diff --git a/src/KArtSell.DbMigrator/0044_consumer_error_handling_and_metrics.sql b/src/KArtSell.DbMigrator/0044_consumer_error_handling_and_metrics.sql
new file mode 100644
index 00000000..915fe053
--- /dev/null
+++ b/src/KArtSell.DbMigrator/0044_consumer_error_handling_and_metrics.sql
@@ -0,0 +1,106 @@
+-- Migration 0044: Consumer Error Handling & Monitoring Infrastructure
+-- AEG-VS-01-05: Event/Job/Inbox - Part 4 Stage 3 (Error Handling + Monitoring)
+-- Created: 2026-08-17
+-- Purpose: Dead-letter queue for failed messages, metrics for observability
+
+BEGIN;
+
+-- 1. DEAD LETTER MESSAGE TABLE
+-- Captures consumer errors: logs failed messages, retry attempts, last error details
+CREATE TABLE IF NOT EXISTS building_blocks.dead_letter_message (
+    dead_letter_id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
+    message_id UUID NOT NULL,
+    event_type VARCHAR(100) NOT NULL,
+    payload_json JSONB NOT NULL,
+    correlation_id UUID,
+    error_message TEXT NOT NULL,
+    error_stacktrace TEXT,
+    attempt_number INT NOT NULL DEFAULT 1,
+    status VARCHAR(50) NOT NULL DEFAULT 'RETRYING'
+        CHECK (status IN ('RETRYING', 'FAILED', 'ARCHIVED')),
+    last_error_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    created_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT CURRENT_TIMESTAMP,
+
+    -- Composite unique: prevent duplicate error records for same message+attempt
+    UNIQUE(message_id, attempt_number)
+);
+
+CREATE INDEX IF NOT EXISTS idx_dead_letter_message_id ON building_blocks.dead_letter_message(message_id);
+CREATE INDEX IF NOT EXISTS idx_dead_letter_status ON building_blocks.dead_letter_message(status);
+CREATE INDEX IF NOT EXISTS idx_dead_letter_created_at ON building_blocks.dead_letter_message(created_at DESC);
+CREATE INDEX IF NOT EXISTS idx_dead_letter_correlation ON building_blocks.dead_letter_message(correlation_id);
+
+COMMENT ON TABLE building_blocks.dead_letter_message IS
+    'Dead-letter queue for consumer errors. Captures failed messages, errors, retry attempts.';
+
+COMMENT ON COLUMN building_blocks.dead_letter_message.status IS
+    'RETRYING = will retry later, FAILED = exhausted retries, ARCHIVED = moved to cold storage';
+
+COMMENT ON COLUMN building_blocks.dead_letter_message.attempt_number IS
+    'Retry attempt counter. Max retries = 3. After 3 failures, status = FAILED.';
+
+-- Update inbox schema to track failed messages
+ALTER TABLE building_blocks.inbox_message
+ADD COLUMN IF NOT EXISTS status VARCHAR(50) DEFAULT 'PENDING'
+    CHECK (status IN ('PENDING', 'PROCESSING', 'COMPLETED', 'FAILED')),
+ADD COLUMN IF NOT EXISTS failed_at TIMESTAMP WITH TIME ZONE;
+
+CREATE INDEX IF NOT EXISTS idx_inbox_status ON building_blocks.inbox_message(status)
+WHERE status = 'FAILED';
+
+-- 2. CONSUMER METRICS TABLE
+-- Performance metrics: latency, success/failure rates, per consumer per event type
+CREATE TABLE IF NOT EXISTS infrastructure.consumer_metrics (
+    metric_id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
+    consumer_type VARCHAR(100) NOT NULL,
+    event_type VARCHAR(100) NOT NULL,
+    correlation_id UUID,
+    duration_ms BIGINT NOT NULL,
+    success BOOLEAN NOT NULL DEFAULT true,
+    error_message TEXT,
+    recorded_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT CURRENT_TIMESTAMP
+);
+
+CREATE INDEX IF NOT EXISTS idx_consumer_metrics_consumer_type ON infrastructure.consumer_metrics(consumer_type);
+CREATE INDEX IF NOT EXISTS idx_consumer_metrics_event_type ON infrastructure.consumer_metrics(event_type);
+CREATE INDEX IF NOT EXISTS idx_consumer_metrics_recorded_at ON infrastructure.consumer_metrics(recorded_at DESC);
+CREATE INDEX IF NOT EXISTS idx_consumer_metrics_correlation ON infrastructure.consumer_metrics(correlation_id);
+
+-- Partition by month for efficient retention policies
+CREATE TABLE IF NOT EXISTS infrastructure.consumer_metrics_202608 PARTITION OF infrastructure.consumer_metrics
+FOR VALUES FROM ('2026-08-01') TO ('2026-09-01');
+
+COMMENT ON TABLE infrastructure.consumer_metrics IS
+    'Consumer performance metrics: latency (duration_ms), success rate, error tracking. ' ||
+    'Partitioned by month for efficient querying and retention. Used for dashboards and alerting.';
+
+COMMENT ON COLUMN infrastructure.consumer_metrics.duration_ms IS
+    'Time to execute consumer handler. Includes serialization, network calls, DB writes. Used for SLA monitoring.';
+
+-- 3. CONSUMER ALERT THRESHOLDS
+-- Define alert conditions for degradation (high latency, low success rate)
+CREATE TABLE IF NOT EXISTS infrastructure.consumer_alert_rules (
+    rule_id UUID PRIMARY KEY DEFAULT gen_random_uuid(),
+    consumer_type VARCHAR(100) NOT NULL UNIQUE,
+    p95_latency_ms BIGINT NOT NULL DEFAULT 1000,  -- Alert if p95 > 1s
+    min_success_rate DECIMAL(5, 2) NOT NULL DEFAULT 95.0,  -- Alert if success rate < 95%
+    enabled BOOLEAN NOT NULL DEFAULT true,
+    created_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT CURRENT_TIMESTAMP,
+    updated_at TIMESTAMP WITH TIME ZONE NOT NULL DEFAULT CURRENT_TIMESTAMP
+);
+
+CREATE INDEX IF NOT EXISTS idx_consumer_alert_rules_enabled ON infrastructure.consumer_alert_rules(enabled)
+WHERE enabled = true;
+
+COMMENT ON TABLE infrastructure.consumer_alert_rules IS
+    'Alert thresholds per consumer type. Used to detect performance degradation and high error rates.';
+
+-- Insert default alert rules
+INSERT INTO infrastructure.consumer_alert_rules (consumer_type, p95_latency_ms, min_success_rate)
+VALUES
+    ('IdentityCreatedConsumer', 500, 99.0),
+    ('IdentityAuditConsumer', 1000, 99.0),
+    ('MfaReminderJob', 5000, 95.0)
+ON CONFLICT (consumer_type) DO NOTHING;
+
+COMMIT;
diff --git a/src/KArtSell.Host/Jobs/ConsumerErrorHandler.cs b/src/KArtSell.Host/Jobs/ConsumerErrorHandler.cs
new file mode 100644
index 00000000..05dc003a
--- /dev/null
+++ b/src/KArtSell.Host/Jobs/ConsumerErrorHandler.cs
@@ -0,0 +1,126 @@
+using Dapper;
+using KArtSell.BuildingBlocks.Data;
+using Microsoft.Extensions.Logging;
+using Npgsql;
+using System.Text.Json;
+
+namespace KArtSell.Host.Jobs;
+
+/// <summary>
+/// Handles consumer errors: logs to dead-letter queue, tracks failure metrics.
+/// Transactional: error record written atomically with inbox status update.
+/// Idempotent: message_id + attempt_number ensures no duplicate error records.
+/// </summary>
+public sealed class ConsumerErrorHandler(
+    IDbConnectionFactory connectionFactory,
+    ILogger<ConsumerErrorHandler> logger)
+{
+    private const int MaxRetryAttempts = 3;
+
+    public async Task HandleConsumerErrorAsync(
+        Guid messageId,
+        string eventType,
+        string payloadJson,
+        string correlationId,
+        Exception exception,
+        int attemptNumber,
+        CancellationToken cancellationToken = default)
+    {
+        var conn = await connectionFactory.OpenAsync(cancellationToken) as NpgsqlConnection
+            ?? throw new InvalidOperationException("Failed to open connection");
+
+        await using (conn)
+        {
+            await using var transaction = await conn.BeginTransactionAsync(cancellationToken: cancellationToken);
+
+            try
+            {
+                // Log error to dead-letter queue
+                const string deadLetterSql = """
+                    INSERT INTO building_blocks.dead_letter_message (
+                        message_id, event_type, payload_json, correlation_id,
+                        error_message, error_stacktrace, attempt_number,
+                        last_error_at, status
+                    )
+                    VALUES (
+                        @MessageId, @EventType, @PayloadJson, @CorrelationId,
+                        @ErrorMessage, @ErrorStackTrace, @AttemptNumber,
+                        NOW(), @Status
+                    )
+                    ON CONFLICT (message_id, attempt_number) DO UPDATE
+                    SET last_error_at = NOW(), error_message = @ErrorMessage
+                    """;
+
+                var status = attemptNumber >= MaxRetryAttempts ? "FAILED" : "RETRYING";
+
+                await conn.ExecuteAsync(
+                    deadLetterSql,
+                    new
+                    {
+                        messageId,
+                        eventType,
+                        payloadJson,
+                        correlationId,
+                        errorMessage = exception.Message,
+                        errorStackTrace = exception.StackTrace ?? string.Empty,
+                        attemptNumber,
+                        status
+                    });
+
+                // Update inbox status for failed messages
+                if (attemptNumber >= MaxRetryAttempts)
+                {
+                    const string updateInboxSql = """
+                        UPDATE building_blocks.inbox_message
+                        SET status = 'FAILED', failed_at = NOW()
+                        WHERE message_id = @MessageId
+                        """;
+
+                    await conn.ExecuteAsync(updateInboxSql, new { messageId });
+                }
+
+                await transaction.CommitAsync(cancellationToken);
+
+                LogConsumerErrorMessage(messageId, eventType, attemptNumber, status, exception);
+            }
+            catch (Exception deadLetterEx)
+            {
+                await transaction.RollbackAsync(cancellationToken);
+                logger.LogError(
+                    deadLetterEx,
+                    "CRITICAL: Failed to log dead-letter message {MessageId} (event type: {EventType}). " +
+                    "Original error: {OriginalError}",
+                    messageId, eventType, exception.Message);
+                throw;
+            }
+        }
+    }
+
+    public async Task<bool> IsMessageFailedAsync(Guid messageId, CancellationToken cancellationToken = default)
+    {
+        var conn = await connectionFactory.OpenAsync(cancellationToken) as NpgsqlConnection
+            ?? throw new InvalidOperationException("Failed to open connection");
+
+        await using (conn)
+        {
+            const string sql = """
+                SELECT COUNT(1) > 0
+                FROM building_blocks.dead_letter_message
+                WHERE message_id = @MessageId AND status = 'FAILED'
+                """;
+
+            return await conn.QuerySingleAsync<bool>(sql, new { messageId });
+        }
+    }
+
+    private static readonly Action<ILogger, Guid, string, int, string, Exception?> LogConsumerError =
+        LoggerMessage.Define<Guid, string, int, string>(
+            LogLevel.Error,
+            new EventId(1, nameof(LogConsumerError)),
+            "Consumer error for message {MessageId} (event: {EventType}, attempt: {AttemptNumber}). Status: {Status}");
+
+    private void LogConsumerErrorMessage(Guid messageId, string eventType, int attemptNumber, string status, Exception ex)
+    {
+        LogConsumerError(logger, messageId, eventType, attemptNumber, status, ex);
+    }
+}
diff --git a/src/KArtSell.Host/Jobs/ConsumerMetrics.cs b/src/KArtSell.Host/Jobs/ConsumerMetrics.cs
new file mode 100644
index 00000000..cd83ea6a
--- /dev/null
+++ b/src/KArtSell.Host/Jobs/ConsumerMetrics.cs
@@ -0,0 +1,188 @@
+using System.Diagnostics;
+using Dapper;
+using KArtSell.BuildingBlocks.Data;
+using Microsoft.Extensions.Logging;
+using Npgsql;
+
+namespace KArtSell.Host.Jobs;
+
+/// <summary>
+/// Tracks consumer performance metrics: latency, success/failure rates, throughput.
+/// Records per consumer type with timestamp bucketing (1-minute intervals).
+/// Used for observability dashboards and alerting on degradation.
+/// </summary>
+public sealed class ConsumerMetrics(
+    IDbConnectionFactory connectionFactory,
+    ILogger<ConsumerMetrics> logger)
+{
+    public sealed class ConsumerInvocation
+    {
+        public required string ConsumerType { get; init; }
+        public required string EventType { get; init; }
+        public required Stopwatch Stopwatch { get; init; }
+        public required string CorrelationId { get; init; }
+        public bool Success { get; set; }
+        public string? ErrorMessage { get; set; }
+    }
+
+    public ConsumerInvocation StartInvocation(string consumerType, string eventType, string correlationId)
+    {
+        return new ConsumerInvocation
+        {
+            ConsumerType = consumerType,
+            EventType = eventType,
+            Stopwatch = Stopwatch.StartNew(),
+            CorrelationId = correlationId
+        };
+    }
+
+    public async Task RecordInvocationAsync(
+        ConsumerInvocation invocation,
+        CancellationToken cancellationToken = default)
+    {
+        invocation.Stopwatch.Stop();
+        var durationMs = invocation.Stopwatch.ElapsedMilliseconds;
+
+        var conn = await connectionFactory.OpenAsync(cancellationToken) as NpgsqlConnection
+            ?? throw new InvalidOperationException("Failed to open connection");
+
+        await using (conn)
+        {
+            const string sql = """
+                INSERT INTO infrastructure.consumer_metrics (
+                    consumer_type, event_type, correlation_id,
+                    duration_ms, success, error_message, recorded_at
+                )
+                VALUES (
+                    @ConsumerType, @EventType, @CorrelationId,
+                    @DurationMs, @Success, @ErrorMessage, NOW()
+                )
+                """;
+
+            try
+            {
+                await conn.ExecuteAsync(sql, new
+                {
+                    invocation.ConsumerType,
+                    invocation.EventType,
+                    invocation.CorrelationId,
+                    durationMs,
+                    invocation.Success,
+                    invocation.ErrorMessage
+                });
+
+                if (invocation.Success)
+                {
+                    LogSuccess(logger, invocation.ConsumerType, invocation.EventType, durationMs, null);
+                }
+                else
+                {
+                    LogFailure(logger, invocation.ConsumerType, invocation.EventType, durationMs, invocation.ErrorMessage, null);
+                }
+            }
+            catch (Exception ex)
+            {
+                logger.LogError(
+                    ex,
+                    "Failed to record consumer metrics for {ConsumerType}({EventType}). Duration: {DurationMs}ms",
+                    invocation.ConsumerType, invocation.EventType, durationMs);
+            }
+        }
+    }
+
+    public async Task<ConsumerMetricsSnapshot> GetMetricsSnapshotAsync(
+        string consumerType,
+        int last_minutes = 5,
+        CancellationToken cancellationToken = default)
+    {
+        var conn = await connectionFactory.OpenAsync(cancellationToken) as NpgsqlConnection
+            ?? throw new InvalidOperationException("Failed to open connection");
+
+        await using (conn)
+        {
+            const string sql = """
+                SELECT
+                    COUNT(*) AS total_invocations,
+                    SUM(CASE WHEN success THEN 1 ELSE 0 END) AS successful_invocations,
+                    SUM(CASE WHEN NOT success THEN 1 ELSE 0 END) AS failed_invocations,
+                    AVG(duration_ms) AS avg_duration_ms,
+                    MAX(duration_ms) AS max_duration_ms,
+                    MIN(duration_ms) AS min_duration_ms,
+                    PERCENTILE_CONT(0.95) WITHIN GROUP (ORDER BY duration_ms) AS p95_duration_ms
+                FROM infrastructure.consumer_metrics
+                WHERE consumer_type = @ConsumerType
+                  AND recorded_at >= NOW() - INTERVAL '1 minute' * @LastMinutes
+                """;
+
+            var result = await conn.QuerySingleOrDefaultAsync<MetricsRow>(
+                sql,
+                new { consumerType, lastMinutes = last_minutes });
+
+            if (result == null)
+            {
+                return new ConsumerMetricsSnapshot
+                {
+                    ConsumerType = consumerType,
+                    TotalInvocations = 0,
+                    SuccessfulInvocations = 0,
+                    FailedInvocations = 0,
+                    SuccessRate = 0,
+                    AvgDurationMs = 0,
+                    MaxDurationMs = 0,
+                    P95DurationMs = 0
+                };
+            }
+
+            var successRate = result.TotalInvocations > 0
+                ? (double)result.SuccessfulInvocations / result.TotalInvocations * 100
+                : 0;
+
+            return new ConsumerMetricsSnapshot
+            {
+                ConsumerType = consumerType,
+                TotalInvocations = result.TotalInvocations,
+                SuccessfulInvocations = result.SuccessfulInvocations,
+                FailedInvocations = result.FailedInvocations,
+                SuccessRate = successRate,
+                AvgDurationMs = result.AvgDurationMs ?? 0,
+                MaxDurationMs = result.MaxDurationMs ?? 0,
+                P95DurationMs = result.P95DurationMs ?? 0
+            };
+        }
+    }
+
+    private sealed record MetricsRow
+    {
+        public int TotalInvocations { get; init; }
+        public int SuccessfulInvocations { get; init; }
+        public int FailedInvocations { get; init; }
+        public double? AvgDurationMs { get; init; }
+        public long? MaxDurationMs { get; init; }
+        public long? MinDurationMs { get; init; }
+        public double? P95DurationMs { get; init; }
+    }
+
+    private static readonly Action<ILogger, string, string, long, Exception?> LogSuccess =
+        LoggerMessage.Define<string, string, long>(
+            LogLevel.Debug,
+            new EventId(1, nameof(LogSuccess)),
+            "Consumer {ConsumerType} processed {EventType} successfully in {DurationMs}ms");
+
+    private static readonly Action<ILogger, string, string, long, string?, Exception?> LogFailure =
+        LoggerMessage.Define<string, string, long, string?>(
+            LogLevel.Error,
+            new EventId(2, nameof(LogFailure)),
+            "Consumer {ConsumerType} failed to process {EventType} after {DurationMs}ms. Error: {ErrorMessage}");
+}
+
+public sealed record ConsumerMetricsSnapshot
+{
+    public required string ConsumerType { get; init; }
+    public int TotalInvocations { get; init; }
+    public int SuccessfulInvocations { get; init; }
+    public int FailedInvocations { get; init; }
+    public double SuccessRate { get; init; }
+    public double AvgDurationMs { get; init; }
+    public long MaxDurationMs { get; init; }
+    public double P95DurationMs { get; init; }
+}
diff --git a/src/KArtSell.Host/Jobs/DownstreamConsumerJob.cs b/src/KArtSell.Host/Jobs/DownstreamConsumerJob.cs
index c8819150..a18bacd2 100644
--- a/src/KArtSell.Host/Jobs/DownstreamConsumerJob.cs
+++ b/src/KArtSell.Host/Jobs/DownstreamConsumerJob.cs
@@ -23,6 +23,7 @@ public sealed class DownstreamConsumerJob(
     IdentityCreatedConsumer identityCreatedConsumer,
     IdentityAuditConsumer identityAuditConsumer,
     MfaReminderJob mfaReminderJob,
+    ConsumerErrorHandler errorHandler,
     IClock clock,
     ILogger<DownstreamConsumerJob> logger)
 {
@@ -79,10 +80,13 @@ public sealed class DownstreamConsumerJob(
 
         foreach (var (messageId, hash) in pendingMessages)
         {
+            (string EventType, string PayloadJson) outboxRow = default;
+            string eventType = string.Empty;
+
             try
             {
                 // Fetch outbox message payload
-                var outboxRow = await connection.QuerySingleOrDefaultAsync<(string EventType, string PayloadJson)>(
+                outboxRow = await connection.QuerySingleOrDefaultAsync<(string EventType, string PayloadJson)>(
                     new CommandDefinition(
                         selectOutboxSql,
                         new { MessageId = messageId },
@@ -94,7 +98,8 @@ public sealed class DownstreamConsumerJob(
                     continue;
                 }
 
-                var (eventType, payloadJson) = outboxRow;
+                eventType = outboxRow.EventType;
+                var payloadJson = outboxRow.PayloadJson;
 
                 // Route to appropriate consumer based on event type
                 switch (eventType)
@@ -142,7 +147,30 @@ public sealed class DownstreamConsumerJob(
             }
             catch (Exception ex)
             {
-                logger.LogError(ex, "Failed to process inbox message {MessageId}", messageId);
+                logger.LogError(ex, "Failed to process inbox message {MessageId} (event: {EventType})", messageId, eventType);
+
+                // Log to dead-letter queue for alerting and debugging
+                try
+                {
+                    var payloadJson = outboxRow != default ? outboxRow.PayloadJson : string.Empty;
+                    var correlationId = outboxRow != default ? "tracing-available" : "unknown";
+                    await errorHandler.HandleConsumerErrorAsync(
+                        messageId,
+                        eventType,
+                        payloadJson ?? string.Empty,
+                        correlationId,
+                        ex,
+                        1,  // First attempt
+                        cancellationToken);
+                }
+                catch (Exception deadLetterEx)
+                {
+                    logger.LogCritical(
+                        deadLetterEx,
+                        "CRITICAL: Failed to log dead-letter message {MessageId}. Original error: {OriginalError}",
+                        messageId, ex.Message);
+                }
+
                 throw;
             }
         }