Most discovery cost overruns are not caused just by inefficient review, but by uncertainty introduced long before review ever begins. Early Case Assessment (ECA) is therefore not simply a technology capability. At its best, it is governance applied early through behaviours that guide the supporting technology to reduce unnecessary data before collection begins.
I often compare this to training my dog, Oisín (Ush-sheen). Boundaries must come before freedom. If you open the gate before setting rules, any chaos that follows is not the dog’s fault, it is the result of the process you put in place.
Across matters I’ve supported, discovery costs escalated because decisions about scope were delayed until after collection, when direction had already been set and options had narrowed. By the time review begins, organisations are no longer deciding what matters, they are managing the consequences of earlier uncertainty.
Modern organisations rarely suffer from reviewing too little data. They suffer from collecting too much without clear intent. Over‑collection is often mistaken for defensibility, when in reality it delays insight and increases cost. In practice, this distinction determines whether discovery becomes an investigative process or an exercise in cost management.
Aligning ECA Across the Discovery Lifecycle
In many matters, data is exported from internal systems into review platforms before scope is fully understood. Effective ECA reduces volume at the source, allowing review teams to prioritise the most potentially relevant information from the outset.
Why Traditional ECA Fails
Traditional approaches often fail when:
• Legal and IT are not aligned early, resulting in miscommunication of objectives
• Review platforms are asked to solve scoping problems
• Over‑collection is mistaken for defensibility
A 5‑Stage Progression Aligned to Dog Training and ECA Maturity
- Setting the Boundaries — Scope before export
- Observing Behaviour — Early analysis and hypothesis testing
- Teaching Focus — Remove noise and duplication
- Understanding Meaning — Natural Language Processing (NLP) and conceptual intelligence
- Building Confidence — Insight gained
Step 1: Setting the Boundaries
The strongest matters begin with conversations rather than collections, where legal, IT, and business stakeholders agree what matters before anything is exported.
- Narrowing custodians
- Date ranges
- Locations
- Communication types
These reduces export volumes without excluding potentially relevant data.
As with training Oisín, boundaries introduced early create confidence later. Discovery that begins without scope does not become more defensible, it becomes harder to control.
Early involvement at this stage often reduces export volumes dramatically by aligning legal intent with technical execution before collection begins.
Step 2: Observing Behaviour
Early analysis often feels uncomfortable because it slows the urge to act, but it is where teams begin to understand what the data is actually telling them. Early searches and filtering allow teams to test assumptions and understand data patterns before committing significant review effort.
- Keyword searching
- Date filtering
- Domain analysis
These approaches allow teams to confirm relevant time periods and focus attention on communications most likely to contain risk.
Similar to training a dog, before you make corrections you observe the behaviours, triggers and patterns to fully understand what truly needs your attention.
Step 3: Teaching Focus
Review teams frequently report that the hardest part of discovery is repetition, seeing the same information surface again and again in slightly different forms. A well-trained dog learns not every stimulus deserves a reaction. Discovery should work the same way. Without volume control, review teams repeatedly chase the same information in slightly different forms, the legal equivalent of barking at every car.
- Deduplication
- Near‑duplicate detection
- Email threading
- Removal of junk data
Step 4: Understanding Meaning
These reduce review volumes while preserving relevance. This allows teams to focus review effort only where meaningful signal remains.
Training progresses when a dog understands context rather than simply reacting to commands.
At this stage, teams shift from searching for documents to understanding conversations. NLP and categorisation allow teams to move beyond keyword reactions toward conceptual understanding, revealing how information connects across datasets.
- Document categorisation
- Language detection
Both assist with defensible prioritisation and reduce manual guesswork.
Step 5: Building Confidence
The difference between reactive and confident legal strategy becomes visible when insight replaces uncertainty and decisions begin arriving earlier.
- Document summarisation
- Timeline visualisation transform large data sets into understandable narratives
Both support faster GC‑level decision‑making. Organisations that delay judgement inherit cost and complexity; those that apply structure early gain clarity before scale.
Like training Oisín, discovery succeeds when guidance comes before freedom. The objective is not to restrict exploration, but to ensure effort is directed where it matters most. When discovery is trained from the start, Teams spend less time digging and more time understanding what they have found.
From Collection to Clarity
Because good discovery, like good training, begins long before action.
ECA is often described as an efficiency exercise, but in practice it is a decision‑making discipline grounded in structured behaviour. Organisations that define scope early do not simply reduce data volumes, they reduce uncertainty. Discovery shifts from reacting to information toward directing investigation.
When structure is applied at the beginning, legal teams gain visibility earlier, business stakeholders gain confidence sooner and General Counsel can make informed strategic decisions before cost and complexity scale beyond control.
Like training Oisín, structure does not restrict exploration, it enables it. The objective is never to limit discovery, but to ensure effort is directed where it matters most.
When discovery is trained from the start, teams spend less time digging blindly and more time understanding what they have found.
