Using longitudinal surveillance records from the 2018–2020 eastern DRC Ebola outbreak and archived forecasts from Sierra Leone's Western Area, this methodological study demonstrates that analytical readiness is estimand-specific and vintage-specific — not a global property of a dataset. A reported case-fatality ratio of 0.40 was statistically compatible with true underlying fatality rates spanning 0.10 to 0.80, an eightfold range, even as larger denominators reduced sampling uncertainty without touching structural uncertainty. Seven-day persistence forecast errors swelled to 18 cases during early surveillance recovery — larger than during acute disruption — and four-week trend multipliers reversed direction (0.67→1.29), illustrating that different analytical targets reach readiness asynchronously within the same dataset.
This preprint, not yet peer-reviewed, addresses a persistent blind spot in outbreak response: decision-makers routinely commission analyses before surveillance systems have stabilized or epidemic dynamics have generated sufficient signal. The study's core contribution — the R(θ,v) readiness index that formally links estimand definition, observation-process awareness, and decision-matched validation — is methodologically rigorous and practically actionable. For public health practice, the implications are significant: wider confidence intervals or model non-convergence are informative signals, not failures to be engineered away with more data. The framework is confirmatory of epistemological concerns long voiced in outbreak analytics literature but provides the first operationalized joint criterion. Limitations include retrospective application to a single epidemic and dependency on archived, imperfect records. Independent prospective validation will be essential before adoption in real-time response protocols.