Methodology

How we grade ROI evidence.

Every ROI benchmark in the Case Ledger is graded on a three-tier evidence scale. The grade is always shown alongside the number — you should never have to wonder how confident we are in a figure.

Evidence grades

Primary source

The ROI figure is drawn directly from a primary source that can be independently verified. This is the highest confidence tier.

SEC filings and earnings call transcripts with named figures
Published first-party case studies with specific metrics and methodology disclosed
Direct interviews with named practitioners who ran the deployment
Peer-reviewed academic studies with disclosed dataset and controls

Verified secondary

The figure comes from a reputable secondary source that cites identifiable primary data. The chain is traceable, but one step removed.

Analyst reports (Gartner, McKinsey, Forrester) with disclosed sourcing
Journalism citing named organisations with checkable primary data
Industry surveys with disclosed sample size and methodology
Vendor case studies where the customer organisation is named and corroborating evidence exists

Self-reported

The figure is self-reported by a vendor or deployment team without independent verification. We include C-grade data because it may still be directionally useful, but it is always clearly labelled.

Vendor press releases or product pages with unnamed customer quotes
Blog posts and social media claims without linked primary sources
Conference talk assertions without accompanying methodology
Internal case studies shared without external audit or disclosure

Normalization model

Raw ROI figures from different sources are not directly comparable — they use different baselines, time horizons, and cost inclusions. We normalize all benchmarks to a shared model before publishing a range.

Time horizon

All figures are normalized to a 12-month post-deployment window unless the source specifies a longer period and the methodology supports it.

Cost basis

Figures include direct implementation costs (licensing, development, integration) but exclude indirect organizational change costs unless explicitly stated in the source.

Baseline

ROI is expressed as improvement over the pre-automation baseline process, not over a theoretical optimum. This avoids inflating figures for processes that were already partially optimized.

Range not point estimate

We publish lower and upper bounds rather than a single figure. The lower bound reflects conservative assumptions; the upper bound reflects the source's stated outcome under favourable conditions.

Versioning and corrections

Each benchmark record carries a version identifier. When a source is updated, retracted, or a calculation error is discovered, we publish a correction note alongside the updated figure rather than silently replacing it. Historical snapshots remain accessible.

If you believe a benchmark is incorrectly sourced or graded, contact us at support@caseledger.ai with the record slug and the source you believe is more accurate. We will review and publish a correction if warranted.

How we grade ROI evidence.

Every ROI benchmark in the Case Ledger is graded on a three-tier evidence scale. The grade is always shown alongside the number — you should never have to wonder how confident we are in a figure.

Evidence grades

Primary source

The ROI figure is drawn directly from a primary source that can be independently verified. This is the highest confidence tier.

SEC filings and earnings call transcripts with named figures
Published first-party case studies with specific metrics and methodology disclosed
Direct interviews with named practitioners who ran the deployment
Peer-reviewed academic studies with disclosed dataset and controls

Verified secondary

The figure comes from a reputable secondary source that cites identifiable primary data. The chain is traceable, but one step removed.

Analyst reports (Gartner, McKinsey, Forrester) with disclosed sourcing
Journalism citing named organisations with checkable primary data
Industry surveys with disclosed sample size and methodology
Vendor case studies where the customer organisation is named and corroborating evidence exists

Self-reported

The figure is self-reported by a vendor or deployment team without independent verification. We include C-grade data because it may still be directionally useful, but it is always clearly labelled.

Vendor press releases or product pages with unnamed customer quotes
Blog posts and social media claims without linked primary sources
Conference talk assertions without accompanying methodology
Internal case studies shared without external audit or disclosure

Normalization model

Time horizon

All figures are normalized to a 12-month post-deployment window unless the source specifies a longer period and the methodology supports it.

Cost basis

Figures include direct implementation costs (licensing, development, integration) but exclude indirect organizational change costs unless explicitly stated in the source.

Baseline

ROI is expressed as improvement over the pre-automation baseline process, not over a theoretical optimum. This avoids inflating figures for processes that were already partially optimized.

Range not point estimate

We publish lower and upper bounds rather than a single figure. The lower bound reflects conservative assumptions; the upper bound reflects the source's stated outcome under favourable conditions.

How we grade ROI evidence.

Evidence grades

Primary source

Verified secondary

Self-reported

Normalization model

Versioning and corrections

Start using the ledger today

How we grade ROI evidence.

Evidence grades

Primary source

Verified secondary

Self-reported

Normalization model

Versioning and corrections

Start using the ledger today