We rebuilt the landing site for a 12-person Hungarian B2B SaaS startup on Next.js with static generation, hreflang HU/EN and PostHog A/B testing. Free-trial conversion went from 2.1% to 3.4% in 6 weeks.
Background
The client is a 12-person B2B SaaS startup, an IT-management workflow tool targeting Hungarian and English markets. The product is mature; sales has closed 200+ customers, but the self-serve free-trial flow under-performed.
The existing landing ran on WordPress: 6.2s mobile LCP, Lighthouse 38 on mobile. hreflang wasn't set up; the EN page duplicated the HU one and search engines saw it. Free-trial conversion was 2.1% (signup / landing visitor), while sales believed 3–4% was reasonable.
Challenge
It wasn't only a technical problem. The landing copy and visual structure felt tired: a feature-listing presentation, 4–5 generic testimonials, and a fixed CTA style. The CMO had tried internal A/B testing, but the WordPress platform didn't make it easy.
A critical scope point: sales couldn't pause for 8 weeks of full redesign. They needed lead generation throughout. So the redesign ran on a parallel track next to the existing site, with cutover on the final day.
Approach
We ran two streams: 1) Next.js redesign on Sanity headless CMS, so the CMO could edit content directly. 2) PostHog A/B test infrastructure stood up so post-launch iteration would be fast.
Scope deliberately didn't include the entire site — only landing, pricing, signup flow, and 3 product feature pages. The blog (40+ articles) stayed on WordPress under a subdomain, with hreflang and internal linking sorted. That saved 6 weeks of work.
Solution
The Next.js landing uses the App Router with a mix of RSC + Client Components. Copy and visual content live in Sanity; the CMO edits with real-time preview. The free-trial signup is a 3-step flow (account, company, password) with inline validation and a progress indicator.
PostHog feature flags test in 4 dimensions: hero headline (3 variants), hero CTA copy (2 variants), pricing display (monthly default / annual default), social proof (logo wall / testimonial grid). The full matrix is 24 combinations, but the initial rollout went on 4 main variant groups to get meaningful samples.
Tech decisions
We chose Sanity over Strapi or Contentful because the CMO preferred the real-time preview experience and the Hungarian localization is supported natively. WordPress + WPGraphQL was an option, but the CMO wanted to move to new infrastructure.
We preferred PostHog over Mixpanel and Amplitude: public EU host is available, GDPR compliance is simpler, and feature flags and session replay live on one platform. Plausible analytics ran alongside for privacy-friendly pageview tracking (which the sales team was used to).
Bumps
First hurdle: hreflang sorting for the existing 40+ blog articles. The original WordPress URL structure mixed HU and EN (e.g. /en/blog/post-1 and /blog/post-1-en both pointing to the same content). We built a 301 redirect matrix, but a few external links 404'd for 6 weeks.
Second: A/B test sample size. Free-trial conversion was low (2.1%), so reaching statistical significance took 2–3 weeks of traffic per variant pair. Frustrating for the CMO, but we stuck to the rule: don't expand the variant count until the first set closes.
Third: aligning Sanity content with Next.js static generation. CMO edits → ISR webhook → new build trigger. Latency was 30–60 seconds, which felt too slow for the 'preview' experience. We added a draft-only preview mode that runs in real time on Vercel preview; production only rebuilds on commit.
Results
After 8 weeks free-trial conversion went from 2.1% to 3.4% (PostHog data, statistically significant). The three winning variants: 1) hero headline 'IT operations automation' → 'Replace your IT spreadsheets' — +0.6 pp. 2) Pricing default monthly → annual — +0.4 pp. 3) Social proof: logo wall → 3 testimonial cards with concrete numbers — +0.3 pp.
Lighthouse mobile score went from 38 to 94; LCP from 6.2s to 1.6s. Organic search traffic grew 32% in 3 months (Search Console), especially on long-tail 'IT ops automation' queries. The Sanity editing workflow: the CMO originally pushed every 1–2 weeks; now 3–4 new landing variants per week.
Stack
Previously the landing redesign was planned 'big bang'. Now we run weekly A/B tests and can put numbers behind every iteration. It's a different operating model.
CMO — B2B SaaS
The quote represents typical client feedback; it is not a verbatim citation.
Related
We replaced manual lead handling and email-based customer flows with a Next.js webshop on a WooCommerce catalogue and an n8n + OpenAI lead classifier. Sales recovered 30–40 minutes of daily admin per person.
Read detailsA RAG chatbot with pgvector and OpenAI GPT-4o, integrated into an existing Shopify storefront. 38% of support tickets close automatically; customer satisfaction didn't drop.
Read detailsCase studies are based on real projects, anonymized at our clients' request. Numbers are approximations; specific business data remains confidential.