Commercial Intelligence Platform for Sales Teams via WhatsApp
Friday, October 23, 2026
This applied project report presents a commercial intelligence platform designed to help sales teams and managers work with negotiation data captured in WhatsApp conversations. The project focuses on how conversation history can be organized into actionable information for monitoring opportunities, identifying pending items, and supporting sales decisions.
Project Overview
WhatsApp is widely used as a sales communication channel in Brazil, especially in small and medium-sized companies. In this context, commercial negotiations often happen partly or entirely in chat conversations, but the information remains fragmented and difficult to monitor without a dedicated system.
The proposal described in the report addresses that problem by structuring conversation data, extracting relevant sales events, and turning them into indicators that can support both managers and salespeople.
Challenge
The main challenge is how to support managers and salespeople in tracking commercial negotiations conducted via WhatsApp when there is no automatic record of the sales history and no reliable way to consolidate information manually at scale.
The report explores a platform that can reconstruct the progression of opportunities, calculate monitoring indicators, and identify pending issues while respecting access restrictions and handling personal information appropriately.
Context Analysis
The analysis describes a setting in which sales teams use dedicated WhatsApp numbers for professional activity. In these situations, negotiations may occur without a system that automatically records the conversation trail, making it harder to evaluate the stage of each opportunity.
Key observations include the following:
- Negotiations happen in WhatsApp without automatic organization of the conversation history.
- Managers may not have a tool that consolidates the commercial record or produces indicators without manual reading.
- Salespeople often depend on their own initiative to record information, which can affect data completeness and reliability.
These observations are presented as preliminary and are expected to be validated during the project investigation phase.
Personas
The report defines two hypothetical proto-personas based on preliminary market analysis.
Sales Manager Persona
Carlos represents sales managers in SMEs. He wants to track opportunity progress, identify risks early, and access consolidated, verifiable information without depending only on verbal reports.
His main difficulty is the need to manually read conversations to understand what is happening in negotiations.
Saleswoman Persona
Ana Paula represents sales professionals who manage multiple WhatsApp conversations at different stages at the same time. She needs help prioritizing urgent interactions, avoiding missed follow-ups, and documenting her work through objective records.
Her main difficulty is keeping track of open conversations and commitments without a control system.
Benefits and Justification
The project is justified by the difficulty managers and salespeople face when trying to follow the trajectory of negotiations without automated consolidation. The report highlights several practical issues:
- Manual extraction of negotiation details consumes time and is prone to errors.
- Commercial history becomes fragmented across individual conversations.
- At-risk opportunities may only be identified too late.
- Salespeople may struggle to organize follow-ups and pending tasks.
- Manual recording competes with the time needed for service and selling.
The expected value of the platform is to reduce manual effort, improve visibility, and provide organized information for monitoring and prioritization.
Solution Scope
The project is positioned within software architecture and artificial intelligence, combining systems integration, natural language processing, data storage, and role-based access control.
The proposed platform includes:
- Integration for reading WhatsApp conversation events;
- Natural language processing to extract stages, events, pending items, and indicators;
- Storage of commercial history with personal information handling;
- Dashboards and alerts based on user role and responsibility.
The report also notes that compliance, data handling, and implementation details must be confirmed before deployment.
Research Hypotheses
The document presents testable hypotheses to guide project investigation and evaluation.
- Managers spend significant effort consolidating negotiation information manually.
- The time to first human response is a useful monitoring indicator.
- Pending issues and follow-ups are often not recovered on time.
- Relevant commercial events can be automatically extracted from conversations with useful correspondence to human review.
These hypotheses are intended to be validated through interviews, analysis of authorized conversations, and comparison with human review.
Monitoring and Prioritization
The report identifies three types of alerts that the platform should distinguish:
- No first response to a new contact.
- Pending response to a question or request during negotiation.
- Scheduled follow-up that has not yet been completed.
These alerts support day-to-day management by helping teams focus on the most urgent commercial situations.
Final Considerations
The project proposes an incremental innovation for sales operations that depend heavily on WhatsApp. Rather than replacing the sales process, the platform aims to make commercial activity more visible, measurable, and easier to manage.
Its value lies in converting conversation history into structured intelligence that can support sales managers, individual salespeople, and broader CRM workflows.
| Resultado esperado | Indicador possível |
|---|---|
| Redução do esforço de consolidação manual | Tempo dedicado à consolidação antes/depois |
| Identificação mais ágil de pendências | Quantidade de pendências identificadas por período |
| Melhoria da organização das oportunidades | Avaliação de usabilidade e utilidade pelos usuários |
| Apoio à continuidade do acompanhamento | Frequência de acompanhamentos realizados dentro do prazo |
| Confiabilidade dos registros extraídos | Correspondência entre extração automática e revisão humana |
| Aspecto a investigar | Hipótese | Método de coleta | Evidência necessária | Status |
|---|---|---|---|---|
| Esforço de consolidação manual pelo gestor | H1 | Entrevistas semiestruturadas | Relato de tempo/frequência e percepção de valor | A coletar |
| Relevância do tempo de primeira resposta | H2 | Entrevistas e avaliação técnica | Confirmação de relevância e correção do cálculo | A coletar |
| Frequência de pendências não atendidas | H3 | Entrevistas e análise de conversas autorizadas | Relato de ocorrências e correspondência com extração automática | A coletar |
| Capacidade de extração de eventos comerciais | H4 | Análise de amostra com revisão humana | Taxa de correspondência por categoria de evento | A coletar |
| Risco / Problema | G | U | T | G×U×T | Prioridade | Grau |
|---|---|---|---|---|---|---|
| Dificuldade de consolidação manual não confirmada como problema relevante | 5 | 4 | 3 | 60 | 1º | Alto |
| Integração via Evolution sem modalidade e conformidade definidas | 5 | 5 | 2 | 50 | 2º | Alto |
| Tratamento de informações pessoais nas conversas sem mecanismo definido | 4 | 4 | 2 | 32 | 3º | Médio |
| Extração de eventos pela IA com correspondência insuficiente para uso prático | 4 | 3 | 2 | 24 | 4º | Médio |
| Prioridade | Funcionalidade | Hipótese | Dependência |
|---|---|---|---|
| 1 | Definição do escopo de dados e da integração | — | Pré-requisito geral |
| 2 | Mecanismo de identificação e tratamento de informações pessoais | — | Antes da persistência |
| 3 | Coleta e armazenamento confiável das conversas e eventos | H1, H3 | Integração definida |
| 4 | Vinculação entre vendedor, conversa e oportunidade | H1, H2, H3 | Armazenamento |
| 5 | Definição e cálculo de indicadores | H2, H3 | Vinculação |
| 6 | Dashboards e alertas por papel (gestor/vendedor) | H1, H2, H3 | Indicadores |
| 7 | Extração experimental de eventos e etapas por NLP | H4 | Conversas armazenadas |