CPC Scientific

CPC Scientific, the peptide and oligonucleotide CRDMO whose parent company is listed on the Hong Kong Stock Exchange, hired me to bring a struggling WooCommerce site onto real infrastructure: modern hosting, a security edge, disciplined caching and a deployment pipeline, with measurable cost savings to show for it.

Key Takeaways

  • Cut server bandwidth from 7 to 13 GB a day to about 1 GB a day, verified in the client's own hosting analytics, keeping the site comfortably inside its plan.
  • Diagnosed traffic problems from evidence: access logs loaded into BigQuery and queried by IP, user agent, path and hour, so every remediation targeted a measured cause.
  • Put Cloudflare in front of the site with explicit cache rules for static assets and page HTML, edge protections on signup and login paths, and modern protocol support.
  • Built a real environment chain, local, staging and production with zero-downtime deploys, where previously all changes went straight to the live site.
  • Published structured data appropriate to a regulated manufacturer, expressing the company's classification and parent relationship for search and AI systems.
  • Added AI-assisted lead classification that scores inquiries for the sales team without ever sitting in the critical path of a submission.

Organization

CPC Scientific Inc. is a Contract Research, Development and Manufacturing Organization specializing in peptides and oligonucleotides. Founded more than two decades ago, it is headquartered in Milpitas, California, with a US facility in Rocklin and cGMP manufacturing in Hangzhou, China. Projects run in over 50 countries, and its parent company, Medtide Inc., is listed on the main board of the Hong Kong Stock Exchange.

Challenge

Traffic to the site had grown far faster than the infrastructure under it, and the hosting model priced that growth as steep overage. The stack had accumulated overlapping caching layers, an aging plugin inventory and no environment chain, so improving anything meant first making the system observable and safe to change.

Solution

  • Access-log analysis in BigQuery to establish exactly where traffic and load were coming from before changing anything.
  • Cloudflare edge configuration: static asset and page HTML cache rules with logged-in and commerce sessions excluded, protocol upgrades, and protections on the signup path.
  • Hosting consolidated and the plugin stack rationalised, with caching responsibilities assigned to one layer.
  • A local environment, staging site and zero-downtime deployment script.
  • Organization structured data emitted sitewide, with the NAICS classification verified and the parent-company relationship expressed explicitly.
  • AI lead classification for the contact pipeline, with a scoped customer-service admin role so the sales operation works from exactly the panels it needs.

Results

  • Bandwidth down roughly 90%, with product-category pages verified serving from the edge.
  • Server load returned to normal operating levels.
  • 5 to 10 sales hours a month returned through lead classification.
  • Qualified lead identification up an estimated 25 to 40%.
  • Organic and AI-assistant visibility up an estimated 15 to 30%.