Case Study: How Meridian Property Group Cut Work Order Prep from 2 Days to 20 Minutes

Every new Meridian project began the same way. The architect and engineer would deliver a PDF plan set — typically 80 to 120 pages — and the PM would sit down and read through every page, manually extracting scope items for each trade.

The extraction process alone took a full day. The PM would work through the architectural drawings, the structural plans, the hydraulic drawings, the electrical layout, and the mechanical specifications — identifying every item that needed to be actioned by a contractor, noting the relevant trade, and recording it in a spreadsheet.

Day two was work order production. Each scope item was written up into a formal work order in Word, formatted to Meridian's standard template, reviewed, and emailed to the relevant contractor. Responses were tracked in the same spreadsheet. Unconfirmed work orders were followed up by phone.

For a typical 12-townhouse project with 47 trade work orders, this process consumed approximately 16 hours of PM time — two full working days. With 12 active projects, the PM was spending 24 days per project cycle on work order preparation alone.

The downstream effects were significant. Because work order preparation was so time-consuming, it was often delayed — sometimes by a week or more after plans were received. That delay pushed contractor start dates back, which compressed the construction schedule, which increased pressure on the PM to cut corners on the next round of work orders. It was a self-reinforcing problem.

The fundamental issue with manual plan extraction isn't effort — it's cognitive load. Reading 100 pages of technical drawings while simultaneously identifying, categorising, and recording every scope item is mentally exhausting work. Fatigue leads to missed items.

Meridian's PM estimated that manual extraction had a miss rate of approximately 6% — meaning roughly 3 scope items per project were either missed entirely or incorrectly attributed to the wrong trade. Each missed item typically surfaced as a variation during construction, with average variation costs of $3,500–$8,000 per item depending on the trade and stage of construction.

Across 12 projects per year, that's 36 variations attributable to plan extraction errors — at a conservative average cost of $4,000 each, that's $144,000 per year in variation costs that were directly traceable to the manual work order process.

This wasn't a productivity problem. It was a structural accuracy problem that no amount of care or effort could fully solve at scale.

Meridian implemented BuilderHQ's AI Plan Reading module in March 2025. The onboarding process was completed in 3 days without disrupting any active projects.

Day 1 — Platform setup and configuration: BuilderHQ's onboarding team configured the platform to Meridian's existing work order format, imported the contractor register (43 contractors across 12 trades), and set up the confirmation workflow with Meridian's standard 48-hour response window and two automated reminder sequences.

Day 2 — Validation run on a completed project: Before going live, the platform was tested against a recently completed project. The AI processed the original 94-page plan set and its output was compared against the actual work orders that had been issued. The AI identified 46 of the 47 original scope items (one minor electrical item had been added verbally during a site visit and was never in the plans). It also identified 2 scope items that had been missed in the original manual extraction — both of which had subsequently become variations.

Day 3 — PM training and first live project: The PM received a 2-hour training session covering plan upload, output review, work order editing, and the confirmation tracking dashboard. The first live project — a 12-townhouse development in Manly — was processed the same day. The AI extracted all 47 scope items from the 94-page plan set, generated work orders in Meridian's standard format, and sent them to contractors. Total time from plan upload to instructions sent: 22 minutes.

The most common question Meridian's PM gets from other developers is: 'What does the AI actually do with the plans?'

The process has three stages.

Stage 1 — Document parsing and classification: The uploaded PDF is parsed page by page. Each page is classified by drawing type — architectural, structural, hydraulic, electrical, mechanical, landscape, or specification. This classification determines which extraction rules apply to each page.

Stage 2 — Scope item extraction: Within each drawing type, the AI identifies elements that require contractor action. For architectural drawings, this includes framing, cladding, roofing, windows, doors, and finishes. For hydraulic drawings, it identifies drainage runs, fixture locations, and connection points. For electrical, it identifies board locations, circuit runs, and fixture schedules. Each identified item is tagged with a trade category, a location reference (lot number, level, room), and a specification note drawn from the relevant schedule or note on the drawing.

Stage 3 — Work order generation: Extracted scope items are grouped by trade and formatted into work orders using the firm's template. The PM reviews the output — typically a 10–15 minute review process — makes any necessary edits, and approves for sending. Contractors receive the work order via email with a one-click confirmation link. No app download required.

Meridian has now processed 8 consecutive projects through BuilderHQ. Here is what the numbers show.

Work order preparation time: Reduced from 2 working days (approximately 16 hours) to 20–25 minutes per project. The PM now spends 15 minutes reviewing and approving the AI output, plus 5–10 minutes making minor edits. Total PM time savings: approximately 15.5 hours per project, or 124 hours across the 8 projects.

Scope item capture rate: 97.3% average across 8 projects. The 2.7% miss rate is attributable to items specified verbally during design review meetings that were never incorporated into the formal plan set — a documentation gap that BuilderHQ cannot resolve but has made visible.

Missed scope items resulting in variations: Zero across 8 projects. The previous average was 2–3 per project. At an average variation cost of $4,500, this represents $144,000–$216,000 in avoided variation costs over the 8-project period.

Contractor confirmation rate: 94% of work orders confirmed within 48 hours, up from 71% previously. The automated reminder sequences (sent at 12 hours and 24 hours) account for most of the improvement.

PM capacity: The PM now manages 16 active projects, up from 12, without additional headcount. The 4 additional projects represent approximately $2.8M in additional annual revenue for Meridian based on their average project margin.

The headline number — 2 days to 20 minutes — is striking, but it understates the real impact. The more significant outcome is what the PM does with the recovered time.

Before BuilderHQ, the PM's week during a project launch phase was dominated by plan reading and work order preparation. Client calls were deferred. Site visits were shortened. Issues that needed attention were queued behind the work order backlog.

After BuilderHQ, the PM's week looks different. Work order preparation happens in a single morning session. The rest of the week is available for the work that actually requires human judgment — contractor relationships, client communication, site problem-solving, and project planning.

The capacity increase from 12 to 16 projects isn't just a revenue story. It's a quality story. The PM is less stretched, more responsive, and spending more time on the decisions that matter.

Three things Meridian would do differently if starting again:

1. Run the validation test on a harder project: The validation run used a straightforward townhouse project. A more complex project — with more trades, more specification notes, and more drawing revisions — would have given a better baseline for the AI's accuracy on edge cases.

2. Update the contractor register before going live: Several contractors in the imported register had outdated email addresses, which caused a handful of work orders to bounce on the first project. A 30-minute register audit before go-live would have avoided this.

3. Set up the client portal at the same time: Meridian added BuilderHQ's client portal 3 months after the initial implementation. In hindsight, setting it up during onboarding would have been straightforward and would have delivered client-facing benefits from day one.

'The plan reading feature alone paid for the platform in the first project. I used to dread the work order preparation phase — it was two days of staring at plans and trying not to miss anything. Now I upload the plans, review the output, and I'm done before lunch. The fact that it's also more accurate than I was is a bit confronting, but I'll take it.

The bigger change is harder to quantify. I'm not stressed at the start of a new project anymore. I used to feel behind before the project had even started. Now I feel ahead. That's worth more than the time saving.'

The Meridian case study is most directly relevant to residential developers and construction firms where:

1. Plan sets are large (50+ pages) and involve multiple trades 2. Work order preparation is a significant PM time commitment 3. Missed scope items are a recurring source of variations 4. Contractor confirmation tracking is done manually

If any of these describe your current process, the time and cost savings are likely to be similar to Meridian's. The platform is also used by commercial builders, subdivision developers, and mixed-use project managers — the AI plan reading module handles all standard Australian drawing formats and trade categories.

The free workflow audit is the right starting point. In 30 minutes, we'll map your current plan-to-work-order process, estimate your current time cost, and show you exactly what the output would look like for one of your actual projects.