The brochure is finished
Most corporate websites are still built on an assumption from around 2012: that the site is a brochure, that its job is to explain, and that success looks like a visitor reading a page and forming an impression.
That model has not collapsed so much as been quietly outgrown. The people arriving at your site have already read about you somewhere else. They have seen the comparison thread, asked an assistant to summarize what you do, and formed most of their impression before the first paint. By the time they arrive, they are not browsing. They are checking.
A site built for checking looks different from a site built for browsing. It answers rather than introduces. It makes the specific thing findable rather than making the whole story navigable. And it holds up under a kind of scrutiny that brochures were never designed for.
Three shifts worth building for
1. The page stops being the unit
For years the atomic unit of a website was the page, and a redesign meant rearranging pages. That breaks the moment content needs to appear in more than one place — the same case study on a service page, a landing page, a proposal, an internal deck, and increasingly inside an answer generated somewhere you do not control.
The unit is now the content object: a case study, a role, an insight, a service. Each with real fields, its own lifecycle and a canonical home. Pages become views over that data. This is why the shift from hand-built templates toward structured content has quietly become the most consequential architectural decision in web projects — not because it is fashionable, but because it is the only way the same fact stays correct in eight places at once.
The test is simple. If updating one fact means editing more than one thing, your content is not structured yet.
2. Performance is design, not maintenance
Speed is still routinely handed to engineering as a clean-up task after the design is signed off. That is backwards. Every meaningful performance number is decided by design choices: the hero video, the font stack, the third-party widget, the carousel that ships an animation library to move three images.
Core Web Vitals made this legible, and the metric set has sharpened over time — interaction latency replaced first-input delay as the responsiveness measure in 2024, which moved the pressure from "does it load" to "does it stay usable". That is a design conversation. A layout that shifts while loading is a design failure. A button that takes a third of a second to acknowledge a tap is a design failure. Filing them as engineering debt guarantees they come back with every redesign.
The practical move is to give performance a seat in the design review, with a budget expressed in the same terms as everything else: this template gets this much weight, and a new component that exceeds it has to displace something.
3. Your site has two audiences now
The second audience is machines — search crawlers, assistants, and increasingly agents acting for a person who may never see your layout at all.
This is not a reason to write for robots. It is a reason to be structurally honest. Machines read what people read, but they only read what is genuinely in the document: content rendered on the server rather than assembled after three client-side round trips, semantic headings that describe the real hierarchy, explicit structured data for the things that have a defined shape — organizations, articles, job postings, products — and stable, meaningful URLs.
The overlap with human accessibility is nearly total, which is the useful part. A page a screen reader can navigate is a page a machine can parse. In the EU that overlap now carries legal weight: accessibility requirements for a broad range of consumer-facing digital services became applicable in June 2025, and the baseline they point at is the one good engineering already implied.
What this means for a redesign
A redesign framed as "refresh the visual language" produces a site with the same problems in a nicer typeface. The questions worth asking are structural.
- What is the site actually for? Not "to represent the brand" — which specific decision should it move forward, and for whom?
- What are the content objects? Name them, define their fields, decide who owns each one.
- What is the release cadence? A site that ships once every three years is a brochure. A site that ships weekly is a product, and needs a product's tooling: previews, staging, analytics somebody actually reads.
- What is the performance budget? Per template, agreed before the first mockup.
- What can be deleted? Almost always more than anyone is comfortable with. Most sites carry pages nobody has opened in two years, at a real cost in maintenance and navigational clarity.
Personalization, carefully
Personalization is the most over-promised idea in web strategy, mostly because it is attempted at the wrong layer: surveillance in exchange for a slightly different headline.
The version that works is coarse and honest. Someone arriving from a technical comparison sees the technical page first. A returning visitor who has read three articles on one topic gets the next one surfaced instead of the general introduction. A visitor in a given market sees the pricing and legal context for that market. None of that requires an identity graph. It requires content structured well enough to recombine — the same requirement as everything else on this list.
Measure fewer things, better
Most site analytics collect a great deal and inform nothing. A small set actually tells you whether the site works.
- Time from arrival to first meaningful action, rather than time on page.
- Completion rate for each journey you have deliberately defined.
- Which content objects appear in the paths that end well, and which appear in the paths that end nowhere.
- Real-user performance at the slow end of the distribution, not the average.
The average visitor does not exist. The one at the 75th percentile does, and they are the one deciding whether your site feels fast.
The evolution is structural
The next evolution of your website is not a new visual direction. It is the move from a document you publish to a system you operate: structured content, budgeted performance, machine-legible markup, an honest measurement set, and a cadence that lets you change your mind without commissioning a project.
The visual refresh will still happen, and it should. It will simply be the least interesting part of the work.
