How we removed 18 hours/week of pay-app reconciliation at a $32M Utah GC

A Utah general contractor running $32M in annual volume came to us with one specific complaint: pay-app week was eating their senior project accountant alive. We shipped the build in six weeks. The engagement paid for itself in four months and freed up 18 hours/week of senior accounting time every cycle thereafter. Here's exactly what it looked like.

The client, anonymized

Salt Lake City-headquartered general contractor. Roughly $32M in annual revenue, split across commercial, light-industrial, and a small public-works book. Twenty-eight active jobs at the time of engagement. Office staff of eleven, including one senior project accountant who owned the entire pay-app pipeline.

Tools in place: Procore as the source of truth for commitments, change orders, and subcontractor billing; Sage 300 CRE as the financial book of record; Microsoft 365 for everything else. No native Procore-to-Sage Connector — they had evaluated it twice and concluded the line-item mapping didn't match their cost-code conventions.

The before

Every month, the senior project accountant ran a process that took roughly 18 hours per week during pay-app week — and this was after she had already optimized it down from twenty-four hours over three years of iteration. The process:

• Export subcontractor billing schedules from Procore, by project, to Excel.
• Export committed cost reports from Sage 300 CRE, by project, to Excel.
• Open both side by side. Reconcile vendor name discrepancies (one system had “ABC Plumbing,” the other had “ABC Plumbing & Mechanical”). Reconcile cost-code discrepancies (a code renamed in Procore six months ago, never propagated to Sage).
• Hand-build each subcontractor's pay app PDF. Get partner signature. Email each pay app to the GC's controller for review.
• Re-key approved amounts back into Sage 300 CRE for AP processing.
• Catch the inevitable rejections — wrong amount, missing lien waiver, sub bumped a cost code — and start the relevant pieces over.

The senior accountant's honest description: “I open a spreadsheet on Friday and I close it on Tuesday. I am not doing anything else that week.”

The build

We did the standard Confluxion engagement — discovery, blueprint with fixed price and timeline, build with embedded check-ins, walkthrough, handover. The technical architecture, briefly:

• Ingestion.Scheduled exports from Procore (via their API) and Sage 300 CRE (via their ODBC interface) into a small staging database we provisioned inside the client's Azure tenant.
• Reconciliation.A fuzzy-match layer for vendor names (Levenshtein distance against a maintained alias table) and a cost-code crosswalk maintained by the project accountant in a SharePoint list. Anything that couldn't auto-match flagged for human review, with a one-click resolution UI.
• Pay app assembly. Templated PDF generation per subcontractor, pulling approved billings, lien waivers (matched by document metadata in their existing SharePoint), and the required AIA-style cover sheets. Output dropped into a per-month review folder, sorted by sub.
• Approval workflow. Partner review via SharePoint with comment-and-approve, then automated re-key into Sage 300 CRE for AP processing.
• Reject loop. Common rejection patterns coded as guard clauses up front (missing lien waiver, retainage math error, sub with prior unresolved variance). Rejections drop to a dashboard the accountant clears in minutes, not days.

Total build time: six weeks, on the timeline we quoted at blueprint. Weekly check-ins with the partner and the accountant kept everyone looped in through the build.

The result

• 18 hours/week of senior accounting time removed, every cycle. At roughly $95/hr loaded, that's about $7,400 per cycle, or ~$89,000 per year.
• Pay-app cycle compressed from 5 days to 1 day. Subcontractors started getting paid faster, which improved relationships measurably (their words, not ours).
• WIP report variance eliminated. The system produces the same numbers the WIP report uses, so the two don't disagree at quarter-end.
• Build paid for itself in four months — including both the fixed-price build and the small SharePoint reorganization we did in parallel.
• Engagement total: $36,500, fixed price, scoped at blueprint. The client paid the full amount on time. We have not touched the system in the six months since handover except for two minor cost-code crosswalk updates the accountant requested through the runbook.

What the accountant said

We asked the senior accountant for a quote we could use, and she gave us this one — verbatim:

We don't open a spreadsheet on Friday anymore.

What this engagement was, and what it wasn't

What it was: a focused, fixed-price automation build that retired one of the highest-leverage manual workflows in the firm. The client knew exactly what they were getting, what it would cost, and when it would ship — all before they wrote a check.

What it wasn't: a digital transformation initiative, a software product, or a multi-quarter consulting engagement. We did one thing. We did it well. We left.

Most of our automation engagements look like this. Pay apps, subcontractor compliance, Sage exports, AP triage, vendor compliance tracking. If you're running a similar process and a similar volume — twenty-plus active jobs, a senior accountant doing 12+ hours of manual reconciliation per cycle — there's a good chance the math works the same way for you.

And: an AI audit will surface this kind of workflow as a side effect. The fourth deliverable on every audit is a Workflow Friction Report — ranked list of automation candidates with estimated hours back and fixed-price quotes. We don't require the audit to lead to an automation build. We do require the automation build to start with a real understanding of where the time is actually going.

Client name withheld at the client's request. Dollar figures are exact. Hours are conservative (the accountant's own estimate, not ours).