Dataprev Campaign - Technical Feasibility Analysis

Executive Summary

This document analyzes the technical feasibility of executing a 30 million WhatsApp message campaign using Salesforce Agentforce for Dataprev, targeting Brazilian citizens. The analysis evaluates three delivery timelines (1, 3, and 5 days) against confirmed infrastructure limits, with sensitivity analysis for CTR variations from 2% to 10%.

Key Conclusion: A 5-day delivery window with CTR up to 5% is viable. At baseline 3% CTR, 5 days provides optimal safety margins. CTR above 5% requires extended timelines or additional infrastructure capacity.

Key Findings

Baseline (3% CTR): 5-day scenario is optimal with 17% safety margin
Conservative (2% CTR): Even 3-day scenario becomes viable (1,750 RPM < 2,400 limit)
High engagement (5% CTR): 5 days required (2,625 RPM approaches limit)
Best case (10% CTR): Exceeds infrastructure capacity even at 5 days - requires 8+ day timeline
Infrastructure bottleneck: DE/UACP throughput (2,400 RPM) is the primary constraint across all scenarios

1. Campaign Overview

WhatsApp Messages Sent

30M

Expected CTR (Baseline)

Total Conversations (3% CTR)

900K

Peak Concentration

70%

Peak Window

4h/day

1.1 Campaign Model

The campaign follows a standard conversational AI pattern where the system initiates contact via WhatsApp, and users respond to engage with the Agentforce agent. Based on historical data from similar government campaigns (MEC PROUNI/SISU - Jan/Feb 2026), we apply the following behavioral model:

Key Assumptions (Data-Driven)

Click-Through Rate (CTR): Baseline 3%, with sensitivity analysis from 2%-10%
Historical CTR range: 1.94%-4.7% (MEC campaigns)
Messages per conversation: 3 user messages on average
Conversation duration: 9.38 minutes average
Peak concentration: 70% of daily traffic occurs during 4-hour peak window (4 PM - 8 PM)
Traffic distribution: Bell curve distribution within peak hours

2. Infrastructure Limits

The following limits have been confirmed by Salesforce Product and Engineering teams for the Dataprev account:

Layer	Component	Limit	Notes
Layer 1	Meta WhatsApp	2.5M msg/hour (~694 msg/sec)	Rarely a bottleneck; Unlimited tier assumed
Layer 2	LLM Gateway	7,500 RPM	One-time uplift for Dataprev
Layer 3	DE/UACP Throughput	40 req/sec (2,400 RPM)	PRIMARY BOTTLENECK Includes both new sessions + existing session messages
Layer 3	DE/UACP Sessions	30,000 concurrent	System queues gracefully above this limit

2.1 Critical Understanding: DE/UACP Throughput

The 40 req/sec (2,400 RPM) limit applies to all inbound messages, regardless of whether they initiate a new session or continue an existing conversation. This is a critical constraint confirmed by the engineering team:

"DE inbound, we will double it up, i.e., from 20/s to 40/s. This change allows us to fully utilize 25k/15m for new conversations, and for any already initiated conversations to go through as well (up to 40/s flow)."
- Vrajesh Pothiwala, Salesforce Engineering

3. Calculation Methodology

3.1 Total Campaign Volume (Variable by CTR)

Total Conversations = WhatsApp Messages × CTR
Total User Messages = Total Conversations × 3 messages per conversation

Examples:
• CTR 2%: 30M × 2% = 600K conversations × 3 = 1.8M user messages
• CTR 3%: 30M × 3% = 900K conversations × 3 = 2.7M user messages
• CTR 5%: 30M × 5% = 1.5M conversations × 3 = 4.5M user messages
• CTR 10%: 30M × 10% = 3M conversations × 3 = 9M user messages

3.2 Daily Peak Calculation

Conversations per Day = Total Conversations / Campaign Days
Peak Conversations = Conversations per Day × 70%
Peak User Messages = Peak Conversations × 3 messages

Average RPM = Peak User Messages / 240 minutes (4-hour peak)
Peak RPM = Average RPM × 1.3 (bell curve peak factor)

Concurrent Sessions = (Peak Conversations / 240 min) × 9 min avg duration

4. Baseline Scenario Analysis (3% CTR)

Scenario	Days	Convs/Day	Peak Convs (4h)	Avg RPM	Peak RPM	Concurrent Sessions
1 Day	1	900,000	630,000	7,875	~10,300	23,625
3 Days	3	300,000	210,000	2,625	~3,400	7,875
5 Days	5	180,000	126,000	1,575	~2,000	4,725

4.1 Limit Compliance Analysis (3% CTR)

Scenario	DE/UACP (2,400 RPM)	LLM Gateway (7,500 RPM)	Sessions (30K)	Overall Status
1 Day	❌ FAILS 10,300 RPM (429% of limit)	❌ FAILS 10,300 RPM (137% of limit)	✅ OK 23,625 (79% of limit)	NOT VIABLE
3 Days	❌ FAILS 3,400 RPM (142% of limit)	✅ OK 3,400 RPM (45% of limit)	✅ OK 7,875 (26% of limit)	RISKY
5 Days	✅ OK 2,000 RPM (83% of limit)	✅ OK 2,000 RPM (27% of limit)	✅ OK 4,725 (16% of limit)	VIABLE

5. CTR Sensitivity Analysis

The following analysis demonstrates how different CTR outcomes impact infrastructure requirements. This is critical for understanding the robustness of each timeline scenario and identifying breaking points.

5.1 Peak RPM by CTR and Timeline

Timeline	CTR 2%	CTR 3% (Baseline)	CTR 5%	CTR 10%
1 Day	6,860 RPM ❌ 286%	10,300 RPM ❌ 429%	17,150 RPM ❌ 714%	34,300 RPM ❌ 1,429%
3 Days	2,290 RPM ✅ 95%	3,400 RPM ❌ 142%	5,700 RPM ❌ 238%	11,400 RPM ❌ 475%
5 Days	1,370 RPM ✅ 57%	2,000 RPM ✅ 83%	3,420 RPM ❌ 143%	6,840 RPM ❌ 285%
8 Days	860 RPM ✅ 36%	1,290 RPM ✅ 54%	2,140 RPM ✅ 89%	4,280 RPM ❌ 178%
10 Days	690 RPM ✅ 29%	1,030 RPM ✅ 43%	1,710 RPM ✅ 71%	3,420 RPM ❌ 143%

Note: Percentages indicate peak RPM as % of 2,400 RPM limit. Values shown include ~30% peak amplification factor from bell curve distribution.

5.2 Key Insights from Sensitivity Analysis

            Decision Matrix by CTR Outcome
            CTR ≤ 2% (Conservative): 3-day timeline becomes viable with 95% of limit utilization
CTR = 3% (Expected/Baseline): 5-day timeline recommended with 17% safety margin
CTR = 5% (High Engagement): 5-day timeline at edge of capacity (143% of limit) - requires 8+ days for safety
CTR = 10% (Best Case): Requires 10+ day timeline even at best case - exceeds infrastructure at shorter timelines

        

5.3 Concurrent Sessions by CTR

Timeline	CTR 2%	CTR 3%	CTR 5%	CTR 10%	Session Limit
1 Day	15,750	23,625	39,375	78,750	30,000
3 Days	5,250	7,875	13,125	26,250
5 Days	3,150	4,725	7,875	15,750
8 Days	1,969	2,953	4,922	9,844
10 Days	1,575	2,363	3,938	7,875

Observation: Concurrent sessions are NOT the bottleneck except in the extreme 1-day + 10% CTR scenario. DE/UACP throughput (RPM) is the primary constraint across all realistic scenarios.

6. Visual Analysis - Baseline (3% CTR)

6.1 RPM During 4-Hour Peak Window (3% CTR)

6.2 Concurrent Sessions During Peak Window (3% CTR)

6.3 Combined View - RPM + Sessions (3% CTR, Dual Axis)

7. CTR Sensitivity - Visual Comparison

7.1 Peak RPM by Timeline and CTR

7.2 5-Day Scenario - CTR Comparison

8. Risk Assessment

8.1 Baseline (3% CTR) Risk Analysis

1 Day - ⛔ NOT RECOMMENDED

Exceeds DE/UACP limit by 329% - system will throttle/queue heavily
Exceeds LLM Gateway limit by 37% - responses will be delayed
User experience will be severely degraded due to long wait times
Risk of cascading failures and system instability

3 Days - ⚠️ HIGH RISK

Exceeds DE/UACP limit by 42% during peak hours
System will queue messages, introducing latency
Minimal safety margin - any CTR variance causes failures
If CTR drops to 2%, becomes viable; if rises to 4%, fails badly

5 Days - ✅ RECOMMENDED (for CTR ≤ 3%)

Peak RPM stays below limit with 17% safety margin
Handles CTR variance up to ~3.6% before hitting limits
Provides buffer for unexpected traffic patterns
Ensures consistent, high-quality user experience

8.2 High CTR Scenarios (5% and 10%)

CTR 5% - Extended Timeline Required

At 5 days: Exceeds DE/UACP limit by 43% (3,420 RPM)

Recommendation: Extend to 8 days (2,140 RPM = 89% of limit) for safe execution

CTR 10% - Infrastructure Capacity Exceeded

At 5 days: Exceeds limit by 185% (6,840 RPM)

At 10 days: Still exceeds limit by 43% (3,420 RPM)

Recommendation: 10% CTR would require either:

12+ day timeline to stay within limits
Infrastructure uplift (increase DE/UACP capacity)
Message throttling strategy to spread load beyond peak hours

9. Recommendations

Primary Recommendation: 5-Day Timeline with 3% CTR Planning

Rationale:

Aligns with historical government campaign CTR data (1.94%-4.7%)
Provides 17% safety margin below infrastructure limits
Robust against CTR variance up to ~3.6%
Ensures high-quality, responsive user experience
Minimizes operational risk and emergency interventions

9.1 Contingency Planning

            Real-Time Monitoring Triggers
            If observed CTR < 2.5%: Consider accelerating timeline (potentially compress to 4 days)
If observed CTR = 3-3.5%: Proceed as planned with 5-day timeline
If observed CTR > 3.5%: Activate throttling protocol or extend timeline
If observed CTR > 5%: Immediate timeline extension to 7-8 days required

        

9.2 Alternative Scenarios

Observed CTR	Recommended Timeline	Peak RPM	Capacity Utilization
2%	3-5 days	1,370-2,290 RPM	57-95%
3%	5 days	2,000 RPM	83%
4%	5-7 days	2,000-2,730 RPM	83-114%
5%	8 days	2,140 RPM	89%
10%	12+ days or infra uplift	Exceeds at <10 days	-