How to Diagnose Amazon Connect Outbound Campaign Throttling Using CloudWatch and CloudTrail

Introduction

When Amazon Connect outbound campaigns experience longer-than-expected connect times or silently fail to dial contacts, the root cause is often throttling at the campaign level rather than a flow logic issue. Standard troubleshooting focuses on contact flows and agent availability, but campaign-level rate limits operate beneath that layer and produce symptoms that are easy to misattribute.

This article provides a step-by-step diagnostic methodology to identify which campaign quota is being hit, quantify the impact, and resolve it.

---

Understanding Campaign Throttle Types

Amazon Connect enforces two distinct campaign-level quotas that interact with each other:

1. Concurrent campaign active calls per instance

The maximum number of simultaneous active calls that campaigns can sustain. When this limit is hit, new campaign dial attempts receive a fast busy tone. This quota is shared across all campaigns running on the instance.

2. Campaign calls per second (CPS) rate

The maximum number of new outbound calls that campaigns can initiate per second. When this limit is hit, excess dial requests are throttled and not placed. This quota is also shared across all campaigns on the instance.

These are separate from the instance-level concurrent calls quota. You can have plenty of instance-level capacity but still hit campaign-specific limits.

💡 Key insight: If you run multiple campaigns simultaneously, they collectively consume the shared CPS allocation. Five campaigns running at once do not each get the full CPS limit — they share it.

---

Symptoms

Look for these indicators that campaign-level throttling may be occurring:

• ⏱️ Outbound campaign connect time exceeds expected values (for example, 4-5 seconds instead of under 2 seconds)
• 📞 Contacts are dialed but receive fast busy tones
• 📊 Campaign throughput plateaus despite available agents
• ❌ DisconnectReason shows connection failures not attributable to the destination number
• 📉 Campaign dials per minute metric is lower than expected given available agent capacity

---

Step 1: Check CloudWatch Metrics

Open the CloudWatch console and navigate to Metrics → All metrics → AWS/Connect. Select your instance ID and graph the following metrics over a 24-hour period with 1-minute granularity.

Primary indicators

Metric	What it reveals
ThrottledCalls	Direct count of calls rejected due to campaign rate limiting. If this is nonzero, you are hitting a campaign quota.
CallsBreachingConcurrencyQuota	Calls rejected because the concurrent limit was reached.
ConcurrentCalls	Current active calls. Compare peak to your campaign active calls quota.
ConcurrentCallsPercentage	How close you are to your configured concurrent call limit.
CallsPerInterval	Total calls (inbound + outbound) per second. Reveals if inbound traffic consumes campaign capacity.

Pattern recognition

---

🔴 Pattern A — CPS throttling

ThrottledCalls is consistently nonzero throughout campaign operating hours, but ConcurrentCalls never reaches your concurrent quota.

→ The per-second dial rate is the binding constraint, not total active calls.

---

🟠 Pattern B — Concurrency throttling

CallsBreachingConcurrencyQuota spikes correlate with ConcurrentCalls peaks touching or exceeding your campaign active calls limit.

→ Calls are failing because too many are active simultaneously.

---

🟡 Pattern C — Mixed

Both metrics are nonzero but at different times. CPS limits during ramp-up (morning), concurrency limits during steady-state (midday peak).

→ You may need to address both.

---

CloudWatch Metrics Insights query

Quantify daily throttled calls:

SELECT AVG(ThrottledCalls) as avg_throttled,
       SUM(ThrottledCalls) as total_throttled,
       MAX(ConcurrentCalls) as peak_concurrent
FROM SCHEMA("AWS/Connect", InstanceId)
WHERE InstanceId = 'YOUR_INSTANCE_ID'
GROUP BY BIN(1d)
ORDER BY BIN(1d) DESC

---

Step 2: Correlate with CloudTrail Events

CloudTrail captures the specific API-level rejections that CloudWatch aggregates. This gives you timestamps, error codes, and request context.

Create an Athena table for CloudTrail logs

If you do not already have CloudTrail logs queryable in Athena, create a table:

CREATE EXTERNAL TABLE cloudtrail_logs (
    eventVersion STRING,
    eventTime STRING,
    eventSource STRING,
    eventName STRING,
    awsRegion STRING,
    sourceIPAddress STRING,
    userAgent STRING,
    errorCode STRING,
    errorMessage STRING,
    requestParameters STRING,
    responseElements STRING,
    userIdentity STRUCT<
        type: STRING,
        principalId: STRING,
        arn: STRING,
        accountId: STRING
    >
)
PARTITIONED BY (
    account STRING,
    region STRING,
    year STRING,
    month STRING,
    day STRING
)
ROW FORMAT SERDE 'org.apache.hive.hcatalog.data.JsonSerDe'
LOCATION 's3://YOUR_CLOUDTRAIL_BUCKET/AWSLogs/'

Query for campaign throttle events

SELECT
    date_trunc('hour', from_iso8601_timestamp(eventTime)) AS hour,
    errorCode,
    count(*) AS throttle_count,
    json_extract_scalar(requestParameters, '$.instanceId') AS instance_id
FROM cloudtrail_logs
WHERE eventSource = 'connect.amazonaws.com'
  AND eventName = 'StartOutboundVoiceContact'
  AND errorCode = 'LimitExceededException'
  AND from_iso8601_timestamp(eventTime) >= current_timestamp - interval '7' day
GROUP BY 1, 2, 3
ORDER BY 1 DESC

What to look for

Signal	Meaning
Throttle events concentrated in specific hours	Campaign schedule overlap is causing contention
Error code is LimitExceededException	Confirms a hard quota ceiling, not a transient error
High ratio of throttled to total StartOutboundVoiceContact calls	Significant portion of dial attempts are being dropped

---

Step 3: Identify the Binding Constraint

Overlay your CloudWatch and CloudTrail findings to determine the root cause:

Observation	Binding constraint	Recommended action
ThrottledCalls high, ConcurrentCalls well below quota	CPS rate limit	Reduce concurrent campaigns or request CPS increase
CallsBreachingConcurrencyQuota high, ConcurrentCalls at quota	Concurrent active calls limit	Request concurrent campaign calls increase
Both metrics elevated simultaneously	Both limits	Address CPS first (upstream gate), then reassess concurrency
CloudTrail LimitExceededException but CW metrics are zero	API-level throttle	Check Outbound Campaigns API rate limits (PutDialRequestBatch: 10 TPS)

---

Step 4: Remediate

⚡ Immediate fixes (no quota change required)

Reduce per-campaign bandwidth allocation

Each campaign's dialing configuration includes a BandwidthAllocation setting (0.0 to 1.0). Reducing this from the default spreads the shared CPS budget more evenly across campaigns and prevents any single campaign from starving others.

Stagger campaign start times

If campaigns all start simultaneously (for example, 8:00 AM), they ramp together and spike CPS usage. Offset start times by 10-15 minutes to smooth the demand curve.

Reduce concurrent campaign count

If you run more campaigns simultaneously than your CPS allocation can support, consolidate contact lists into fewer campaigns during peak hours.

📋 Strategic fix (requires quota increase)

To request a campaign quota increase (requires Business, Enterprise On-Ramp, or Enterprise Support plan):

1. Open the AWS Support Center and create a service limit increase case
2. Select Amazon Connect as the service
3. Specify the instance ARN and production account ID
4. Include the following data points:

Data point	Where to find it
Current daily outbound call volume	Connect admin console or CallsPerInterval metric
Number of concurrent campaigns	Campaign management page
Campaign operating hours per day	Campaign schedule configuration
Peak concurrent calls observed	CloudWatch ConcurrentCalls (Maximum statistic)
Daily throttle volume	CloudWatch ThrottledCalls (Sum statistic)

⚠️ Note: Campaign-specific quotas (CPS rate, burst capacity, active campaign calls) may not appear in the Service Quotas console. Contact AWS Support to confirm your current values and request increases.

---

Step 5: Verify the Fix

After implementing changes, monitor the same metrics for a full business week:

SELECT SUM(ThrottledCalls) as daily_throttled,
       MAX(ConcurrentCalls) as peak_concurrent,
       SUM(CallsBreachingConcurrencyQuota) as daily_breaches
FROM SCHEMA("AWS/Connect", InstanceId)
WHERE InstanceId = 'YOUR_INSTANCE_ID'
GROUP BY BIN(1d)
ORDER BY BIN(1d) DESC
LIMIT 14

Success criteria

Compare the 7 days before your change to the 7 days after:

• ✅ ThrottledCalls daily sum reduced by 70% or more
• ✅ CallsBreachingConcurrencyQuota near zero
• ✅ Campaign connect time returning to expected values (check via outbound campaign metrics in the Connect admin console)

---

Quick Reference

Step	Tool	Purpose
1	CloudWatch Metrics	Quantify throttling and identify which quota type
2	CloudTrail + Athena	Get precise timestamps and confirm error codes
3	Correlation	Determine binding constraint (CPS vs. concurrency)
4	Console + Support	Apply immediate fixes and request quota increases
5	CloudWatch Metrics	Validate resolution

---

Key Takeaway

Campaign-level throttling operates below the flow layer. Standard Connect troubleshooting — reviewing flows, checking agent availability, examining disconnect reasons — will not surface campaign CPS exhaustion. You must check ThrottledCalls and CallsBreachingConcurrencyQuota explicitly, then correlate with CloudTrail to confirm which specific limit is being hit.

---

Related Resources

• Monitoring your Amazon Connect instance using CloudWatch
• Amazon Connect service quotas
• Outbound campaign metrics
• Troubleshoot outbound campaign issues in Amazon Connect
• Best practices for Amazon Connect outbound campaigns