Building an analytics system from scratch for a B2B product (Yavendió)
Context
Yavendió is a platform that automates the sales flow in conversational channels, mainly WhatsApp, for businesses with high volume of queries and small commercial teams. At the time of my joining, the company did not have a structured metrics system. Performance analysis was done manually, which limited business visibility and the ability to make informed decisions.
Problem
Performance was measured through manual processes:
- counting conversations directly from the interface
- individual chat review to identify sales
- manual recording in Google Sheets
This generated multiple limitations:
- lack of real-time visibility
- inability to reliably measure conversion
- difficulty identifying which clients or leads performed best
- no segmentation or attribution capability
- high operational cost in analysis
Decisions were made with incomplete information and through a slow and inefficient process.
Approach
Prioritization of implementation speed over technical perfection, with the goal of generating visibility as soon as possible.
- direct access to the database (PostgreSQL)
- building queries for data extraction
- design of a base dataset for analysis
- implementation of a simple pipeline (DB → Google Sheets → Looker Studio)
The goal was to quickly create a reliable source of information for the business.
Implementation
The company's first operational analytics system was developed:
- data extraction and transformation from PostgreSQL
- automated dataset updates (hourly frequency)
- dashboard construction in Looker Studio
Initial metrics:
- conversation volume (by client and date)
- lead segmentation (cold / warm / hot)
- temporal traffic evolution
System evolution:
- implementation of sales tracking through database flags
- initial conversion calculation
- incorporation of attributes such as lead source
- creation of personalized dashboards per client
Extensions:
- Slack integration for operational alerts
- dashboard use by Customer Success, Marketing and Sales teams
- incorporation of metrics access as part of the sales pitch
Result
- first centralized and automated metrics system in the company
- elimination of manual analysis processes
- enabling continuous business monitoring
Impact on operations:
- client prioritization based on performance
- improved decision-making by Customer Success team
- more informed optimization of acquisition campaigns
Business impact:
- use of metrics as value proposition in sales
- increased client engagement through personalized dashboards
Quantitative Impact
- Analysis time reduction: from ~2 manual hours → less than 5 minutes
- Elimination of: manual chat counting, individual conversation review
- Acquisition improvement: estimated 20–30% increase in qualified leads
Adoption
The system was adopted across the organization:
- daily use by founders and operational teams
- integration into Customer Success workflow
- recurring use by clients
It became a central reference source for business status, including the main metric (conversation volume), directly linked to revenue.
Evolution
System growth led to structural decisions:
- migration to Metabase for greater scalability
- democratization of data access within the organization
- foundation for subsequent product analytics implementation (PostHog)
Learnings
- implementation speed is critical in contexts without data
- understanding what metrics the business needs is more complex than building the system
- a functional MVP can generate high impact if it solves a real gap
- adoption is as important as technical implementation
- analytics systems can evolve into commercial advantages