The false trade-off
There is a stubborn belief in our industry that creative websites and SEO performance are in tension. The argument goes: SEO demands fast pages, semantic HTML, lots of indexable content, and conventional structures. Creative work demands custom layouts, animation, hero typography, and visual ambition. The two cannot fully coexist, so you pick a side.
We think this is wrong. The best creative sites we have shipped in the last two years also rank in the top three for their target keywords. The trade-off is illusory if you treat SEO as a craft constraint from the first sketch, not as a finishing layer applied at QA.
What actually moves the needle in 2026
Search has changed enough since 2020 that a lot of "SEO best practice" content is now outdated. Here is the short version of what matters today:
Core Web Vitals are non-negotiable. Largest Contentful Paint under 2.5s, Interaction to Next Paint under 200ms, Cumulative Layout Shift under 0.1. Pages that fail these get visibly demoted in the SERP, especially on mobile. The good news is that a properly built Next.js site clears these effortlessly. The bad news is that a creative site full of unoptimised hero videos and unbounded animation triggers will not.
Search intent is the primary signal. Google has gotten extremely good at matching pages to intent. A landing page targeting "best CRM for solo consultants" needs to actually be the best page on the internet for that query. It cannot just contain those words; it has to answer the question. Listicles that summarise five options outrank single-product brochures even when the brochure is better-written.
E-E-A-T (Experience, Expertise, Authority, Trust) is what they look for in content. Pages that demonstrate first-hand experience — author bios with credentials, original photography, dated case studies, named customers — outrank pages that read like generic AI output. This is one of the few cases where a creative agency has a structural advantage; we make first-hand artefacts for a living.
Schema markup is leverage. Structured data unlocks rich snippets, knowledge-panel placement, and better AI search visibility. A site with proper Article, BreadcrumbList, FAQPage, Organization, and Service schema is dramatically more legible to search engines than the same site without it.
LLM search visibility is a real channel now. ChatGPT, Perplexity, and Google's AI Overviews answer a meaningful share of queries without sending traffic. Pages that get cited inside those answers are the new top-of-funnel. The optimisation here is partly old (clear semantic HTML, well-structured headings) and partly new (llms.txt, content that's quotable in passages, factual claims that pair with sources).
How we bake SEO into creative work from day one
The trick is to treat SEO not as a layer added at the end but as a constraint that shapes the design from the first wireframe. A few practical examples:
Headlines do double duty. The hero <h1> is both a piece of brand expression and a search-optimised heading. We never let designers write a hero headline without checking it against the keywords for that page. "We make digital things" is a designer's hero. "Senior brand and web design subscription for ambitious teams" is the same hero rewritten for both impact and search.
Animations are accessible. Every animation we ship can be paused or skipped, loads after the page is interactive, and respects prefers-reduced-motion. This is good for users and good for search — Google penalises animations that block paint or hijack the main thread.
Images carry alt text and meaningful filenames. A hero photo named hero-final-v3-FINAL.jpg is invisible to image search. A hero photo named figma-design-tools-workspace.jpg with a thoughtful alt text is indexable. We rename every asset before it hits the repo.
Internal linking is intentional. Every page links naturally to two or three related pages elsewhere on the site. Not in a "related content" widget at the bottom — inside the prose itself, where the link is contextual. This is how authority compounds across a site rather than concentrating on the homepage.
Sitemaps and robots.txt are not afterthoughts. Every site we ship has a generated sitemap, a thoughtful robots.txt, an llms.txt for AI crawlers, and IndexNow integration so Bing and Yandex know about new content within seconds of publishing.
What we measure
We instrument every site we ship with three SEO health checks running on a schedule:
Lighthouse CI in the deployment pipeline. A regression in Performance or Accessibility scores breaks the build before it ships.
Core Web Vitals field data via the Chrome User Experience Report. Lab metrics tell us about the platonic page; field data tells us about the actual experience our real users are having.
Search Console weekly snapshot. Impressions, clicks, average position by URL, segmented by query. We watch for sudden drops, which usually mean a recent change broke indexability.
We do not chase ranking changes obsessively. The job is to ship a site whose pages deserve to rank, and to make sure no technical accident is preventing that. Every site we have built this way has crossed page one for its target keywords within six months of launch. There is no magic here — there is only the discipline of treating search as part of craft, not as marketing's problem after the site goes live.
The new front
The next year will likely shift more search volume to LLM-powered answer engines. Sites that get cited there will compound traffic and authority faster than sites that just rank on classic Google. Optimising for AI citation is not a different skillset — it is the same skillset (clear prose, named claims, well-structured pages) applied with the awareness that the reader might be a model rather than a human. The brands that show up in the AI answer are the brands that the AI has decided to trust. That trust is built the same way human trust is built: clarity, expertise, and time.