Real Estate & Workplace Solutions · Intelligent Automation
Invoice Extraction Now Takes 5 Minutes, Not 24, Saving $132K a Year
A global real estate and workplace solutions firm processed more than 1,000 non-standard invoices a month by hand. We built an intelligent automation pilot using optical character recognition (OCR) and robotic process automation (RPA) to pull data straight into Workday. Extraction fell from 24 minutes to 5, saving about $132K a year.
$132K/yrSaved at 1,000 invoices a month
40%Less processing time so far, 80% targeted24 to 5 minExtraction time per documentThe Challenge
The firm processed non-standard supplier invoices and utility bills by hand, extracting each one and reconciling it against meter readings. That took about 24 minutes per document and carried a high error rate. With nothing wired into the Enterprise Resource Planning (ERP) system, staff then keyed the same data in a second time. At more than 1,000 invoices a month, manual processing alone cost six figures a year, before a single late-payment penalty or duplicate payment. Every non-standard format added friction, and double-keying into Workday was time the team could not get back. At that volume the cost repeated every month.
What We Did
We started with the invoices themselves, building an OCR workflow that reads each document and pulls the data out. That cut extraction from about 24 minutes to 5 per document, removing the manual keying that came first in the process. We then added RPA bots to structure the extracted data and run it through payment workflows, reconciling energy and water bills against meter readings before anything is processed. The data flows straight into Workday, so nobody keys it a second time.
The Solution
OCR reads each invoice and pulls the data. RPA structures it, runs it through payment workflows, and reconciles energy and water bills against meter readings. Everything flows straight into Workday, so extraction takes 5 minutes per document instead of 24, and processing time overall is down 40%.
The Outcome
Extraction time per document fell from about 24 minutes to 5, and processing time is down 40% so far. At 1,000 invoices a month that is about $132,000 a year, and closer to $264,000 a year at twice the volume. The team no longer keys the same invoice twice, so those hours go back into closing the books instead of data entry. A machine-learning layer now in development targets an 80% cut in processing time overall, a roadmap target rather than a delivered result.
Related cases
Teams Ship Releases 70% Faster Without Loosening Change Control
A Singapore-based insurer with 5,000-plus employees coordinated every release by hand across separate change, deployment and security tools. We unified those workflows inside ServiceNow and integrated Octopus and Tufin, so an approved change triggers deployment automatically. Deployment cycles are down 70%, with full visibility and no loss of control.
Field Reps Act in the Moment, Saving $100K a Year in Overstock
A retail execution software platform connects national big-box retailers with a freelance workforce. It ran on siloed, batch data across route, territory and schedule optimization, and across Photo-AI, its AI image validation. We replaced that with a real-time Data Vault 2.0 lakehouse. Field reps get timely in-store insight, and the platform banks $100K a year in overstock savings.
The Team Cleared 40% of Its Backlog Without a Single New Hire
A Kuwaiti bank operating under a zero-new-hire mandate had to improve service anyway. Noblq took full ownership of its ServiceNow platform as a dedicated managed-services team, with Generative AI (GenAI) automating identity checks and approvals running natively in Microsoft Teams. The incident backlog fell 40%, with no new headcount.
Still closing the books on manual invoice work?
Noblq works through your process with you and applies the right technology to the way you actually work, to automate what used to need people. You get a system shaped to your business, sized to the return, and owned by you.