How a marketing services company automated 50,000 monthly responses and reduced operating costs

Industry

Marketing

Company type

Marketing services company with more than 1,000 available service options

Business area

Lead response and customer service

Automation type

Automated first and second responses to sales inquiries

Syntropic Ops services

Process design, automation, systems integration, exception handling, and monitoring

Tools

Freshdesk, n8n, OpenAI, and Slack

Approximate volume

50,000 automated responses per month

Key result

Responses delivered in under 30 minutes and minimum estimated savings of $10,000 per month

The company receives a large volume of emails from potential customers requesting information about its marketing services.

The primary challenge was the size of its catalog. With more than 1,000 possible service options, each inquiry could relate to a different offering and required the team to accurately identify what information the lead needed.

Before the automation was implemented, these initial interactions were handled by a team of approximately 10 to 20 agents working across different schedules.

Each agent was assigned to specific areas of the company. As a result, response speed depended on the availability of the person responsible for that category.

When an email arrived outside an agent’s working hours, the inquiry could remain unanswered until that person became available again.

The growing volume of messages also created a scalability problem. To maintain response capacity, the company had to hire more employees and train them on an extensive catalog of services.

In addition to staffing costs, the company incurred expenses related to training, supervision, and shift coverage.

Experienced agents were also spending a significant portion of their time on repetitive, early-stage interactions instead of focusing on more advanced conversations, exceptional cases, or later stages of the customer journey.

Before the automation, the process worked as follows:

1. A lead sent an email requesting information.
2. The message was registered in Freshdesk.
3. The email was assigned to or reviewed by an agent responsible for that area.
4. The agent interpreted which service the customer was requesting.
5. The agent manually searched for the relevant information.
6. The agent prepared and sent a response.
7. If the assigned agent was unavailable, the email remained pending.
8. Follow-up interactions also depended on the team’s availability.

The process worked, but it required a large operational structure to handle an increasing volume of inquiries.

The company needed a way to manage the first stages of the conversation without increasing the number of agents at the same rate as lead volume.

The main objective was to reduce the manual work involved in sending the first and second responses to leads.

The company wanted to:

• Reduce the need to hire additional agents.
• Lower training costs.
• Respond faster.
• Maintain coverage regardless of each specialist’s working hours.
• Free trained agents from repetitive interactions.
• Allow the team to focus on later stages of the customer request.
• Scale response capacity without increasing headcount proportionally.
• Preserve human intervention for cases that genuinely required it.

The goal was not to eliminate agents entirely, but to use their time more effectively.

Syntropic Ops designed an automation that analyzes every new email received in Freshdesk.

When an inquiry arrives, Freshdesk sends the email data to n8n through a webhook. From that point, the message passes through several filters and validation steps.

The first goal of the system is to determine whether the email contains an information request that can be answered automatically.

The automation analyzes words, phrases, and patterns associated with the company’s services. These rules help identify which type of service the customer is requesting.

When the message is ambiguous or the rules do not produce a sufficiently clear result, OpenAI is used as an additional interpretation layer.

Once the service has been identified, the system automatically searches the company’s information source for the relevant content.

If the appropriate service is found, the system prepares a response and sends it through Freshdesk. This keeps the conversation recorded within the same platform used by the support team.

The system can manage the initial lead interactions without requiring an agent to review and answer every message manually.

When the automation cannot find the requested service or cannot determine the correct response with sufficient confidence, it does not force an answer.

Instead, it sends a Slack notification so an agent can review the inquiry and respond as quickly as possible.

This approach automates repetitive cases while preserving human review for exceptions.

1. 01

   Email Received

   A lead sends an inquiry requesting information about one of the company’s services. The email is registered in Freshdesk.

2. 02

   Webhook Trigger

   Freshdesk sends the new message data to n8n through a webhook. This event automatically starts the workflow.

3. 03

   Initial Validation

   The automation reviews the message content and determines whether it is an inquiry that can be processed automatically. It also filters out messages that should not receive an automated response.

4. 04

   Service Identification

   The system analyzes phrases, keywords, and patterns to identify which of the company’s more than 1,000 service options the customer is requesting.

5. 05

   AI-Assisted Interpretation

   When the message is ambiguous, OpenAI helps interpret the content and determine the intent behind the inquiry. Artificial intelligence acts as a supporting layer rather than the only decision-making mechanism.

6. 06

   Information Retrieval

   Once the service is identified, n8n automatically searches the company’s database or knowledge source for the relevant information.

7. 07

   Response Preparation

   The system uses the retrieved information to prepare the response that will be sent to the lead.

8. 08

   Delivery Through Freshdesk

   The response is sent through Freshdesk, preserving the conversation history in the original ticket.

9. 09

   Exception Handling

   If the service cannot be found or the message cannot be classified correctly, the system sends a Slack notification. An agent can then review and answer the case.

10. 10

    Logging and Control

    Each operation maintains a status that indicates whether the response was processed, sent, or escalated for human review.

The solution was to combine several methods:

• Trigger phrases.
• Patterns extracted from real emails.
• Business rules.
• Validation steps.
• OpenAI-assisted interpretation.
• Human review for ambiguous cases.

This combination created a more controlled and reliable system than one based exclusively on artificial intelligence.

The automation now processes approximately 50,000 responses per month.

Confirmed results include:

• Responses sent in under 30 minutes.
• An estimated 50% reduction in the staffing required to manage these interactions.
• Less dependence on individual agent schedules.
• Greater agent availability for complex cases.
• Reduced need to hire and train additional employees.
• More consistent coverage across different schedules.
• Automatic escalation of exceptional cases.
• Fewer repetitive tasks.
• Greater capacity to process a high volume of leads.

The company estimates that each automated response avoids between $0.20 and $0.50 in operational labor costs.

With approximately 50,000 responses per month, the calculation is as follows:

This places the estimated operational savings between $10,000 and $25,000 per month.

The minimum annual estimate is $10,000 × 12 months = $120,000 per year. Under the higher estimate, savings could reach $25,000 × 12 months = $300,000 per year.

These figures are operational estimates based on response volume and avoided labor costs. They should not be presented as audited financial results.

The impact of the automation goes beyond the cost of answering emails.

Before the project, an increase in inquiries required the company to expand its team, organize additional schedules, and train agents on a catalog of more than 1,000 services.

This model directly linked business growth to higher operating costs.

With the automation in place, a significant portion of the initial interactions can be processed without increasing the number of agents proportionally.

This allows the existing team to handle a greater volume of leads.

Trained agents can now spend more time on:

• Advanced inquiries.
• Exceptional cases.
• Specific customer needs.
• Later stages of the customer request.
• Conversations requiring human judgment.
• Improving the customer experience at higher-value touchpoints.

The company also reduced its dependence on individual specialists.

A lead no longer has to wait for the agent assigned to a specific category to become available. The system can process the request and send a response in under 30 minutes.

This improves response speed and reduces the risk of losing opportunities because of delays.

The process is also more scalable. The company can absorb increases in volume without immediately hiring additional agents to answer the same repetitive questions.

To operate at a high email volume, the automation required more than a simple connection between applications.

The system includes several reliability mechanisms:

• A queue that controls the frequency of calls to Freshdesk.
• Processing intervals aligned with API limits.
• Automatic retries for temporary failures.
• Status checks before repeating an operation.
• Duplicate-response prevention.
• Filters that determine which messages should be automated.
• Rules that complement AI-based interpretation.
• Slack alerts for unresolved cases.
• Human intervention for exceptional situations.
• Status tracking for each process.

These controls allow the automation to operate consistently without requiring constant manual supervision.

Automating the first and second responses to leads involved much more than connecting Freshdesk to an artificial intelligence model.

The result depended on designing a complete system capable of classifying messages, identifying services, retrieving information, managing API limits, executing safe retries, and escalating exceptions to the team.

The automation enabled the company to process approximately 50,000 responses per month, reduce the staffing required for these interactions by about 50%, and maintain response times below 30 minutes.

It also freed trained agents to focus on later stages of the process and on conversations that genuinely require human involvement.

With minimum estimated operational savings of $10,000 per month, the project shows how a well-designed automation can reduce costs, increase operational capacity, and improve response speed without losing control over exceptional cases.