Recursos · 26

Ejecute un experimento A/B que respalde una decisión acertada

Especificar hipótesis, asignación, métrica principal y barreras de seguridad antes de comparar variantes.

Actualizado · 6 min

Laboratorio de medición ficticio sin personas: dos aparatos negros idénticos, lámpara de latón y cuaderno burdeos Ilustración editorial generada con IA

Lo que esta guía ayuda a lograr

  • Preregistrar el protocolo
  • Consultar asignación
  • Vigilar barandillas
  • Decidir con incertidumbre

Comprobación rápida

  • ¿Qué hipótesis concreta se podría refutar?
  • ¿Qué unidad recibe la variante?
  • ¿La asignación observada coincide con el plan?
  • ¿Qué daño amerita una pausa?
  • ¿Se establecieron la duración y los segmentos antes de que se vieran los resultados?

Método paso a paso

  1. 1

    Indique el cambio propuesto

    Escriba el mecanismo esperado, la población objetivo y el efecto en una métrica principal. Describir los posibles riesgos de accesibilidad y servicio.

    Entregable: hipótesis fechadas y criterios de éxito.

  2. 2

    Definir asignación

    Elija la unidad de aleatorización y mantenga la asignación estable entre las visitas. Excluya bots y especifique segmentos fuera de alcance antes del lanzamiento.

    Entregable: protocolo de asignación y exclusiones.

  3. 3

    Instrumento y prueba

    Verificar eventos de exposición, conversión y guardarraíl para ambas variantes. Pruebe los estados móviles, de rechazo de consentimiento y de error.

    Entregable: Registro de QA de ambos recorridos.

  4. 4

    Comprobar proporción de muestra

    Comparar las asignaciones observadas y esperadas antes de leer el efecto del tratamiento. Una discrepancia puede exponer defectos de asignación, medición o orientación.

    Entregable: verificación de ratios y diagnóstico de discrepancias.

  5. 5

    Analizar en el tiempo planificado

    Respetar la ventana elegida y reportar incertidumbre, métrica primaria, daños y análisis de segmentos planificados. Inspeccione las sorpresas sin seleccionar sólo el número favorable después del hecho.

    Entregable: análisis con limitaciones e incertidumbre.

  6. 6

    Decidir y dar seguimiento

    Asignar un propietario para implementación, abandono u otra prueba. Después del lanzamiento, verifique que el efecto y las barreras de seguridad sigan siendo consistentes.

    Entregable: decisión archivada y verificación post-lanzamiento.

Indicadores de gestión

IndicadorQué midePrimera acción
AsignaciónProporción de muestra esperada versus observadaDiagnóstico antes del análisis
ExposiciónUsuarios elegibles realmente expuestosReparar segmentación
EfectoCambio e incertidumbre en métricas primariasAplicar regla de decisión planificada
BarandillasErrores, latencia y abandono por variantePausar tratamiento nocivo

Errores comunes

  • Parar cuando un número parece favorable por primera vez
  • Recopilando muchas métricas sin una principal
  • Ignorar la discrepancia en la proporción de muestras
  • Tratar un pequeño segmento como prueba general

Preguntas frecuentes

¿El ratio de muestreo es una formalidad menor?

No. Una discrepancia inexplicable puede hacer que la comparación no sea confiable.

¿Se debe comprobar la conversión diariamente?

El seguimiento operativo ayuda, pero cambiar la duración o la regla de decisión en respuesta a los resultados socava la interpretación planificada.

¿Qué pasa si el resultado es incierto?

Informar la incertidumbre y el tamaño del efecto plausible, luego decidir si otro experimento tiene valor real.

Distribución A/B: datos y SRM

Las notas detalladas de esta edición están disponibles en inglés.

A conversion difference becomes difficult to interpret when compared groups no longer match the assignment plan. Sample Ratio Mismatch (SRM) is a data quality signal: investigate allocation before evaluating the product change. This dossier connects calculation, instrumentation and decisions without treating a p-value as overall validation of an experiment.

In 2020 Microsoft Research describes four stages where SRM can arise and an ExP reference threshold p < 0.0005 using a chi-squared test. The educational CIE reader below uses a two-sided exact binomial test for two 50/50 groups. It does not reproduce ExP’s algorithm; the historical threshold is illustrative, not a universal standard.

Microsoft Research · Diagnosing Sample Ratio Mismatch · Published 2020-09-14 · consulted 2026-10-23

Four places to investigate a counting discrepancy

Four places to investigate a counting discrepancy
Stage documented by ExPCIE review questionOriginal practical investigation
AssignmentCheck the unit and rule that assign each participant.Replay the same identity across visits; compare configuration versions and change times. A reassigned unit no longer represents two independent observations.
ExecutionCompare what each variant actually allows to execute.Reproduce redirects, incomplete loading, errors and cancellation. Record when exposure is logged and whether the variant changes that moment.
Log processingReconcile inputs, exclusions and joins.Count units before and after each transformation. Check missing keys, duplicates, bot filters and telemetry loss by group.
AnalysisExamine the population retained for comparison.Compare the assigned population with the analysed subset. A filter triggered by an action available only in B selects a population affected by treatment.

Define what the two counts represent

Define what the two counts represent
Reader assumptionConsequence for interpretation
Two mutually exclusive variantsCount each unit once, in A or B. Repeated page views or events are not distinct participants.
Fixed 50/50 allocationEvery unit has the same probability of assignment to A and B. A 90/10 allocation, changing ratio or enforced quota needs a different model.
Independent trialsUnits linked through an organisation, household or reassignment can violate the binomial model. Choose the unit in the protocol.
Fixed window and filtersCompare the same interval and deduplication and eligibility rules. Do not adapt exclusions to obtain a desired result.
Bounded aggregate countsAccept 0 to 5,000 units per arm, at most 10,000 overall. This is an interface limit, not a minimum statistical power requirement.
Quality check onlyThe reader computes neither effect, conversion, power nor required sample size. A result without an SRM signal does not establish causality.

NIST/SEMATECH · Binomial distribution · e-Handbook §1.3.6.6.18 · consulted 2026-10-23

Lector de reparto 50/50

Enter the two observed counts or load a fictional example. This checks allocation, not conversion counts. Calculation stays in the browser, without transmission, storage, import or export.

Under H₀, X follows Binomial(N, 0.5), with N = A + B and k = min(A, B). The reader computes p = min(1, 2 × Σᵢ₌₀ᵏ C(N,i) / 2ᴺ). This doubles the smaller tail, valid here by 50/50 symmetry. Sums are evaluated in logarithms to retain very small probabilities; displayed values are rounded.

A p-value describes how this allocation compares with the assumed model; it is not the probability that the experiment is valid or that a bug exists. A small value does not locate the cause. A large value does not establish error-free data. Repeated checks and multiple segments change false-alert control: define a suitable monitoring rule before operational use.

Eight educational examples readable without JavaScript

R01 · Two balanced groups
Two balanced groups
Count ACount BExpected reading
5050p = 1 under the model. This checks neither instrumentation, conversion nor causality.
R02 · 60/40 across ten units
60/40 across ten units
Count ACount BExpected reading
64p = 0.75390625. A visibly different ratio alone does not establish a signal at this small sample size.
R03 · Same ratio, hundredfold sample
Same ratio, hundredfold sample
Count ACount BExpected reading
600400The ratio remains 60/40, but the p-value becomes very small. Sample size changes statistical interpretation.
R04 · One empty arm, ten units
One empty arm, ten units
Count ACount BExpected reading
010p = 0.001953125 remains above the illustrative 0.0005 threshold. No alert at this threshold does not certify both arms are working.
R05 · One empty arm, twenty units
One empty arm, twenty units
Count ACount BExpected reading
020p = 0.0000019073486328125: a signal under the reference threshold. Investigate before interpreting an effect.
R06 · No observed units
No observed units
Count ACount BExpected reading
00No p-value: no observations. Show unavailable data rather than a reassuring result.
R07 · Extremely small probability
Extremely small probability
Count ACount BExpected reading
50001Displaying log₁₀(p) avoids presenting numerical rounding to zero as a mathematical impossibility.
R08 · Leading space is not repaired
Leading space is not repaired
Count ACount BExpected reading
5050Input rejected. The reader accepts only ASCII integers from 0 to 5,000, without spaces, signs, separators or redundant leading zeros.

Before resuming effect analysis

Before resuming effect analysis
Suggested checkRecord or result to retain
TimelineDiscrepancy onset, assignment versions, deployments and collection changes.
Stage countsAssigned, exposed and analysed counts; differences explained using the same unit.
IdentitiesEvidence of unit stability and explicit treatment of duplicates, session changes and sign-ins.
TelemetryComparable events, delivery errors and losses by variant, device and version.
FilteringEligibility and exclusion rules, with their position before or after treatment.
DecisionSupported cause, fix, still-usable period or new experiment, owner and review date.

Avoid artificially balancing groups by deleting observations to remove an alert. Reconstruct the data path first. An effect metric, guardrails and uncertainty remain necessary after repairing counts. In the final brief, separate the statistical signal, demonstrated cause and product decision.

Sources, editions and credits

Sources: Microsoft Research, article dated 14 September 2020; NIST/SEMATECH, binomial distribution §1.3.6.6.18. Questions, recipes and eight cases are CIE educational material, not results from real experiments. No institutional chart, logo or visual is reproduced.

Microsoft Research · Diagnosing Sample Ratio Mismatch · Published 2020-09-14 · consulted 2026-10-23

SHA-256 : d28a6a66b3f1d014f3aafce6c0239b6a6477049b2d4582f06f4d6aaae651ee75

NIST/SEMATECH · Binomial distribution · e-Handbook §1.3.6.6.18 · consulted 2026-10-23

SHA-256 : d899609cde11a15ebd745f6313931da10e9511a0dff64f39b027e5b9ad103ad8

Referencias oficiales

Las referencias respaldan el método. Adapte los controles a su contexto; no constituyen certificación. Los títulos de referencia originales y los documentos fuente pueden estar en otro idioma.

Referencias consultadas el .