Extract Cash Flow from PDF — Automated PDF to Excel Cash Flow Extraction | Sparkco

Product overview and core value proposition

Sparkco automates extraction of cash-flow data from any PDF into Excel-ready spreadsheets with formulas, consistent formatting, and full audit trails. For teams that need PDF to Excel precision and speed, Sparkco applies AI document parsing and data extraction to deliver structured, analysis-ready outputs in minutes, not hours. Built for FP&A, accounting, treasury, and transaction advisory teams that demand accuracy, repeatability, and traceability.

What it does: Sparkco reads cash flow statements and related schedules in PDFs and converts them to clean Excel models—with mapped line items, correct sign conventions, roll-forward logic, and linked subtotals. It identifies operating, investing, and financing sections, extracts amounts and notes, and applies spreadsheet formulas so FP&A can plug outputs directly into models and dashboards.

Why it matters: Manual PDF data entry typically takes 30–60 minutes per statement and carries 1–5% error rates according to industry benchmarks. Sparkco reduces processing time to about 1–2 minutes per document and targets 99.5–99.9% accuracy, cutting rework and review. That’s an 80–90% time reduction and a 70–95% drop in errors, driven by OCR, layout understanding, and validation rules that reconcile totals and cash roll-forwards.

Business outcomes: Finance teams close faster, forecast more often, and audit more easily. Typical quick wins include reducing data-entry time by 80–90%, cutting monthly reconciliation by 10–20 hours, and accelerating the close by 1–3 days. FP&A gains more time for scenario analysis; accounting improves consistency and ties-out faster; treasury sees earlier cash visibility; transaction advisory can process diligence packets at 3–5x throughput with a reproducible audit trail.

• Time: 80–90% faster cash-flow extraction versus manual PDF to Excel entry (30–60 minutes down to 1–2 minutes per statement).
• Accuracy: reduce manual entry error rates (1–5% industry average) to a 99.5–99.9% accuracy target with validation and reconciliations.
• Auditability: automatic logs, versioned templates, and traceable fields improve audit readiness and policy compliance.
• Scalability: 3–5x more documents processed per FTE per month via batch processing and review queues.

Top measurable benefits and KPIs

Key features and capabilities

Technical overview of document parsing and PDF automation features mapped to finance outcomes. Each capability includes definition, operation, accuracy, edge-case handling, and a micro-scenario with measurable impact.

This section details core document parsing features for financial PDFs, with realistic accuracy ranges, limitations, and direct benefits to finance teams adopting PDF automation and table extraction.

Feature-to-benefit mapping

Intelligent PDF parsing (OCR + layout analysis)

OCR combines text recognition with layout segmentation to locate paragraphs, tables, headers, and footers. In practice it auto-detects language, de-skews pages, and normalizes contrast before extraction.

• Benefits: cuts manual entry and improves consistency across document parsing pipelines.
• Accuracy: 90–99% field-level on digital PDFs; 85–95% on 300 dpi scans.
• Edge cases: rotation, stamps, watermarks; mitigations include de-skew, binarization, and page masking.
• Micro-case: Bank statements (100 pages); daily balances extracted and opening/closing reconciled; 6 hours saved; errors reduced ~70%.

Table detection and tabular extraction

Algorithms mix deep detectors with line/whitespace heuristics to find tables, headers, and spanning cells. Cell structure is reconstructed with ruled-line tracing and text block grouping.

• Benefits: high-fidelity table extraction enables automated reconciliations and variance analysis.
• Accuracy: table detection 92–98%; cell extraction 88–96% on clean, bordered layouts.
• Edge cases: borderless or nested tables; fallbacks include column projection and header semantic inference.
• Micro-case: AP aging report; 10k rows extracted; reconciliation time cut from 3h to 45m.

Multi-page document stitching

Stitches tables across page breaks using carry-forward amounts, repeating headers, and sequence cues. Preserves row order, merges continued sections, and validates roll-forwards.

• Benefits: eliminates manual copy/paste across pages.
• Accuracy: 90–97% when headers or carry-forward markers persist.
• Edge cases: out-of-order scans; mitigated by page number detection and balance checks.
• Micro-case: 50-page GL export; continuous ledger built; rework reduced 40%.

Automated field mapping to Excel templates

Maps extracted fields to named ranges and headers in Excel via schema matching and fuzzy header similarity. Confidence thresholds route low-confidence mappings for review.

• Benefits: accelerates reporting and close tasks.
• Accuracy: >95% on known templates; 85–93% on unseen but similar formats.
• Edge cases: renamed headers; mitigated with synonyms and unit normalization.
• Micro-case: Cash-flow template; auto-populated sections; 2 hours saved per entity.

Formula and formatting propagation

Inserts data while preserving workbook formulas, named ranges, and styles. Recalculates and checks totals/subtotals for drift against extracted figures.

• Benefits: prevents broken formulas and keeps reporting visual standards.
• Accuracy: reference integrity checks detect >99% broken links in test suites.
• Edge cases: volatile macros and external links; mitigate with protected ranges and pre-checks.
• Micro-case: KPI workbook; 12 sheets updated; zero formula breaks; 30 minutes saved per refresh.

Entity and line-item recognition (cash inflows/outflows, operating/investing/financing)

Classifies transactions and lines using NER and rules to tag inflow/outflow and cash flow categories. Normalizes vendors, accounts, and memo text for consistent downstream use.

• Benefits: speeds cash flow assembly and segment reporting.
• Accuracy: 90–97% for common descriptors; lower on sparse memos or mixed languages.
• Edge cases: ambiguous merchants; mitigations include bank code dictionaries and reviewer prompts.
• Micro-case: 12k transactions auto-classified; variance ties in 15 minutes; manual effort down 70%.

Exception handling and human-in-the-loop review

Confidence scoring routes low-certainty fields to review; corrections feed active learning. Reviewer actions are logged and re-applied to similar future documents.

• Benefits: raises quality while containing risk.
• Accuracy: post-feedback lift of 2–5% on recurring documents.
• Edge cases: large exception spikes from new layouts; mitigated by template onboarding.
• Micro-case: Invoice totals mismatch flagged; reviewer fixes tax field; prevents export defect.

Batch processing and scheduled workflows

Parallel workers process files from S3/SFTP with retry and backpressure. Schedules trigger nightly or hourly runs and emit status webhooks.

• Benefits: predictable throughput and SLAs.
• Accuracy: deterministic runs with checksum verification on outputs.
• Edge cases: burst traffic; mitigated by autoscaling and queue timeouts.
• Micro-case: 500 statements nightly; 95th-percentile completion <2 hours.

Audit trail and change history

Immutable logs capture source file hash, model versions, reviewer actions, and diffs across exports. Supports trace-back from any cell to origin page and coordinates.

• Benefits: audit readiness for SOX and internal controls.
• Accuracy: complete lineage via cryptographic hashes.
• Edge cases: PII retention; mitigated with redaction and role-based access.
• Micro-case: External audit requests lineage; evidence assembled in minutes.

Export options (Excel, CSV, API)

Exports validated data to Excel workbooks, CSV extracts, or a REST API with schema validation. Supports idempotent retries and duplicate detection.

• Benefits: easy integration with ERP and BI pipelines.
• Accuracy: schema checks catch missing or mis-typed fields before export.
• Edge cases: API rate limits; mitigated with pagination and backoff.
• Micro-case: Push to NetSuite and S3; zero duplicates; 20-minute ETL removed.

Buyer evaluation checklist

Use this concise checklist when comparing document parsing and PDF automation solutions.

• Report field-level and cell-level accuracy on your document samples (digital vs scanned).
• Demonstrate table extraction on borderless and multi-header tables with merged cells.
• Show multi-page stitching with carry-forward reconciliation and subtotal validation.
• Validate mapping into your actual Excel templates with formulas preserved.
• Require confidence scores, exception routing, and reviewer feedback loops.
• Assess throughput under batch loads and view SLA metrics and retries.
• Inspect audit logs for source hashes, versioning, and cell-to-origin traceability.
• Test exports (Excel/CSV/API) with schema validation and idempotency.

How it works: Upload, parse, and export to Excel

A precise, stage-based ETL for PDF to Excel document conversion that ingests, OCRs, parses, maps, validates with confidence scoring, enriches data, and exports Excel with formulas preserved.

This guide explains the full PDF to Excel pipeline used by finance and operations teams to extract cash flow from PDF, invoices, and statements at scale. It prioritizes latency, accuracy, and auditability, with human review for ambiguous fields and export options that preserve spreadsheet formulas.

Below you will find steps, latency expectations, confidence thresholds, error handling, and UX copy so you can deploy with predictable SLAs.

1) Step-by-step workflow: PDF to Excel

The ETL is organized into eight stages with measurable inputs, outputs, latency, and failure modes.

1. Ingest: Single file, bulk upload, or API ingest queues PDFs and images for processing.
2. Pre-processing: OCR-oriented cleanup (de-skew, denoise, auto-rotate, language detection).
3. Parsing: Layout analysis, table and line-item extraction, NLP-based label detection.
4. Mapping: Auto-map entities to Excel templates; allow user overrides and saved rules.
5. Validation: Confidence scoring at field and document level; route low-confidence to human review.
6. Enrichment: Vendor lookup, currency conversion, date normalization, and business rule checks.
7. Export: Generate Excel (XLSX) with formulas preserved, CSV, JSON, and API webhook delivery.
8. Monitoring: Track latency, confidence, throughput, and exception rates with audit logs.

Stage specifications

Confidence thresholds and actions

Confidence scoring and human review

Each token, field, and document receives a confidence score (0–1). Rules combine OCR, parsing features, and business checks (e.g., totals sum, tax rate bounds). Items under thresholds are routed to a human review queue with side-by-side document view, bounding-box highlights, and field-level history.

Audit logs capture original value, confidence, editor, change reason, and timestamp for compliance.

• Human review UX copy: Review Exceptions, Accept All High-Confidence, Approve and Export, Send to Reprocessing, Assign Reviewer
• Reviewer tools: Rotate Page, Split Document, Merge Pages, Override Mapping, Save Template Rule
• Escalation: if reviewer cannot resolve, mark Needs Rescan and notify uploader

Latency expectations (per page and per batch)

Per-page (p50): pre-processing 0.3–0.8 s + OCR 1–3 s + parsing 0.5–1.5 s = 1.8–5.3 s. GPU OCR or cloud OCR reduces to ~1.3–3.5 s.

A 5-page invoice pack typically reaches Excel in 10–20 s (p50) and 30–45 s (p95), excluding human review.

Batching: 100 documents x 5 pages with 10 parallel workers completes in ~2–5 minutes wall-clock, assuming steady-state compute and I/O.

Export to Excel and other formats

Excel export writes values to designated input cells or named ranges; template formulas, pivot tables, and references remain intact and recompute on open. CSV and JSON are available for downstream ETL, and webhook payloads enable real-time integrations.

• UX copy: Download Excel, Export CSV, Send Webhook, Choose Template, Recalculate Before Save
• Excel specifics: preserve formulas; lock non-input cells; support multiple tabs; support dynamic tables
• Webhook: POST JSON to configured endpoint with signature, file id, and link to XLSX

UX copy and screenshots

Use clear, action-oriented labels and provide visual confirmation for each stage. Suggested screenshot placeholders below.

• Buttons: Upload PDFs, Start Batch Extraction, Review Exceptions, Open Mapping Editor, Approve and Export, Retry Failed
• Empty states: Drop files here to convert PDF to Excel, or Paste URL, No exceptions. Great job!
• Tooltips: Confidence 0.91 (below 0.95 threshold). Click to review., This field was auto-mapped from saved rule.

Common problems and mitigations

• Poor scan quality: enforce 300 DPI; enable binarization and de-noise; ask for original PDF if confidence <85%.
• Rotated or skewed pages: auto-rotate and de-skew; expose Rotate Page in review.
• Complex tables (merged cells): use structure-aware table extraction; fallback to line-item heuristics with reviewer confirmation.
• Password-protected PDFs: prompt for password at upload; otherwise reject with reason.
• Mixed languages: run language detection per page; route to language-specific OCR models.

ETL diagram text

Source Connector -> Pre-Processor -> OCR Service -> Layout Parser -> Table/Line-Item Extractor -> NLP Labeler -> Mapper (Excel templates, named ranges) -> Validator (confidence thresholds, business rules) -> Enricher (FX rates, vendor master, date normalization) -> Exporter (XLSX/CSV/JSON/Webhook). Data transforms: binary file -> rasterized pages -> tokens with bbox -> structured fields/tables -> mapped Excel cell bindings -> validated record -> enriched dataset -> exported artifacts.

Key questions answered

• How long from upload to Excel? Typical 5-page PDF: 10–20 s p50, 30–45 s p95 without human review.
• How are ambiguous fields resolved? Confidence thresholds route to Review Exceptions; reviewer selects the correct candidate and can Save Template Rule.
• Does the export preserve formulas? Yes. Only input cells are written; formulas recalculate on open or server-side if Recalculate Before Save is enabled.
• Can I extract cash flow from PDF statements? Yes. Map source line items to cash flow template rows; totals must reconcile before export.
• What happens on failure? The job is retried with exponential backoff; persistent errors are surfaced in the queue with Retry Failed and detailed logs.

Supported document types and real-world examples

A concise, practical guide to CIM parsing, bank statement conversion, and PDF to Excel workflows with field mappings, formulas, and parsing tips.

This compendium outlines supported document types for cash flow extraction and shows how each is transformed from PDF to Excel with concrete mappings, example snippets, and robust parsing tips. Variations exist by jurisdiction and issuer; verify line-item nomenclature and units before loading to models.

• Confidential Information Memoranda (CIMs)
• Audited financial statements
• Bank statements
• Invoices
• Payment advices / remittances
• Payroll reports
• Clinical/medical records with billing and cash flows

CIM (Confidential Information Memorandum)

Typical layout: narrative plus multi-year financial tables (income, balance sheet, cash flow), pro forma adjustments, EBITDA add-backs, and footnotes spanning pages. Extraction targets: pro forma cash flow by activity, EBITDA adjustments, working-capital bridges, and forecast periods.

• Footnotes: map note numbers to EBITDA add-backs using XLOOKUP on a notes table.
• Multi-page tables: unify headers across page breaks; validate row continuity by line-item labels.
• Currency: normalize to a base currency via an FX rate table with effective dates.
• Sign rules: use parentheses as negatives; treat dashes as 0.

Sample Excel mapping (1-sheet summary from a 3-page CIM)

Audited financial statements

Typical layout: auditor’s report, primary statements (income, balance sheet, cash flows), and notes. Extraction targets: cash flows by activity, interest and taxes paid, non-cash adjustments, and reconciliation items (IFRS vs US GAAP labels).

• Parentheses mean negatives; convert to signed numbers before loading.
• Presentation currency may differ from note disclosures; track FX.
• Discontinued ops often separated; exclude from operating cash if required.
• Totals may repeat across pages; deduplicate by hash of line label+period.

Sample mapping to Excel (cash flow detail)

Bank statements

Typical layout: account header, statement period, running ledger of debits/credits, daily balances, and sometimes check images. Extraction targets: value date, description, debit/credit, balance, check numbers, fees.

• Rolling balances can reset at page breaks; recompute in Excel.
• Distinguish posting date vs value date for cash timing.
• Normalize descriptions (strip reference IDs, merge wrapped lines).
• Handle OCR artifacts: 1 vs l, dot commas, and whitespace.

• Excel rolling balance (Balance F2): =IF(ROW()=2,OpeningBalance,F1+E2-D2)
• Month-to-date cash inflows: =SUMIFS(E:E,A:A,">="&EOMONTH(TODAY(),-1)+1,A:A,"<="&EOMONTH(TODAY(),0))
• Fee classification: use a keyword table and XLOOKUP on Description

Sample mapping to Excel (transaction-level)

Invoices

Typical layout: seller/buyer headers, invoice metadata, line-item table (description, qty, unit price), taxes/discounts, totals, and payment terms. Extraction targets: header fields, line-item amounts, taxes, currency, and due dates.

• Multi-line descriptions wrap; join lines until next row with Qty/Price.
• VAT/GST vs sales tax: capture rate and jurisdiction.
• Stamped/handwritten notes can overwrite totals; prefer subtotal + tax + discount recomputation.
• Currency codes may differ from symbols; map via ISO 4217 table.

• Line total (G2): =E2*F2
• Due date (K2): =B2+J2
• Currency normalization (L2): =G2*XLOOKUP(I2,FX[Currency],FX[Rate])
• MTD normalized sales: =SUMIFS(L:L,B:B,">="&EOMONTH(TODAY(),-1)+1,B:B,"<="&EOMONTH(TODAY(),0))

Sample mapping to Excel (header + line items)

Payment advices / remittances

Typical layout: payer, deposit date, remittance lines referencing multiple invoices, deductions/fees (e.g., bank charges), and net amounts. Extraction targets: per-invoice allocation, fees, short pays, and method (ACH, wire, lockbox).

• EFT addenda/835 remittance codes map to adjustments; keep a code dictionary.
• One payment can reference many invoices; maintain one row per allocation.
• Bank fees deducted at source: capture in Fees/Deductions for true cash received.
• Remittances span pages; use statement ID + sequence for grouping.

Sample mapping to Excel (allocation-level)

Payroll reports

Typical layout: pay period header, employees with gross pay, taxes, deductions, net pay, and employer costs. Extraction targets: cash disbursement dates, net pay totals, tax remittances, and off-cycle runs.

• Separate cash vs non-cash perks; exclude non-cash from cash flow.
• Off-cycle payments: flag for forecasting; dates may differ from period end.
• YTD columns reset annually; compute MTD from Payment Date.
• Employer taxes paid separately; map to financing or operating cash per policy.

• MTD payroll cash: =SUMIFS(G:G,B:B,">="&EOMONTH(TODAY(),-1)+1,B:B,"<="&EOMONTH(TODAY(),0))
• Accrual estimate: =GrossDailyRate*MAX(0,NETWORKDAYS(PeriodEnd+1,MonthEnd))

Sample mapping to Excel (cash-focused)

Clinical/medical records with billing and cash flows

Typical layout: encounters with CPT/HCPCS, charges, payer adjudication (ERA/EOB), adjustments/write-offs, patient responsibility, and payment postings. Extraction targets: charge vs allowed, paid amounts by payer, adjustments, patient payments, and payment dates.

• Map CARC/RARC codes to adjustment categories (contractual, denial, copay).
• UB-04/CMS-1500 red forms require tuned OCR; capture field anchors.
• Split claims into service lines; one check may pay multiple claims.
• Protect PHI: mask identifiers when exporting to shared Excel.

Sample mapping to Excel (service-line)

Use cases and target users

Who should use this platform, why it matters, and how it fits common finance and deal workflows. Focus: document automation, data extraction, and PDF to Excel pipelines with measurable time and accuracy gains.

Teams that repeatedly turn unstructured PDFs, scans, and spreadsheets into analysis-ready data benefit most. Typical expectations: 200–20,000 documents per week, 20–120 fields per document, and 97–99.5% target accuracy with human-in-the-loop. Outcomes: faster closes, better liquidity visibility, and shorter diligence cycles.

The platform’s strengths are high-accuracy data extraction, PDF to Excel export, schema validation, template learning, and APIs for end-to-end document automation.

Operational benchmarks

FP&A teams

Persona: Senior Financial Analyst/FP&A Manager running monthly/quarterly close, variance analysis, and forecast updates. Pain: manual PDF to Excel, scattered files, and late actuals.

FP&A use cases

Corporate accountants

Persona: GL/Staff Accountants handling reconciliations, journal support, AP/AR backups. Pain: manual keying from invoices and statements, audit-ready trails.

Corporate accounting use cases

Treasury

Persona: Treasury Analyst/Manager responsible for cash positioning, liquidity forecasting, and bank connectivity. Pain: manual bank statement ingestion and delayed visibility.

Treasury use cases

Transaction services / IBD teams

Persona: M&A advisors and TS professionals creating models and memos. Pain: slow CIM digestion and manual normalization of historicals.

M&A advisory use cases

Private equity deal teams

Persona: Investment Associates/Operating Partners tracking portfolio KPIs and lender reporting. Pain: inconsistent monthly packs across companies.

PE use cases

Audit firms

Persona: External auditors executing PBC intake and substantive testing. Pain: manual sampling support and inconsistent evidence formatting.

Audit use cases

Document automation specialists

Persona: RevOps/IT automation engineers owning pipelines. Pain: brittle templates, limited governance, and slow change control.

Automation engineering use cases

Technical specifications and architecture

Technical architecture and specifications for a document parsing API designed for secure, scalable PDF automation architecture. Covers layers, data flow, security controls, performance benchmarks, deployment footprints, and API/webhook patterns.

Architecture diagram narrative: The PDF automation architecture is layered into Ingest, Processing/ML, Storage, API/UX, Audit/Logging, and Security/IAM. Documents arrive via HTTPS upload, pre-signed URLs (S3/GCS/Azure), or watched folders for on-prem. The Ingest layer normalizes formats, validates signatures, antivirus-scans, and enqueues work (Kafka/SQS/RabbitMQ). Processing workers handle preprocessing (deskew, denoise, binarize), page segmentation, OCR via pluggable engines (on-prem Tesseract; cloud connectors to AWS Textract or Google Vision), layout analysis, ML-based field extraction, and schema validation. Outputs include structured JSON, tokens with bounding boxes, and confidence scores, persisted alongside source files and lineage metadata. API/UX exposes synchronous endpoints for small documents and asynchronous jobs with webhooks for batches, while Audit/Logging records every event and access. Security/IAM enforces encryption, key management, RBAC, SSO, and data residency.

This design emphasizes a high-availability document parsing API and a resilient PDF automation architecture that scales horizontally, isolates tenants, and supports both cloud-native and on-premise deployments.

Architecture layers and data flow

Technical specifications

OCR engines: Pluggable. On-prem default Tesseract 5.x (LSTM) with language packs; cloud connectors to AWS Textract (AnalyzeDocument/Expense) and Google Vision OCR. Accuracy varies by domain: clean typed text typically 94–99% with cloud OCR, structured table/line-item extraction generally stronger in Textract; Tesseract is cost-effective on-prem but sensitive to noise.

Supported formats: PDF (native and scanned), TIFF (single/multi-page), PNG, JPEG; optional Office ingestion (DOCX/XLSX) via server-side conversion. Max default file size 50 MB (configurable), up to 500 pages/document.

API and rate limits: REST/JSON over HTTPS. Default 600 requests/min per API key, burst 1200 for 60 seconds; 50 concurrent jobs/key (raise via contract). Synchronous limits: up to 10 pages or 10 MB; otherwise async job is required.

Latency and throughput (typical): Single-page PDF p50 900 ms, p95 2.5 s (cloud OCR); Tesseract on 8 vCPU worker p50 1.4 s, p95 3.0 s. Throughput: Tesseract 20–40 pages/min per 8 vCPU worker; cloud OCR with 50 parallel jobs achieves 1,000+ pages/min aggregated. End-to-end batch latency dominated by OCR and network I/O.

Storage and retention: Results and artifacts retained 30 days by default (configurable 0–365); ephemeral mode keeps only streaming memory and deletes source on completion. Data residency zones per tenant.

Observability: Structured logs (JSON), metrics (p50/p95 latency, queue depth), distributed tracing, and full audit trails (who/when/what) exportable to SIEM.

• Deployment footprints: Cloud-managed (autoscaling containers, S3/GCS, managed KMS) or on-prem (Kubernetes or VMs).
• On-prem minimum: 3 nodes — API (4 vCPU/8 GB), worker (8 vCPU/32 GB), storage (MinIO 4 vCPU/16 GB + Postgres 2 vCPU/8 GB) + queue (2 vCPU/4 GB).
• On-prem HA: 6–10 nodes with 3+ worker nodes, replicated DB, and erasure-coded object storage.
• Horizontal scaling: Add workers; partitions via queues; idempotent job processing.

Security and compliance

Encryption: TLS 1.2+ in transit, optional mTLS for service-to-service. AES-256-GCM at rest in object storage and databases; per-tenant keys via KMS; optional HSM-backed keys and FIPS 140-2 validated modules.

Identity and access: RBAC with fine-grained scopes (ingest, read-results, admin). SSO via SAML 2.0 and OAuth 2.0/OIDC; SCIM for user provisioning. IP allowlists and signed URLs for artifact access.

Logging and audit: Immutable audit trails with hash-chaining; 1-year default retention (configurable). PII redaction for logs.

Compliance: Managed cloud offering maintains SOC 2 Type II and ISO 27001 certifications; on-prem/self-hosted deployments can inherit customer controls and are provided with security implementation guides.

• Data handling: Optional customer-managed keys (CMK), field-level redaction, and zero-retention mode.
• Data residency: Region pinning; no cross-region replication unless enabled.
• Privacy: GDPR-ready data subject controls; access logs available to tenants.

Scalability and deployment options

Cloud scaling: Stateless API tier behind a gateway; worker autoscaling based on queue depth and p95 latency SLOs. Multiple OCR backends can be load-balanced per profile.

On-prem scaling: Add workers per queue partition. GPU acceleration optional for layout/vision models; Tesseract remains CPU-bound. Benchmark representative samples before capacity planning.

• Concurrency limits: Soft limit 2,500 concurrent jobs/tenant; hard safety valve at 10,000 per region (raise via support).
• Batch processing: Multipart manifests and ordered webhooks; resume via idempotency keys.
• Disaster recovery: Daily snapshots of metadata DB; versioned object storage with lifecycle policies.

API design and examples

Extraction request (async recommended for large inputs):

{ "input_url": "https://s3.amazonaws.com/bucket/invoice.pdf", "file_format": "pdf", "pages": "1-5,10", "profile": "invoice_v1", "async": true, "webhook_url": "https://example.com/hooks/doc-complete", "metadata": { "tenant_id": "acme-co", "doc_ref": "INV-10023" } }

Response (202 Accepted):

{ "job_id": "job_01HF7A2ZQW", "status": "queued", "estimated_cost_cents": 12 }

Synchronous extraction (small files):

{ "file_b64": "...", "profile": "generic_document", "async": false }

Sync response (truncated):

{ "status": "succeeded", "pages": 3, "entities": [{"type":"invoice_number","value":"10023","confidence":0.98}], "tokens": [{"text":"ACME","bbox":[0.12,0.08,0.23,0.11]}] }

Webhook delivery on completion:

{ "event": "extraction.completed", "job_id": "job_01HF7A2ZQW", "status": "succeeded", "p95_latency_ms": 2300, "pages": 12, "results_url": "https://api.example.com/v1/jobs/job_01HF7A2ZQW/results", "metadata": { "tenant_id": "acme-co", "doc_ref": "INV-10023" } }

Webhook behavior: Signed with X-Signature: HMAC-SHA256(secret, body). Retries with exponential backoff up to 10 attempts; duplicate-safe via X-Idempotency-Key. 2xx acknowledges; 4xx are not retried; 5xx retried.

• Endpoints: POST /v1/extractions, GET /v1/jobs/{id}, GET /v1/jobs/{id}/results, POST /v1/webhooks/test
• HTTP semantics: Idempotent submissions via Idempotency-Key header; pagination on listings using cursor tokens.
• Errors: Structured problem+json with trace_id; common codes 400, 401, 403, 413, 429, 500.

Questions answered

• Can this run on-premise? Yes. Kubernetes or VM-based deployments are supported with Tesseract by default; adapters can call Textract/Vision if egress is permitted.
• What encryption standards are used? TLS 1.2+ in transit; AES-256 at rest with KMS-managed keys; optional HSM and FIPS 140-2 validated crypto.
• What are the API rate limits? Default 600 requests/min per key, burst 1200 for 60 s; 50 concurrent jobs/key; adjustable by plan.
• How fast does it process? Typical single-page p95 2–3 s; throughput ranges 20–40 pages/min per 8 vCPU worker on Tesseract; cloud OCR scales to 1,000+ pages/min with parallelization.
• What file types are supported? PDF, TIFF, PNG, JPEG; optional DOCX/XLSX via conversion.
• What compliance is available? Managed cloud: SOC 2 Type II and ISO 27001; on-prem inherits customer’s certifications and controls.

Integration ecosystem and APIs

Connect your systems to automate PDF to Excel at scale. Prebuilt connectors, robust APIs, reliable webhooks, and SDKs deliver secure, end-to-end integration.

Our integration ecosystem streamlines PDF to Excel by meeting your data where it lives. Use out-of-the-box storage and ERP connectors, or build custom flows with our public APIs, webhooks, and SDKs to automate ingestion, mapping, extraction, and delivery.

• SEO focus: APIs, integration, PDF to Excel

Core API endpoints

Feature list and connectors

Deploy quickly with prebuilt connectors, then extend via APIs for custom logic.

• Content storage: SharePoint, OneDrive, Google Drive, Dropbox, Box
• Email ingest: Microsoft 365/Exchange (Graph API), Gmail (Gmail API)
• ERPs: NetSuite, SAP S/4HANA and ECC (via iDoc/BAPI/OData gateways)
• Accounting: QuickBooks Online, Xero, Sage Intacct
• RPA and workflow: UiPath, Automation Anywhere, Power Automate, Zapier, Make
• Data transport: SFTP, HTTPS pre-signed URLs
• Custom: Public APIs, webhooks, and SDKs for tailored integrations

Public APIs

Authenticate with an API key (recommended for server-to-server) or OAuth 2.0 Client Credentials. Scope keys to least privilege and rotate regularly. Use Idempotency-Key for uploads and reprocess calls.

• SDKs: Python, JavaScript/TypeScript, Java, .NET, Go
• Content types: multipart/form-data for uploads; JSON for control endpoints
• Idempotency: header Idempotency-Key recommended for POSTs

Sample payloads

Webhooks and reliability

Register webhooks per environment. We sign payloads with HMAC-SHA256; verify header X-Signature against your secret. Respond with 2xx within 10s.

• Events: job.completed, job.failed
• Retries: exponential backoff (approx. 30s, 2m, 10m) up to 6 attempts, then DLQ
• Security: HTTPS only, IP allowlist optional, verify signature, store secrets in a vault

Developer quickstart (3 steps)

1. Upload a PDF: POST /v1/files/upload with file and optional mapping_id. Expect 10–30s for typical files; receive job_id.
2. Check status: GET /v1/jobs/{job_id}/status every 2–3s (or use webhooks). Stop polling when status=completed or failed.
3. Download Excel: GET /v1/jobs/{job_id}/result?format=xlsx. Save the file to storage or forward to your ERP.

Integration playbooks

• Scheduled shared-drive ingestion: Connect SharePoint/OneDrive/Google Drive/Dropbox/Box. Run every 5–15 minutes, upload new PDFs, move processed files to /processed, and tag with job_id. Use checksums to avoid duplicates.
• Trigger from email attachments: Use Graph/Gmail filters to forward attachments to a staging folder or pre-signed upload. Parse subject/sender to choose mapping_id. Post results back to the original thread or ERP via API.
• API-first batch processing: Stream files from an object store, throttle to 20–50 RPS, set Idempotency-Key per file. Poll with backoff, consume webhooks, and bulk-download results to S3/Blob. Reprocess failures with updated mappings.

Troubleshooting tips

• 401/403: Invalid key or scope. Verify Authorization header and OAuth scopes.
• 415: Unsupported media type. Use multipart/form-data for uploads; PDF only for source.
• 429: Back off per Retry-After; reduce concurrency. Enable client-side jitter.
• 5xx/timeout: Retry with exponential backoff and idempotency. Check /v1/health.
• Webhook missing: Confirm HTTPS, firewall allowlist, and certificate chain. Verify X-Signature. Check your 2xx acknowledgments within 10s.

Pricing structure and plans

Simple, transparent pricing for PDF to Excel and document extraction with clear tiers, predictable overages, and ROI you can measure.

Choose a plan that fits your volume today and scales tomorrow. Our pricing is anchored to market-leading OCR/document AI benchmarks so you get premium extraction without surprises. Competitor pricing ranges from $0.0015–$0.07 per page depending on features and volume; our tiers stay competitive while adding purpose-built PDF to Excel workflows, integrations, and support.

Every plan supports per-page, per-document, per-field, or subscription billing so you can align spend with value. Overages are automatic at discounted rates, and you can switch plans anytime. A 14‑day free trial with 500 pages lets you validate accuracy, speed, and cost before you commit.

Plan comparison and ROI scenarios

What’s included by tier

Pay-as-you-go is ideal for teams exploring PDF to Excel or seasonal spikes. Standard adds predictable pricing, higher API limits, and popular integrations. Enterprise unlocks SSO, advanced security, priority SLAs, on‑prem deployment, and white‑glove services.

• Integrations: Zapier, Make, Google Drive, SharePoint, S3, Slack, Webhooks, and REST API.
• Security: SOC 2 program, encryption at rest/in transit; Enterprise adds SSO/SAML, SCIM, audit logs, VPC options.
• Services: custom templates, dedicated onboarding, and mapping experts to accelerate time-to-value.

Overages, billing models, and terms

Billing models: choose per-page, per-document, per-extraction-field, or subscription. Per-document pricing starts at $0.10 per document (up to 5 pages) plus $0.02 per additional page; per-field starts at $0.002 per extracted field.

Overages: billed automatically at the rates shown above; usage resets monthly; no rollovers. Contracts: Pay-as-you-go and Standard are monthly, cancel anytime. Enterprise is annual (or multi‑year) with quarterly true‑up. Typical enterprise ranges: $2,500–$15,000/month plus discounted usage; on‑prem licensing from $60,000/year.

How to choose

• Under 1,000 pages/month or irregular use: Pay-as-you-go for lowest commitment.
• 1,000–15,000 pages/month with integrations: Standard for predictable pricing and scale.
• 15,000+ pages/month, SSO, or on‑prem: Enterprise for security, SLAs, and fastest throughput.

Enterprise procurement steps

1. Discovery: workflow review, volume forecast, and success criteria.
2. Security and compliance: questionnaire, SOC 2 package, architecture review.
3. Pilot: 2–4 week proof-of-value with success metrics and data samples.
4. SOW and pricing: finalize tier, commit volume, and add-on services.
5. Legal: MSA, DPA, and InfoSec approvals.
6. PO and go-live: provisioning, onboarding, templates, and success plan.

FAQ: billing and pricing

• Do I pay for failed pages? No charge for system errors; retried pages bill once.
• Can I mix billing models? Yes—subscribe for baseline volume and use per-page for bursts.
• Is there a free trial? Yes, 14 days and 500 pages.
• How are pages counted? Each PDF page processed; per-document option covers up to 5 pages.
• What drives enterprise cost? Monthly page volume, custom templates, SLAs, security (SSO/VPC), and on‑prem deployment.

Implementation and onboarding

A concise, practical onboarding guide for procurement, IT, and power users to implement document-to-spreadsheet automation from pilot to full rollout.

This onboarding plan prioritizes a scoped pilot, measurable validation, and a phased rollout to scale document-to-spreadsheet and PDF to Excel automation with low risk.

Follow the timeline, checklist, and roles below to accelerate time-to-value while maintaining governance and accuracy.

Timeline and milestones (Gantt-style)

Pilot checklist and acceptance

• Sample set: 150–300 docs across 5–10 templates, including edge cases.
• Use real data for invoices, POs, contracts, bank statements.
• Target accuracy: header fields 95%+, line-item fields 90%+, table row match 88%+.
• Straight-through processing (no human touch): 60%+ in pilot, trend improving.
• Cycle time: ≤5 minutes per document average including human review.
• Security: SSO enabled, roles/permissions verified, audit log on.
• Integrations: ERP/GL export to CSV/XLSX/Sheets validated.
• Human-in-the-loop queue active with confidence thresholds.
• Acceptance criteria documented and signed by FP&A, IT, Procurement.
• Go/no-go: metrics met for 2 consecutive pilot weeks.

Configuration steps

1. Upload sample templates and historical PDFs; label 20–30 gold-standard docs.
2. Map common fields (vendor, dates, amounts, line items, cost centers).
3. Set confidence thresholds (auto-approve 95%+, route to review 80–95%, reject <80%).
4. Enable human review queues, dual-approval for payments, and audit trails.
5. Configure document-to-spreadsheet exports (XLSX/CSV), naming, destinations.
6. Connect SSO, provision roles, and set data retention and PII redaction.

Training plan

• 1-hour product demo: capture, review, export, metrics.
• 2–3 hands-on workshops by role (procurement, FP&A, power users).
• Admin training (60–90 min): mappings, thresholds, queues, SLA dashboards.
• Office hours during pilot and first rollout wave.

Roles and responsibilities

Success metrics

• Accuracy (field/line-item/table).
• Straight-through rate and exception rate.
• Cycle time per document.
• User adoption and review backlog age.
• Cost per document vs baseline.
• Integration success and data freshness SLAs.

Escalation matrix

FAQ: common onboarding blockers

• Q: Not enough sample documents? A: Pull last 3–6 months, include variants and edge cases.
• Q: Low accuracy on tables? A: Add labeled examples, tighten column anchors, adjust thresholds.
• Q: Review backlog grows? A: Raise auto-approve threshold only for high-confidence fields and add reviewers.
• Q: Export mismatches ERP? A: Reconcile field types, date/number formats, and chart-of-accounts mapping.

Post-implementation optimization

• Retrain with corrected reviews weekly in month 1, then monthly.
• Add new template variants and vendors via controlled playbooks.
• Tune thresholds by BU to balance accuracy and throughput.
• Automate quality alerts for drift and SLA early warning.
• Quarterly business reviews to expand use cases and ROI.

Customer success stories and case studies

Three anonymized case study examples show how document automation and PDF to Excel workflows improved M&A CIM review, treasury bank-statement reconciliation, and AP invoice processing with measurable before/after results, ROI, and process changes.

These case studies highlight conservative, benchmark-based outcomes across core PDF to Excel and document automation scenarios. Each includes a clear problem, deployed solution, quantified results, and a direct customer quote.

Before/after metrics and deployment timeline

Case study: M&A due diligence (CIM extraction)

Customer profile and problem: A mid-market private equity firm (50 employees; VP of M&A sponsor) manually re-keyed data from PDF CIMs to Excel for comps and models. Volume averaged 6 CIMs/month at 8 hours/CIM with 2% keying errors and inconsistent tables.

Solution deployed: Document automation with PDF to Excel extraction, custom CIM templates, multi-table capture, and redaction. Integrated with Box, Excel, and Salesforce; SSO and audit trails enabled an exception-based review workflow.

Quantified results: Time per CIM fell from 8 h to 2 h; monthly manual hours dropped from 48 to 12 (-75%). Accuracy improved from 92% to 98.7%, and straight-through processing reached 82%. Estimated savings $72k/year with 6-month payback.

Customer quote: Automation turned our CIM review into a same-day exercise without sacrificing quality, noted the VP of M&A.

Case study: Treasury reconciliation (bank statements)

Customer profile and problem: A global manufacturer’s treasury team (8 staff; Director of Treasury sponsor) reconciled 500 bank statements/month across 12 banks. Manual entry averaged 15 minutes/statement with a 3% error rate and a 6-day month-end close.

Solution deployed: Bank-statement parser with PDF to Excel/CSV export, format normalization, and rules-based matching. Integrated to ERP and a treasury workstation via API; alerts routed exceptions to a shared queue.

Quantified results: Processing time dropped to 2 minutes/statement (-87%), saving 108 hours/month. Errors fell from 3% to 0.2%; straight-through rate exceeded 85%; month-end close shortened from 6 to 3 days. Estimated annual labor savings about $45k with 4-month payback.

Customer quote: We now reconcile in hours, not days, and our audit binders practically build themselves, said the Director of Treasury.

Case study: Accounting operations (invoice automation)

Customer profile and problem: A regional distributor’s AP team (5 staff; Controller sponsor) processed 4,000 invoices/month, mostly emailed PDFs. Manual entry consumed 40 hours/month per clerk; error rate was 1.5% with frequent duplicates.

Solution deployed: Invoice capture with PDF to Excel line-item extraction, 3-way match to POs and receipts, duplicate detection, and GL-coding rules. Integrated with NetSuite; introduced auto-approval thresholds and exception queues.

Quantified results: Manual entry time fell from 40 to 4 hours per clerk (-90%); team hours dropped from 200 to 20/month. Straight-through reached 78% and errors fell from 1.5% to 0.4%. Estimated annual savings $75k; ROI 2.7x with 5-month payback.

Customer quote: We stopped typing and started managing exceptions. Close is smoother, and vendors get paid faster, noted the Controller.

Support, security, and documentation

Clear support tiers, sample SLAs, and a transparent security posture for PDF automation—plus links to the documentation you need.

This section outlines support options and escalation, security controls with proof points, and where to find documentation for PDF automation.

Support tiers and escalation

Choose the support level that fits your team. Enterprise customers receive a dedicated CSM and priority SLA. All customers have access to the status page and public roadmap.

• Community (no-cost): Community forum, knowledge base, product updates; email acknowledgment next business day; support hours Mon–Fri 9am–6pm local.
• Standard (business): Support portal + email; 24x5 coverage; P1/P2/P3 response targets per SLA; service credits for missed uptime; status page subscriptions.
• Enterprise: 24/7 global P1 hotline, dedicated CSM, prioritized queue, quarterly business reviews; 99.9% or 99.99% uptime SLA; custom DPA and security reviews.

• Escalation path: Tier-1 support → duty engineer → on-call incident lead → incident commander → executive sponsor.
• Triggers: If no workaround in 2 hours (P1) or 1 business day (P2), auto-escalate to next level; customer can request manual escalation via portal or hotline.

Sample SLA targets (response and updates)

Security and compliance

Our security controls are designed for sensitive financial documents and PDF automation at scale. Evidence and reports are available on request under NDA.

• Encryption: AES-256 at rest with cloud KMS and annual key rotation; TLS 1.2+ (TLS 1.3 preferred) in transit; HSTS and perfect forward secrecy.
• Data residency: Customer-selectable US or EU regions; data stored and processed in-region; logically isolated tenants.
• Access controls: SSO via SAML 2.0/OIDC; RBAC with least privilege; SCIM provisioning; audit logs retained 12 months.
• Vulnerability management: External penetration testing twice per year; quarterly vulnerability scans; remediation targets P1 72 hours, P2 14 days.
• Secure SDLC: Mandatory code review, dependency scanning, secrets management, infrastructure-as-code with change approvals.
• Business continuity: Encrypted backups daily; 35-day retention; RPO 24 hours, RTO 4 hours; multi-AZ deployment.
• Compliance: SOC 2 Type II attested; latest report available under NDA. ISO 27001 certification in progress; target completion Q4 2025. GDPR-compliant processing with DPA available.

Documentation resources

Find everything you need to build, integrate, and troubleshoot PDF automation.

• API reference: https://docs.example.com/api
• Integration guides (ERP/AP/GL): https://docs.example.com/integrations
• Developer SDKs (Python, Node, Java): https://github.com/example/pdf-automation-sdks
• Mapping template library: https://templates.example.com/library
• Troubleshooting knowledge base: https://support.example.com/kb
• Video tutorials and webinars: https://videos.example.com/pdf-automation
• Release notes and changelog: https://docs.example.com/changelog
• Status and uptime: https://status.example.com
• Support portal: https://support.example.com

Recommended support workflows for finance teams

Use these workflows to keep invoice and statement extraction accurate and auditable.

• Submit a parsing exception: Open a ticket in the support portal and attach the source PDF, redacted if required, plus your expected fields and output schema.
• Request a new template: Provide 3–5 representative PDFs, vendor name, and required fields; we create or extend a mapping template from the library.
• Escalate an accuracy issue: Include job IDs, confidence scores, environment (prod/sandbox), and recent changes; invoke P2 if workarounds are available, P1 if blocking payables.

1. Collect evidence: PDF sample, job/run ID, timestamps, expected vs actual fields.
2. Classify severity (P1/P2/P3) using the SLA table above.
3. Submit via support portal or call the P1 hotline (Enterprise) with all artifacts.
4. Track updates on the ticket and status page; join the incident bridge if P1.
5. Verify the fix in sandbox, then production; close the ticket with acceptance notes.

Competitive comparison matrix

An analytical competitive comparison for PDF to Excel and document extraction buyers. The matrix contrasts Sparkco with AWS Textract, Google Document AI, ABBYY Vantage, Adobe PDF Services, Docparser, and Nanonets across accuracy, stitching, formulas, integrations, API, security, deployment, pricing transparency, and support.

Use this competitive comparison to shortlist vendors for PDF to Excel automation. It emphasizes table accuracy, multi-page stitching, and whether exports preserve formulas and formatting—critical for financial and operational reporting.

PDF to Excel and document extraction: competitive comparison matrix

Where competitors win

A balanced competitive comparison shows clear areas of strength beyond Sparkco.

• AWS Textract: Hyperscale, pay-as-you-go, and deep AWS integrations; ideal when you already orchestrate data in S3/Lambda [aws.amazon.com/textract].
• Google Document AI: Strong structured-document accuracy and tight GCP integration for ML pipelines [cloud.google.com/document-ai].
• ABBYY Vantage: Best-in-class language coverage and enterprise-grade on-prem deployments with robust table capture [abbyy.com/vantage].
• Adobe PDF Services: Familiar tools and fast ad hoc PDF to Excel exports for business users [developer.adobe.com/document-services].

Where Sparkco wins

Sparkco differentiates on analyst-grade Excel output and finance-specific automation in this competitive comparison.

• Formula-preserving Excel exports: Reconstructs sums/subtotals and maintains styles so spreadsheets remain analysis-ready.
• Specialized templates: Cash-flow, P&L, and CIM table parsers tuned for messy, multi-page exhibits common in financial diligence.
• Auto-stitched tables: Multi-page tables are unified with header continuity, reducing downstream engineering.
• Transparent limits: Sparkco supports fewer languages than ABBYY and achieves best accuracy with light template configuration on novel layouts.

Rebuttals to common objections

• Textract/DocAI are cheaper: True on per-page rates, but TCO rises with custom stitching, post-processing, and QA. Sparkco’s formula logic and stitching cut engineering hours.
• We already run ABBYY on‑prem: Keep ABBYY for broad OCR; add Sparkco for finance workstreams needing analysis-ready Excel with formulas.
• Adobe export is enough: Great for one-offs, but it lacks APIs, stitching, and formula logic required for repeatable, auditable data flows.

Buyer checklist for PDF to Excel vendor selection

• Measure table accuracy on messy, multi-page samples with merged cells and footers.
• Verify multi-page stitching and header carry-forward across breaks.
• Confirm Excel output: values only or reconstructed formulas and formatting.
• Assess API maturity, webhooks, SDKs, and idempotent retries.
• Review security: SSO, audit logs, data residency, and compliance attestations.
• Check deployment options (cloud, VPC, on‑prem) and latency/throughput SLAs.
• Ensure pricing transparency and model fit (usage vs seats).
• Validate support tiers, response SLAs, and solution engineering availability.

FAQ, resources, and call-to-action

Your high-impact FAQ for PDF to Excel document automation—concise answers, trusted resources, and clear next steps.

FAQ (collapsible Q&A)

Each item below is a collapsible Q&A. Click to expand for quick facts and links to deeper docs.

• Q: How is pricing structured? A: Tiered SaaS based on pages processed, features, and support; volume discounts and annual savings available. See pricing: /docs/pricing
• Q: How accurate is PDF to Excel extraction? A: 90–99% on clean, structured docs; lower on scans or complex tables. Confidence scores and human review included. Accuracy guide: /docs/accuracy
• Q: Where is my data stored (data residency)? A: Choose US, EU, or APAC; data remains in-region per your selection. Residency and retention: /docs/data-residency
• Q: Do you offer on-prem or private cloud? A: Yes—Kubernetes-based deployment for VPC or on-prem with feature parity to cloud. Deployment options: /docs/deployment
• Q: How easy is integration? A: REST API, SDKs (Python, JS), webhooks, and no-code connectors. Typical build is hours, not weeks. Integrations: /docs/integrations
• Q: How fast can we ramp? A: Most teams ship a first workflow in 1–2 days; broader rollout in 1–2 weeks. Quickstart: /docs/quickstart
• Q: Is there a free trial? A: Yes—14 days, 500 pages, API at 5 req/s, and watermarked exports on free tier. Start: /signup/trial
• Q: What sample files do you need? A: 5–10 real PDFs and your target Excel/CSV schema; redact PII unless under NDA. Sample checklist: /docs/samples
• Q: Can you handle complex layouts and tables? A: Yes—table structure detection, multi-column pages, variable templates; route low-confidence items to review. Complex docs: /docs/layouts
• Q: How do you secure data? A: SOC 2 Type II, ISO 27001, encryption in transit/at rest, SSO, RBAC, audit logs, configurable retention. Security whitepaper: /resources/security-whitepaper.pdf

Resources

• Type: Demo request — Audience: Buyers and evaluators — Link: /request-demo
• Type: API docs — Audience: Developers — Link: /docs/api
• Type: Implementation guide — Audience: Solutions engineers and admins — Link: /docs/implementation
• Type: Security whitepaper (PDF) — Audience: Security and compliance — Link: /resources/security-whitepaper.pdf
• Type: Case study (Fintech AP automation, PDF) — Audience: Finance ops leaders — Link: /resources/case-studies/fintech-ap.pdf
• Type: Case study (Logistics billing, PDF) — Audience: IT and operations — Link: /resources/case-studies/logistics-billing.pdf

Call to action

Pick your next step below—sales-led for tailored scoping, or self-serve to validate PDF to Excel in your environment.

• Button: Request a personalized demo — What to expect: Book in under 60 seconds; confirmation within 1 business day; a 30–45 minute session covering your use case, live PDF to Excel on your files, and a clear rollout plan and quote. Prepare: 2–3 sample PDFs, target Excel/CSV fields, monthly volume, systems to integrate, and security questions. Link: /request-demo
• Button: Start a free trial — What to expect: Instant access; 14 days; 500-page limit; API 5 req/s; watermarked exports on free tier; includes starter PDF to Excel templates and sample data. Prepare: Create workspace, upload 5–10 sample files, map fields, set a webhook or export to Excel, invite a teammate. Link: /signup/trial