Case Study — Series B SaaS, EU

Customer details anonymized at the customer’s request. Numbers and technical details are real and were verified from inside mainland China before and after the engagement.

The problem

The customer’s global marketing site used a typical European B2B SaaS stack: Webflow CMS, Cloudflare CDN, Intercom for chat, HubSpot for forms, Vimeo for product demo videos, Google Fonts for typography, and Google Analytics + Google Tag Manager for measurement. From London, Frankfurt, or New York the site loaded fully in under 2 s. Their China sales contact, a partner sales director who travels to Shanghai monthly, kept reporting that the page would “sometimes load and sometimes not.”

What we measured from a residential connection in Pudong, on the customer’s public homepage:

• First Contentful Paint: 9.2 s (vs. 1.1 s from London)
• Largest Contentful Paint: not reached within 30 s on most loads
• 14 of 86 third-party requests failed completely. Among them: Google Fonts (CSS file blocked), Vimeo embed (player JS blocked), Google Analytics (collect endpoint blocked), HubSpot Forms (script blocked), Intercom (widget blocked), reCAPTCHA (challenge blocked), 2 of their 3 Cloudflare-hosted product images (significant tail latency).
• Page rendered without primary fonts (text in fallback Arial), without product video (gray box), and without working “Contact Sales” form (HubSpot loader hung).
• The Chinese sales lead testing it described it as “empty and broken-looking.”

What we did

Step 1 — Audit and architecture (days 1–3)

Multi-region testing from Shanghai, Beijing, Chengdu, and Guangzhou confirmed the same failure pattern. We chose Path C — the customer didn’t have a mainland China entity, didn’t want to incorporate one, and didn’t need to reuse their existing domain for the China launch. We allocated a subdomain on a domain we own, hosted under our mainland China entity, with our company as the operating entity of record. ICP filing was filed under our entity.

Step 2 — Mirror & resource adaptation (days 4–10)

We mirrored 18 of their roughly 30 marketing pages (homepage, 6 product pages, 4 industry pages, pricing, 3 company pages, blog index, 3 blog posts they wanted preserved). For each blocked or slow third-party resource, we made a deliberate choice:

• Google Fonts → self-hosted as WOFF2 inside the mirror, preloaded on critical pages.
• Vimeo product demo → transcoded to MP4, served from China CDN with a small custom HTML5 player.
• Google Analytics + GTM → replaced with a minimal first-party analytics endpoint that the customer’s data team could later mirror back to GA via server-side import.
• HubSpot Forms → replaced with a static form posting to a small bridge service that we operate; submissions are forwarded to HubSpot via API on the customer side.
• Intercom widget → removed from the China site (intentional choice: their China sales preferred email anyway).
• reCAPTCHA → replaced with a lightweight server-side honeypot, sufficient for the volume.
• Images → pulled to China CDN, converted to WebP, sized appropriately for typical screen sizes.

Step 3 — Localization (days 8–14)

The customer chose to do light localization, not a full translation: the homepage hero, pricing page CTAs, and the contact page were translated. Everything else stayed English to match the customer’s brand voice. We built it as a per-page toggle so the customer can later add Chinese to additional pages without re-engaging us.

Step 4 — Deploy & verification (days 15–18)

ICP filing came back approved on day 13. We pointed DNS, switched to HTTPS, and ran a final verification from four mainland regions. Final numbers, from Pudong residential 4G:

• First Contentful Paint: 1.4 s (down from 9.2 s — 6.6× faster)
• Largest Contentful Paint: 2.1 s (was unreachable)
• Failed third-party requests: 0 (was 14)
• Page render: complete, with correct fonts, working video, working contact form, and working analytics

What changed for the customer

The customer agreed to share the following business outcomes from the first 60 days post-launch, on the condition we wouldn’t name them publicly:

• The China sales contact stopped having to email PDF brochures because the site “works again.”
• The China-region partner kickoff used the new URL on its slides; the URL was forwarded to 23 prospect attendees.
• 7 inbound inquiries arrived via the new China-region contact form in the first 60 days. The customer estimates 2 of these are real opportunities at deal sizes that exceed the entire project cost in year one.
• One downstream effect: the customer’s Chinese partner now uses the URL on a public sales page, which led to additional referral traffic the customer didn’t plan for.

What we didn’t do

We didn’t touch the customer’s global Webflow site. We didn’t migrate their CMS. We didn’t do paid Baidu campaigns or any kind of outreach — the gain came purely from fixing the technical reality of opening a URL from inside China. We also did not localize the entire site; we localized exactly enough to support a Chinese-speaking partner’s sales conversations.

What this looks like commercially

The customer paid the standard price: USD $5,000 setup plus USD $500/month thereafter. Total first-year cost including 12 months of operations: USD $11,000. The customer renewed the Hosting and Operations subscription at month 12 with no re-quoting. We continue to push small content updates monthly and add 1–2 new product pages per quarter without an additional fee.

Could you get similar results?

The honest answer is: probably yes if your site is similar in shape (informational marketing site, mainstream third-party stack, no login/payment/UGC), and probably less dramatic if your site doesn’t lean as heavily on third-party scripts. Best way to find out is the same way this customer started: