Futureman Labs
All case studies

Construction / Manufacturing

Running the Ops Stack for a Brick-Cladding Manufacturer

A brick-cladding manufacturer ran sales out of one rep's inbox and the factory out of a whiteboard and one person's memory. Over four months the website, the lead funnel, the sales layer over the CRM and a production app were built and are now run week to week. This engagement is active.

4 to 10 a day
Inbound web inquiries
345
Quotes issued since Aug 2025
Whiteboard and memory
Scheduling before
Weekly, ongoing
Cadence

Client

The same Australian engineered-cladding manufacturer, seen across the whole business rather than the pipeline alone. The product is sold to architects, engineers, builders and installers rather than delivered as a contracting service, and the team is small: a founder, an operations lead, one business development manager, and a handful of factory and marketing staff. This study covers the four-month build across the website, the lead funnel, the sales layer and the factory floor. Anonymized, active engagement.

Stack

XeroClose CRM, later Cloze CRMVercelSupabaseResendGoogle Chat botWhatsApp via UnipileClaude CodeGA4 and Search Console

The problem

Two halves of the business were running on the same fragile pattern: everything important lived in one person's head. On the sales side, the single business development manager's inbox was the source of truth for every project, so the rest of the team had to call him for an update and the CRM was filled in late, by hand. On the factory side, the cutting queue, the colour sequencing that stops one brick contaminating another, and the staggered material orders were all held by the operations lead, mostly on scrap paper. An earlier off-the-shelf app had been tried and abandoned, used for nothing but a basic cut list. The website added to it: slow to change, missing the technical information buyers wanted, and with no analytics behind it.

What we built

Four connected pieces, shipped in sequence rather than as one big launch. A new website on a custom stack, with gated technical guides, a labour and materials calculator, and a chat bot the team can update by messaging a bot in Google Chat. A sales layer over the CRM that reads the rep's email and CRM through their APIs, answers questions with links back to the source email, and runs a nightly scan that proposes CRM updates into a review tab for a human to approve before anything is entered. Automated nurture sequences by lead type, sent from a separate subdomain, so the rep gets one notification per lead instead of answering the same questions by hand. And a production app that turns accounting quotes into cutting, order and packing lists, schedules jobs on a Gantt, logs daily cut quantities, handles goods-in and timesheets, and gives floor staff a view with no pricing in it.

What changed

Inbound went from roughly one inquiry a day with no advertising to between four and ten web inquiries a day, with the rep counting nine in a single day in August. The technical manual flow now attracts architects and takes the repetitive phone calls off the rep. CRM updates moved to daily or twice daily, which is what made pipeline analysis trustworthy in the first place. The production app is in daily use by the operations lead, a full yard stock take means on-hand inventory is accurate, and every won job now lives in it rather than in CRM stages that were never meant for production. Timesheet approval runs on a weekly email link through to payroll.

The takeaway

This is what an embedded engagement looks like when it works: not one system delivered and handed over, but a weekly cadence where the next most expensive manual step gets removed, then the next. The pieces compound because they share the same data. The measure is not how impressive any single build is, it is whether the business still needs a particular person to be reachable before anything can happen.

If you've got 1% of stuff that needs any type of manual brain power, it's better than the whole process.

Think you have a similar system half-built, or want one built and run for you?