Design as the argument a site makes
B2B website design is the arrangement of pages, words, and evidence so that several people with different jobs can each get what they need from one site and then agree with one another. That is a different craft from visual identity, and different again from information architecture, which is covered under B2B website architecture.
The confusion worth clearing first is between design and decoration. A site can be beautifully executed — considered type, restrained palette, confident photography — and still fail every buyer who visits, because none of those choices answers the question the visitor arrived with.
The test is not whether the design community admires the work. It is whether a technical evaluator can find the integration details, a finance reviewer can understand how you charge, a procurement officer can find your terms and certifications, and any of them can send a page to a colleague and have it make sense without a covering note.
Designing for a committee, not a persona
Business purchases are decided by groups. Gartner's May 2025 survey of 632 B2B buyers describes buying teams of five to sixteen people across as many as four functions, found that around 74% of buying groups show what it calls unhealthy conflict, and reported that groups reaching consensus were 2.5 times more likely to describe the deal as high quality.
- The technical evaluator wants specifications, integration and API detail, stated limits, supported versions, and ideally a way to try something without a sales conversation.
- The economic buyer wants the outcome, the risk, the cost structure, and comparable customers. Feature lists do not help; a case study with a measured result does.
- The procurement or security reviewer wants terms, certifications, insurance, data-processing detail, an accessibility statement, and references. These pages generate no leads and lose deals by their absence.
One finding should change how personalisation gets designed: group-level relevance improved consensus by 20%, while individual-level relevance reduced it by 59%. A site showing each visitor a different story hands a committee three stories to reconcile.
Navigation labelled in the buyer's vocabulary
Most corporate navigation is an org chart with rounded corners. The top level names business units, the second names product families, and the labels are internal terms new employees took a month to learn. Buyers arrive knowing their problem and not your taxonomy.
The symptoms are reliable: a top-level Solutions menu that could mean anything, an Industries section duplicating half of Products, and a mega-menu with forty items and no hierarchy.
Four cheap sources tell you what the labels should be. Site-search logs record the words visitors use when navigation fails them. Sales call recordings and lost-deal notes carry the vocabulary of people who did not buy. And a card sort with eight to ten actual buyers exposes disagreement about grouping in an afternoon.
Then run a first-click test: give someone a realistic task and see whether their first click lands in the right branch. A menu that fails on first click will not be rescued by better copy further down.
Above the fold, and the hero-section fashion cycle
The top of a business homepage has been through several fashions: the full-bleed stock photograph over a slogan, the autoplaying video background, the enormous single line of type, the rotating carousel, and lately the abstract generated render. Each was briefly compulsory.
What belongs in the first screen is duller and has not changed. Who you are. What you do, in words a stranger in your industry would use. Who it is for. One next step.
The fashionable options carry a measurable cost, because the hero is usually the largest element on the page. Google's Core Web Vitals are LCP, INP, and CLS, assessed at the 75th percentile and segmented separately for mobile and desktop, with documented thresholds of LCP within 2.5 seconds, INP of 200 milliseconds or less, and CLS of 0.1 or less. A large hero image or video is very often the LCP element, and a carousel or late-injected cookie banner is a common source of layout shift. Note too that INP cannot be measured by Lighthouse: a perfect Lighthouse score is not the same as passing Core Web Vitals.
Evidence design: what actually does the persuading
In a considered purchase, persuasion is done by evidence rather than adjectives. That means treating evidence as a content type with a structure, not as testimonial decoration.
- Case studies stating the situation, the constraint, what was done, and a result someone could verify. Where the client cannot be named, name the sector and size. A case study that could describe any company describes none.
- Specifications and comparison tables, published rather than gated. Evaluators disqualify suppliers whose limits they cannot find, and they do it silently.
- Pricing transparency, which needs no price list. Publishing the unit you charge by, what drives the number up, and what a typical engagement includes qualifies buyers in both directions.
- Documentation treated as a marketing asset — in technical categories, the most persuasive material a company owns.
Customer logos with no story behind them read as borrowed credibility. Awards matter less than marketing teams believe, with one exception: the certifications procurement is required to check.
Accessibility, as a legal matter and a practical one
Accessibility is measured against the Web Content Accessibility Guidelines, the W3C standard for making web content usable by people with disabilities. In the United States, accessibility claims about websites are commonly brought under the Americans with Disabilities Act, and public-sector and enterprise procurement frequently ask for a conformance report; the specifics vary by jurisdiction and are properly a question for counsel.
The practical argument is stronger anyway, because most of what accessibility requires is what makes a site work for everyone: form fields with real labels and errors that say what to fix, visible focus states for keyboard users, sufficient contrast, headings in a sensible order, captions on video, alt text describing function rather than repeating a filename, and never using colour alone to carry meaning.
Overlay widgets promising compliance from a single script are a bad purchase: they do not fix the underlying markup, they have been widely criticised by disability advocates, and buying one is not remediation. Fix the code. In B2B the commercial argument is procurement — a supplier who cannot produce an accessibility statement loses tenders it never hears about.
Design systems, and why big-bang redesigns fail
A redesign that ships as a single event concentrates every risk on one date. Every URL changes at once, every template is new, every measurement baseline resets, and if conversions drop there is no way to know which of two hundred changes caused it.
The alternative is a design system — a documented set of tokens, components, and page types marketing can assemble without a developer — plus incremental release. Ship the new template for one page type, keep the URLs, measure for a few weeks, then do the next. The programme looks longer on paper and finishes sooner in practice, because it is not blocked on universal agreement.
Two things make it work. Content types must be defined before components, or you build a library that cannot express a case study. And URL changes should be decoupled from design changes; when they are unavoidable, they belong in a planned migration with a mapped redirect for every URL.
Settling design arguments, and the "looks dated" complaint
Design review in most companies is a taste contest the most senior person present wins. The fix is procedural rather than diplomatic: agree in writing, before anyone sees a design, what job the page has, who it is for, and how you will know it worked. Then review against that brief. "I do not like the blue" is not a finding; "an evaluator cannot find the integration list from here" is.
Useful evidence is mostly qualitative, because most business sites lack the traffic for statistical testing. Five-participant usability sessions surface the majority of serious problems. First-click tests settle navigation disputes. Be honest about A/B testing at low volumes: a test needing six months to reach significance is not a decision-making tool.
Treat "the site looks dated" as a symptom, not a diagnosis. It usually means the content describes products you no longer lead with, the mobile experience is broken, pages are slow, or an executive saw a competitor's homepage last week. Each has a cheaper, more targeted fix than a rebuild.
Frequently Asked Questions
What makes a good B2B website?
One that lets several different people evaluate you without help. In practice: application pages written in the buyer's language rather than your product taxonomy, specifications and limits published rather than gated, case studies specific enough to be checked, a clear explanation of how you price, documentation an engineer can use, and a next step appropriate to where the visitor is.
What it does not require is visual novelty. In business categories, good website design for B2B means clarity and evidence rather than craft.
What does a B2B website redesign cost?
Cost is driven by scope rather than design quality, and the drivers are predictable: how many distinct page types must be designed and built, how many pages need rewriting rather than migrating, whether the CMS is being replaced, whether URLs are changing and therefore need a mapped redirect plan, how many languages and regions are in scope, and how many stakeholders hold approval.
The variable that inflates budgets most is content. Teams scope a design project and find halfway through that the pages need writing, which is the larger job.
Should a B2B website show pricing?
Not necessarily a price list, but almost always an explanation of how you price. Publishing the unit you charge by, the factors that move the number, and what a typical engagement includes removes a common reason for buyers to leave and disqualifies enquiries your sales team would have spent weeks on.
The usual objection is competitive exposure. In most categories competitors already know roughly what you charge, and the people the silence deters are qualified buyers wondering whether you were in their range.
How do we stop the CEO overruling design decisions on taste?
Change what is being reviewed. Agree the page's job, audience, and success measure in writing before any design is shown, and structure the review as a check against that brief rather than an open reaction. Bring evidence into the meeting — a first-click test, five usability sessions, the site-search log, the objections sales hears — because taste arguments end when there is data in the room.
It also helps to name who decides. Reviews without a tiebreaker produce designs owned by nobody.