Case Study 01: Navigating Complexity

How to find the story when you can't read all the documents.

60 documents, 2,220 pages, evidence-first navigation.

---

All evidence excerpts in this case study are unaltered Accounter output, validated at the character level against source documents.

The Challenge: Information Overload

An investigator is tasked with understanding the UK's High Speed 2 (HS2) rail project, a decade-long initiative with a history of delays, budget revisions, and scope changes. The source material comprises 60 documents from four different public bodies, totaling 2,220 pages of annual reports, audit findings, and committee transcripts.

Reading this volume of material is not feasible. A manual review would require weeks, and it would be nearly impossible to connect disparate facts spread across thousands of pages. The core challenge is not just finding facts, but finding the narrative threads hidden within the noise.

---

The Process: From 2,220 Pages to 648 Signals

Accounter processed the entire 60-document corpus, reading exhaustively through body text, appendices, and footnotes. The process was designed not to find a single answer, but to create a permanent, navigable map of the evidence.

Step	Action	What Happens
01	Build the Document Library	60 PDFs uploaded. Metadata extracted automatically. Investigator reviews and corrects labels before processing.
02	Run Extraction	Every page read. Substantive claims captured with verbatim source text. 18 minutes for this corpus.
03	Review the Ledger	2,253 signals extracted. Auto-tagged by theme (Risk, Governance, Finance) and rated by salience.
04	Filter by Salience	Filter to High and Exceptional to surface the most significant claims first.
05	Navigate	648 high-focus signals ready to explore by source, theme, or salience.

Instead of a mountain of pages, the investigator now has a structured evidence base where every signal is linked to its original, verbatim source text. No paraphrasing, no summarization—only the evidence.

---

The Navigation: Three Paths to Insight

With 648 high-quality signals, the question becomes: "Where do I start?" The Signal Ledger can be filtered, sorted, and searched in multiple ways—by source document, by tag, by salience, by date, by keyword, and more. Below, we explore three common starting points and what each reveals.

Path 1: By Source — "Where's the action?"

Not all documents are created equal. The source view immediately shows that Public Committee transcripts are the richest seam, with 262 signals at 1.47 per page—nearly double the density of Annual Reports. This is where the arguments are happening: conflicting statements, direct questioning, and candid admissions.

Path 2: By Tag — "What are the recurring problems?"

Accounter automatically tags each signal based on its content. The tag view reveals the project's persistent fault lines at a glance: Risk dominates with 211 signals, followed by Governance, Funding, Finance, and Delivery. Clicking any tag filters the ledger instantly, letting the investigator dive deep into a specific issue.

Filtering to Governance reveals 79 signals. The most recent parliamentary report contains a signal that hints at serious problems:

Path 3: By Salience — "Show me the smoking guns."

For the fastest path to the most critical information, the investigator can filter for Exceptional salience signals. This instantly narrows the field from 648 signals to just 34. These are the most significant, high-impact events and statements: project cancellations, major cost escalations, safety incidents, and damning admissions.

Note: We're filtering to High and Exceptional here—but don't discard those Medium signals. When following a specific thread (as we'll show in Case Study 02), routine procurement announcements become crucial context for the story.

Scanning the Exceptional list, one name appears repeatedly: Euston station. The budget blowout, the construction pause, the £6.5 billion in cancelled scope—all tied to this single location.

This is a thread worth pulling. In Case Study 02, we follow the Euston story from first warning to cancellation.

Explore the evidence for yourself (648 signals)

Browse the live database that built this case study. Every signal is verbatim, traceable, and filterable by source, theme, or salience.

---

The Outcome: From Overload to Insight in Minutes

By using Accounter to navigate the document set, the investigator has, in a matter of minutes, achieved what would have taken weeks of manual effort:

• Identified the most fertile ground for investigation — Public Committee transcripts, with their conflicting statements and direct questioning.
• Mapped the recurring fault lines — Risk, governance, and cost emerge as the dominant tags across a decade of documents.
• Isolated a shortlist of critical signals — 34 Exceptional-rated signals ready for immediate follow-up, with Euston emerging as a clear thread.
• Built a permanent evidence base — The Signal Ledger persists. New documents can be added, new runs conducted, and the evidence revisited at any time.

Accounter reads exhaustively—body text, appendices, and footnotes receive the same scrutiny—so no signal is missed. Every finding is grounded in verbatim text, validated at the character level. There is no inference or hallucination.

The investigator has not just found isolated facts; they have discovered the shape of the story. The next step is to take one of these threads and pull on it.

---

Next Steps: Following the Thread

This case study demonstrates how Accounter enables rapid navigation and discovery. The investigator is now equipped with multiple, high-value starting points.

One thread stands out: across multiple source groups, the name Euston Station appears repeatedly—in warnings, cost revisions, and heated committee exchanges. It's a thread rich enough to warrant its own investigation.

Case Study 02: Following the Thread shows how an investigator takes this single line of inquiry and uses a targeted keyword run to build a detailed, evidence-grounded timeline of how the budget spiraled out of control.

Document Corpus Summary

Source Group	Documents	Pages	High+ Signals
Public Committees	30	558	262
HS2 Ltd Annual Reports	12	1,169	170
National Audit Office	9	368	161
Bi-annual Parliamentary Reports	9	125	55

All source documents are publicly available government records.

---