Scope & Boundary Definition.
Scope determines the territory in which a research claim is allowed to be true. Boundaries define what enters the evidence system, what remains outside it, which conditions must match and when collection must stop. Without them, “more data” quietly becomes a different study.
Scope is not page length. It is the domain of valid inference.
A result can be technically correct and methodologically misleading when it is interpreted outside the population, time, market, source state or depth that produced it.
Scope and boundary definition is the explicit specification of the conceptual, observational, contextual, temporal and operational limits within which evidence is collected and conclusions remain valid.
The sample is the subset actually observed under a documented selection rule.
A filter changes a view; a boundary determines whether an observation belongs in the study.
It states where the evidence stops speaking rather than pretending the evidence is universal.
Four territories. Four different meanings.
Researchers frequently collapse the theoretical universe, eligible population, requested records and usable observations into one number. They must remain separate.
Conceptual universe
Everything that could theoretically belong to the phenomenon: all relevant queries, pages, entities, sources or generated answers.
Eligible population
Everything meeting the declared definitions, location, language, source, time, type and quality rules.
Requested population
Records submitted to a provider, crawler or model before failures, null states, caps and unavailable observations.
Analyzed sample
Usable returned observations that survived declared validation and missing-data treatment.
A complete scope closes six escape routes.
Each boundary blocks a different kind of silent study drift. Missing one may change the meaning of every result.
Construct boundary
Define what “visibility”, “coverage”, “competitor”, “citation” or “entity” means operationally—and what it does not mean.
Unit boundary
Specify whether one observation is a query, result, URL, canonical page, domain, entity, prompt, answer or source relation.
Eligibility boundary
Write inclusion, exclusion, deduplication and missing-state rules before records are inspected.
State boundary
Fix geography, language, device, engine, interface, model, source and other observation-changing conditions.
Time boundary
Separate snapshot, collection window and longitudinal interval. Never merge asynchronous states as simultaneous evidence.
Depth and stopping boundary
Fix result depth, request cap, crawl depth, prompt repetitions, saturation rule, budget and termination event.
One boundary architecture. Four research environments.
Every tab contains complete fallback content in the HTML. JavaScript changes the active protocol only after execution is confirmed.
Fixed query corpus under synchronized result conditions
Decide admission before seeing the winner.
Rules written after inspection can be unconsciously adjusted to preserve a preferred story. Predeclared rules make inclusion, exclusion and exception handling inspectable.
The observation represents the declared construct and unit without reinterpretation.
Location, language, source type, device, interface and time satisfy the protocol.
Source, timestamp, identifier and relevant returned state can be preserved.
Normalization, identity resolution and quality rules do not produce an unresolved conflict.
Example: paid result inside an organic visibility study.
Example: mobile observations silently mixed with desktop results.
Canonicalization collapses alternate representations of the same analytical unit.
Failure remains counted in the missingness record even when excluded from analysis.
Four precise scopes. Four explicit claim ceilings.
These examples show why scope is part of the meaning of the result, not metadata appended after analysis.
A new request can silently become a new study.
When an observation crosses a declared boundary, do not append it for convenience. Reject it, record it as an exception or open a versioned extension with its own comparability statement.
Collection ends by rule, not by fatigue or preference.
The stopping condition belongs in the protocol before the first observation. Different research systems require different termination logic.
All eligible units processed
Use when the bounded population is known and small enough to attempt complete collection.
STOP WHEN eligible_processed = eligible_totalPredeclared count reached
Use when sample size and selection procedure are fixed before observation.
STOP WHEN valid_n = declared_nNo qualifying new category
Use only with a documented novelty definition and consecutive-observation rule.
STOP AFTER k runs without new classBudget or request ceiling
A practical cap is valid when disclosed and reflected in the conclusion—not hidden as completeness.
STOP AT request_cap; label truncatedThe scope should be reconstructable from one record.
This minimum manifest travels with the dataset so later analysts do not need to infer which observations were eligible or comparable.
Every method node. One controlled research route.
MTH/03 closes the study territory. MTH/04 determines how observations will be selected and collected inside that territory.