CaseStudies

The challenge

Alcantara Safety Solutions needed more than a brochure site. As a one-person consultancy selling custom safety documentation — WorkSafeBC-aligned safety plans and programs for B.C. construction, landscaping, warehousing, and civil contractors — the founder’s growth bottleneck was prospecting: finding the right contractors, keeping track of who had been contacted, and following up consistently, all while doing the actual client work.

The project had three distinct problems to solve: visibility (deep, genuinely local SEO coverage across the communities the business serves, not a single generic services page), pipeline (a usable database of real prospects with enough context per lead to make a first call relevant), and workflow (a simple, private system the owner could run daily from a phone or laptop without drowning in CRM complexity).

A 60+ page marketing site engineered for local search

• ✦20 dedicated city pages, each with unique content, FAQs, and structured data (JSON-LD) targeting searches like "safety consultant in [city]" — not doorway pages, but genuinely localized content.
• ✦Full static site on Vercel with custom security headers, immutable asset caching, and a sitemap covering every URL.
• ✦Conversion-focused contact flow, kept deliberately simple and untouched once it was working.

A private leads engine (custom CRM)

A full-stack application reachable at the client’s own domain, with everything a one-person sales operation needs and nothing it doesn’t:

• ✦An 8,930-lead pipeline sourced from public B.C. records (construction, landscaping, warehousing, civil & excavation), each lead carrying company, contact, phone, city, and a plain-language rationale for why they’re a fit.
• ✦A daily work queue — instead of staring at a database, the owner gets a prioritized list: highest-priority leads first, oldest untouched first, with one-tap actions to call, mark contacted, schedule a follow-up, or snooze a lead for three days.
• ✦Bulk operations — select any set of leads to change status or clean up the pipeline in one action, with an activity log that records every change automatically.
• ✦Lead enrichment — per-lead research helper that opens targeted lookups (company contact info, LinkedIn, local directories), plus dedicated website/LinkedIn fields with an automatic "enriched" timestamp.
• ✦Saved filters — any filter combination (e.g. "high-priority Vancouver construction") reloads in one click.
• ✦Full CSV export of any filtered view — the entire pipeline, not just the current page.
• ✦A 20-template outreach kit covering the full cycle: cold emails, follow-up sequences, phone scripts (including a gatekeeper script), voicemails, LinkedIn messages, referral outreach, seasonal campaigns, and proposal follow-ups. All editable in place — the owner can add template #21 without calling us.

Security done properly

• ✦Owner-only access with email/password authentication (secure scrypt hashing), session cookies hardened for cross-domain use.
• ✦Role-gated API — every data endpoint requires authentication.
• ✦Standing rules throughout the engagement: no test enquiries to the live business, no invented claims on the marketing site, no credential exposure in deliverables, no deploys without authorization.

The technical story

• ✦Static + full-stack hybrid architecture. The marketing site stays a fast, free-to-host static deploy on Vercel; the leads engine’s frontend is served from the same domain while its API proxies to a hosted backend. The owner visits one URL; the complexity stays invisible.
• ✦A Git/CI rescue. Mid-project, the hosting account’s Git connection had silently severed — pushes deployed nothing. We diagnosed it from deployment metadata, reconnected the repository programmatically, and restored automatic builds.
• ✦A framework-version landmine. The admin UI shipped with Tailwind v4-only syntax inside a v3 project, silently breaking layouts. We traced it through rendered CSS, converted the utilities, and verified the fix visually.
• ✦A stale-deployment failure mode. When the hosted backend stopped refreshing on republish, we stopped depending on it for the frontend — moving the app shell to the client’s own infrastructure and keeping only the API proxied. Single point of failure removed.