Fragmented ESG evidence stalled reporting — Azure AI into Workiva for attested reports

Overview

Environmental, Social, and Governance (ESG) reporting at a logistics firm depended on emails, utility bill PDFs, and travel receipts stitched together near deadlines. Evidence was inconsistent, controls were informal, and auditors flagged gaps in provenance and disclosure alignment. Intelligex implemented document extraction with Azure AI Document Intelligence (formerly Form Recognizer), mapped outputs into Workiva for disclosure controls, and routed attestations to facility managers before executive sign-off. ESG packages assembled from the same governed dataset with attached evidence and approvals, which reduced errors, shortened review cycles, and made board and investor discussions more straightforward.

Client Profile

• Industry: Logistics and transportation
• Company size (range): Multi-region operator with warehouses, fleets, and corporate offices
• Stage: Established private company enhancing ESG maturity
• Department owner: Strategy, Analytics & Executive Leadership (Office of the CFO and Corporate Sustainability)
• Other stakeholders: Facilities and Fleet Operations, Procurement, Travel/Expense, Internal Audit, Legal & Compliance, IT/Data, Investor Relations

The Challenge

ESG disclosures combined utility invoices, fuel statements, fleet telematics exports, and employee travel data. Each business unit forwarded documents by email or shared drive links. Analysts keyed values into spreadsheets, reconciled units and billing periods by hand, and attached screenshots as support. When internal audit or external reviewers asked for source evidence or control records, the team retraced steps through inboxes and file shares.

Reporting frameworks evolved while controls lagged. Operations teams collected what was available rather than what standards required. Emissions factors, location-based versus market-based electricity methods, and activity boundaries were applied inconsistently. The firm already used Workiva for financial reporting and board packs, but ESG artifacts lived outside it. Leaders wanted a way to capture data from real documents, anchor calculations in a repeatable model, and enforce attestations and approvals in a system that could stand up to review.

Why It Was Happening

Evidence capture and controls were fragmented. Utility statements, fuel receipts, and travel exports arrived in varied formats with inconsistent metadata. Manual entry introduced errors, and version control was absent once spreadsheets circulated. Even when policies existed, the process did not enforce them at the point of intake.

Disclosure ownership lacked a workflow. Facilities managers and regional leaders understood consumption patterns, but they were not prompted to attest to completeness or period coverage before numbers moved to executive review. Workiva housed final narratives, yet it did not receive structured, sourced data with control logs attached. Without a governed pipeline and review gates, the team relied on heroics close to deadlines.

The Solution

We built a governed ESG data pipeline that extracted values from source documents, mapped them to a standard model aligned to the Greenhouse Gas (GHG) Protocol, and pushed controlled datasets into Workiva for disclosure and sign-off. Azure AI Document Intelligence handled extraction from invoices and receipts. Exceptions flowed to a human reviewer with the original document side by side. Facility and fleet managers attested to coverage and anomalies before Finance and Sustainability approved disclosure packages. Nothing was replatformed: repositories and email remained intake channels, Workiva remained the disclosure and control system, and the new layer orchestrated extraction, validation, and approvals.

• Document extraction using Azure AI Document Intelligence to parse utility bills, fuel statements, and travel receipts
• Standardized ESG data model aligned to the GHG Protocol Corporate Standard for scopes, activity types, units, and emission factors
• Mapping and unit normalization (kWh, therms, liters, gallons, miles) with location- and market-based methodologies captured as attributes
• Integration to Workiva ESG for disclosure controls, evidence attachment, and tasking (Workiva ESG)
• Attestation workflow for facility managers and fleet leads to confirm completeness and period coverage before roll-up
• Human-in-the-loop review queue for low-confidence extractions and edge cases, with decisions logged and reused
• Controlled calculations for emissions and intensity metrics with factor versioning and audit trails
• Role-based permissions and redaction for sensitive data (account numbers, personal identifiers)
• Operational dashboard showing document intake status, extraction quality, attestation progress, and outstanding approvals
• Publication gates: Sustainability and Finance sign-off before investor or board materials are released

Implementation

• Discovery: Cataloged source documents by region and site (utilities, fuel, travel/expense), identified required fields for each activity type, and mapped current disclosures to reporting frameworks and investor expectations. Assessed existing Workiva workflows and control requirements.
• Design: Defined the ESG data model and document templates, including unit normalization and activity boundaries. Selected extraction models and confidence thresholds. Designed the attestation and approval steps in Workiva, including evidence attachment requirements and escalation paths.
• Build: Configured Azure Document Intelligence to extract fields from frequent invoice formats; added rule-based post-processing for dates, units, and line-item aggregation. Built the transformation layer aligned to the GHG Protocol model. Integrated with Workiva to create controlled datasets, tasks, and certification workflows with evidence links.
• Testing and QA: Ran historical periods through the pipeline, compared outputs to prior filings, and reconciled differences. Tuned extraction confidence thresholds and exception categories. Verified that attestations captured completeness and that approvals reflected control ownership.
• Rollout: Began in observe-only mode, generating controlled datasets and evidence packages while the legacy spreadsheet process continued. After validation, enabled required attestations and approvals. Phased regions and facilities onto the workflow with a clear handoff schedule.
• Training and hand-off: Delivered short guides for facilities and fleet leads on document prep and attestations; trained Sustainability and Finance on Workiva tasks, evidence, and sign-offs; briefed Internal Audit on control design and sampling; and assigned stewardship for factor libraries and data model updates.

Results

ESG data flowed from source documents into a governed model with evidence attached at the record level. Facility managers confirmed coverage and anomalies before metrics rolled up, and Sustainability and Finance signed off inside Workiva with a visible control trail. Review sessions focused on insights and narrative rather than chasing files or debating units.

During board and investor reviews, the team referenced a controlled snapshot with a clear lineage from document to disclosure. Audit inquiries were addressed by opening the Workiva record with attached evidence, extraction logs, factor versions, and approvals. The scramble to assemble late-stage decks diminished, and disclosures became more consistent across periods and regions.

What Changed for the Team

• Before: Values were keyed from PDFs and emails into spreadsheets. After: Document extraction populated a standardized model with evidence attached.
• Before: Attestations were informal or absent. After: Facility and fleet leaders certified completeness before roll-up and executive sign-off.
• Before: Emissions methods differed by site and period. After: Methodology and factors were versioned and applied consistently with audit trails.
• Before: Workiva housed only final narratives. After: Controlled datasets, tasks, and approvals lived in Workiva alongside disclosures.
• Before: Audits required inbox searches and reconstructions. After: Reviewers saw the original document, extraction, and approval history in one place.

Key Takeaways

• Start with evidence: extract from source documents into a governed model rather than reconciling spreadsheets late.
• Align to a recognized standard such as the GHG Protocol so scopes, methods, and factors are transparent and repeatable.
• Route attestations to operational owners before executive sign-off; local confirmation prevents downstream rework.
• Keep existing repositories and Workiva; add extraction, normalization, and control orchestration around them.
• Version everything that affects results—factors, methods, templates—and capture approvals in the same system as disclosures.

FAQ

What tools did this integrate with?
We used Azure AI Document Intelligence to extract fields from utility bills, fuel statements, and travel receipts. Outputs mapped to a standardized model aligned with the GHG Protocol and flowed into Workiva ESG for disclosure controls, evidence, and sign-offs. Existing repositories and email remained intake channels, and the firm’s BI tool displayed operational dashboards.

How did you handle quality control and governance?
Quality gates included extraction confidence thresholds, rule-based validations for units and periods, and a human-in-the-loop queue for exceptions. Attestations by facility and fleet leads confirmed completeness, and Sustainability and Finance approvals locked controlled datasets in Workiva. Emissions factors and methodologies were versioned, and each disclosure carried evidence, factor versions, and approver identity in an audit trail.

How did you roll this out without disruption?
We ran the new pipeline alongside the existing spreadsheet process for a period, compared results, and tuned models. Once stakeholders were comfortable, we enabled required attestations and approvals by region. Workiva remained the disclosure platform, repositories stayed in place, and the orchestration layer standardized intake and controls without changing where teams store documents.

How were different scopes and methodologies handled?
The model captured scope and method attributes for each record. Electricity supported location- and market-based methods; fleet and travel captured activity data and applied the correct factors. Method choices and factor versions were stored with each record so reviewers could see exactly how a value was derived and adjust methods consistently across periods if policy changed.

What about sensitive data in invoices and receipts?
Extraction masked account numbers and personal identifiers, and role-based permissions limited who could view full documents. Evidence stored in Workiva used redacted copies where appropriate. Legal and Internal Audit reviewed privacy controls as part of the rollout, and access logs demonstrated who viewed what and when.