Case Study

Case Study: How We Built a 98-Score Bilingual Site for a NY Tech Firm

Noviant Inc. came to us with a 41 Lighthouse score, zero Chinese-language organic presence, and a website that was costing them enterprise clients. Here is the full technical breakdown of how we turned it around.

The Client

Noviant Inc. is an ICT and AI consulting firm headquartered in Midtown Manhattan. Their clients are mid-market and enterprise companies in financial services, healthcare, and logistics. Average contract value: $85,000. Their website — when they came to us — had a Lighthouse performance score of 41 on mobile.

Three Problems Costing Them Real Money

Problem 1: Invisible to Chinese-language search. Approximately 30% of Noviant's target client base — financial services firms in NY, LA, and SF — have Chinese-speaking decision-makers who search in both English and Chinese. Noviant had zero Chinese content, zero Chinese-language backlinks, and zero organic presence in that search segment.

Problem 2: Performance destroying credibility. A 41 Lighthouse score on mobile meant LCP of 6.2 seconds and INP of 480ms. When enterprise procurement teams evaluated Noviant against competitors during RFP processes, a slow website signaled technical incompetence — fatal for an AI consulting firm.

Problem 3: No conversion architecture. The site had no clear calls to action, no case study pages, and no contact flow optimized for enterprise buyers. Traffic existed but converted at 0.4%.

Technical Solution

We rebuilt the site on Next.js 15 with App Router and next-intl for i18n, generating static pages at build time for all locale/route combinations. Key technical decisions:

  • Static generation (SSG) for all content pages — eliminates server latency for the content types that matter most to SEO
  • React Server Components for data-heavy sections — zero JavaScript shipped to the client for sections that do not need interactivity
  • CJK font subsetting — reduced Chinese font payload from 4.2MB to 287KB using glyphhanger and a custom build script
  • Edge CDN deployment via Vercel — assets served from 40+ global edge locations, cutting TTFB from 820ms to 68ms
  • Correct hreflang implementation — alternate tags in the document head for /en/ and /zh-CN/ variants of every page
  • Hero image optimization — WebP with AVIF fallback, explicit dimensions, fetchpriority="high", preloaded in document head

Results After 90 Days

  • Lighthouse performance score: 41 → 98
  • LCP: 6.2s → 1.4s
  • INP: 480ms → 76ms
  • CLS: 0.31 → 0.02
  • Organic traffic: +340% (driven by Chinese-language keyword indexing)
  • Qualified leads per month: +280%
  • Site conversion rate: 0.4% → 2.9%

Three Lessons for Any B2B Tech Firm

Lesson 1: Your website is part of your product demo. For a technology consulting firm, a slow or broken website is a direct contradiction of your value proposition. Fix it before you lose the next RFP.

Lesson 2: Bilingual SEO in a major US metro is low-competition, high-value territory. Most competitors have not touched it. Six months of bilingual content compounds into a defensible organic moat.

Lesson 3: Enterprise conversion architecture is different from consumer. Long sales cycles need case studies, credibility signals, and frictionless contact flows — not hero carousels and stock photography.