Reducing churn with sentiment detection

Reducing churn with sentiment detection

Introduction

Introduction

Traffic Today is a performance marketing agency managing dozens of high-value client accounts across multiple platforms. Their growth brought complexity in managing client communications and performance data at scale.

Company name

Traffic Today

Year

2025

Company size

120

Industry

Online Marketing

Scope of work

/

Consultancy

/

AI Agents

/

AI Workflows

Timeline

12 weeks

Introduction

Traffic Today is a performance marketing agency managing dozens of high-value client accounts across multiple platforms. Their growth brought complexity in managing client communications and performance data at scale.

Company name

Traffic Today

Year

2025

Company size

120

Industry

Online Marketing

Scope of work

/

Consultancy

/

AI Agents

/

AI Workflows

Timeline

12 weeks

Challenges

Challenges

Challenges

As Traffic Today scaled, so did the complexity of managing client communications and performance data.

Account managers were juggling Slack messages, emails, VOIP calls, Asana task updates, and marketing analytics all scattered across multiple tools.

This made it difficult to:

  • Maintain real-time awareness of every client’s status

  • Spot urgent issues before they escalated

  • Provide managers with an accurate, high-level overview of all clients without hours of manual digging

The lack of unified client intelligence meant that high-intent messages could be buried, daily trends went unnoticed, and client context was fragmented across tools.

Approach

Approach

Approach

We looked at ways to combine all client data into positive directions.

Step 1 – Discovery & Data Mapping

We mapped every communication and data source used in daily operations. This included Slack, email, VOIP calls, task tracking tools, and marketing analytics platforms like GA4, Google Ads, and Search Console.

Step 2 – System Architecture & Ingestion Pipelines

We designed Nessy, a per-client AI agent embedded directly into Slack. Each agent has its own secure, isolated memory to store communications, analytics data, and contextual metadata.

Step 3 – Real-time Intelligence & Daily Updates

Nessy continuously ingests data streams and updates itself via a daily cron job. Managers and staff can query Nessy in real-time for client summaries, past decisions, urgent tasks, or trend spotting.

Step 4 – AI Layer with Human Feedback

We implemented a “mark as important” feature so humans can flag high-value communications. This allows the AI to prioritise and focus on key actions, ensuring it captures sentiment, urgency, and intent accurately.

Solutions

Solutions

Solutions

We deployed Nessy, a Slack-native Client Intelligence Agent, for 100+ active clients.

Nessy connects all communication and analytics sources into one searchable, conversational interface.

It automatically:

  • Captures importance ratings, creates sentiment scores, urgency levels, and topics discussed

  • Summarises client activity daily

  • Detects and highlights key actions, questions, and decisions

  • Performs cross-client trend analysis for management visibility

Current integrations include Slack, Email (via CC), and VOIP call data. We combined integrations with Asana, Google Sheets reporting, WhatsApp, and additional marketing data sources (Meta, TikTok, Ahrefs).

The strong thing about this use case is that we made it scalable to work on Slack, Teams, or whatever chat you prefer, creating a flexible layer.

You now have the ability to also ask @Nessy questions on the actual date, so it can reply and report back on the data it gathers.

Results

Results

Results

Nessy has eliminated the need for managers to manually check multiple tools just to understand client progress.

Account managers now receive daily briefings with the most important updates for each client. Managers gain instant, cross-client performance insights without waiting for weekly reports. Urgent issues are spotted within minutes instead of days.

Over the first month, Nessy processed an average of ~1,200 messages per client (≈ 49,000 total messages) across all integrated channels, automatically classifying and summarising them for context and action.

Key Results

60

60

minutes/week saved per manager per client

100%

100%

of relevant comms automatically summarised

100+

100+

Onboarded client channels on Nessy

Have a project in mind?

Talk to our team

Tell us about your project—whether it’s an AI Agent, Workflow, or Project.

Have a project in mind?

Talk to our team

Tell us about your project—whether it’s an AI Agent, Workflow, or Project.

Have a project in mind?

Talk to our team

Tell us about your project—whether it’s an AI Agent, Workflow, or Project.