Extrusure

Before choosing any software, we mapped Extrusure's full commercial lifecycle end to end: enquiry, technical clarification, supplier pricing, cost calculation, customer quotation, negotiation, order confirmation, manufacturing, logistics, delivery, and payment. We evaluated ERPNext, Zoho, HubSpot, and Salesforce against that map, and concluded the actual problem was broader than CRM: Extrusure needed one system connecting customers, products, quotations, orders, and accounting, not another isolated database. We built the first phase around ERPNext on Frappe Cloud, deliberately limited to what the business needed immediately rather than everything the platform could theoretically do.

Industrial equipment orders can stay open for three months or more, with several deals moving at once. We designed a custom pipeline with more than twenty stages to reflect that, and made a distinction most small-business CRMs miss: a customer's relationship status (enquiry, client, regular client) is a different thing from where a specific deal sits (new enquiry, quotation requested, advance paid). Combining the two into one field would have made the system confusing within weeks.

We rebuilt Extrusure's quotation process inside ERPNext to match the format customers already recognized, rather than forcing a generic template on them. Underneath it, we built the landed-cost calculation logic international trade requires (equipment cost, packing, inland transport, freight, customs and duty, currency), run as consistent business logic instead of a spreadsheet formula rebuilt by hand for every deal. Vendor pricing goes through a structured audit step before it becomes a customer quotation. AI is used to extract, recompute, and cross-check numbers; the final commercial decision stays with a person, because an AI producing a plausible but wrong quotation is a risk we're not willing to take on a client's behalf.

Extrusure's product knowledge lived across operation manuals, historical quotations, and supplier documents. Rather than importing every technically valid part into the system, we separated the complete technical catalogue from the commercially active item master, so the system reflects what customers actually ask for and order, not everything a manual happens to list.

A single pipeline couldn't represent Extrusure's reality: the same customer can have a confirmed order in production, an after-sales issue open, and a new enquiry waiting on a supplier quote, all at once. We built separate tracking for enquiries, larger projects, and confirmed orders, each showing its current stage, the next action, and who it's waiting on, so a live order doesn't go quiet just because it's already been paid for.

We rebuilt Extrusure's website (Next.js, TypeScript, Tailwind CSS, deployed on Cloudflare Workers), covering dozens of product, service, and industry pages built on structured data instead of hardcoded pages. We migrated the domain's DNS and hosting away from a website builder without disrupting the company's live email, a detail that's invisible when it goes right and very expensive when it doesn't. We also moved the company's documents and communication onto infrastructure the company owns, rather than a personal account.

We connected AI tooling directly to the operational system, so records could be inspected, structured, and extended faster, with a person still reviewing anything commercially sensitive. We used the same disciplined approach to reconstruct years of scattered historical business records into a reliable evidence base, keeping confirmed facts separate from probable ones, which now feeds an early management reporting dashboard.