if they can’t name it. The journey fragments, the brand disappears, and the website stops being the front door and becomes the thing you route people around.
5. Nobody owns what happens after launch
Most redesigns aren’t ruined during the build. They’re ruined in the year after.
Pages get hidden instead of deleted — and nobody deindexes them, so outdated content keeps surfacing in search and AI answers long after you’ve disowned it. The blog stops the day the last person who cared leaves. Old ads keep driving paid traffic to abandoned landing pages. No publishing rules, no owner, no review cycle — just slow rot, until you’re back at the top of this article choosing another agency from another portfolio.
A redesign is a decision-making exercise, not a design exercise
If you want the redesign to actually improve things, settle these before anyone opens design software:
- What job does this website perform for the business?
- What must a first-time visitor understand — and how fast?
- What are the performance and SEO requirements, in writing?
- What content gets deleted (not hidden, deleted)?
- Which tools stay, and how will they feel like one site?
- Who owns governance after launch?
Most of those get settled in a content outline — the page-by-page plan of what the site needs to say, written before design starts.
Beauty without clarity is art. Clarity without structure is a pamphlet. Structure without beauty is engineering.
A real website needs all three — and 93% of failed redesigns skipped at least one. (Fine, I made that number up. But you’ve seen the sites. You know.)
If you want the other half of this — the five things a redesign should never make worse — I wrote that one too.
Here’s the encouraging bit: both of those clients ended their calls in the same place — not defensive, but curious. “Don’t pull your punches,” one told me. “It can only improve the outcome for people.”
So I won’t. If you’re planning a redesign and want someone to pressure-test the brief before you sign anything — that’s a conversation I genuinely enjoy. Book a quick chat and bring your list of requirements. Or your lack of one.
