Skip to content

Case 05 · CRM · Products · Operations

A useful dashboard ends in an action, not a chart.

I am building ZoDesk dashboards that answer what is happening with opportunities and what to do about it: where they come from, whether anyone attends them, where they drop and whether they end in a sale. Every rule has a threshold, an action and a record of what the team decided.

Project
Monitoring and action dashboards
Context
ZoDesk · car dealership in Panama
By Antonio Duque
Reviewed against the code and documentation on

01

Context and need

With several intake channels (forms, WhatsApp, portals, social), it was impossible to know which leads were left untouched, which channel went cold, or whether advertising produced buyers or just conversations.

The opposite risk existed too: dashboards full of metrics nobody turns into decisions.

02

My responsibility

  • I defined what to measure, the matching rules, the thresholds and the actions, and directed the build and review of each dashboard in ZoDesk.

03

Strategy and decisions

  1. Separate the funnel gates: arrived, contacted, replied, still in conversation, bought.
  2. Match lead and sale with explicit rules (same phone within a 120-day window) and state the limits on screen.
  3. Exclude bulk imports from time-of-day analysis so imports are not mistaken for demand.
  4. Rules that appear and disappear on their own, and team decisions stored with author and date.

04

What I built

  • Daily financing chat funnel: opened, replied, saw their result, left details; where people stop per question, and who is online now.
  • An eight-stage lead journey from arrival to sale, with its drop-offs explained (“why” and “what to do”), median time to first response, and repeats.
  • A lead monitor by channel, time slot and salesperson, with automatic rules and recorded decisions.
  • Ads-and-leads and money dashboards.
  • Normalized loss reasons: price, no down payment, no reply, financing, payment too high, bought elsewhere, fake data.

Real screenshots

Screens from the real system. ZoDesk ones come from its demo mode, with an invented dealership; those marked “Live” are public pages captured on November 14, 2026.

ZoDesk panel: each card opens the detail behind the number.

ZoDesk panel: each card opens the detail behind the number.

ZoDesk metrics panel with cards for tasks, leads, sales, receivables, paperwork and expenses.

05

How it works

  • System
  • Team
  1. System

    Sources

    Channel derived from origin, UTM and click ID.

  2. System

    Data quality

    Repeat leads from the same person and bulk imports flagged separately.

  3. Team

    Response

    Assignment, first contact and timing; touches are measured from November 14, 2026.

  4. System

    Outcome

    Loss reasons and matching with sales by phone and time window.

  5. System

    Rule

    An out-of-threshold indicator triggers a concrete recommendation.

  6. Team

    Decision

    The team accepts or dismisses the action, recorded with author and date.

Demo

From indicator to action

This is how the ZoDesk lead monitor shows the funnel: each light is a lead moving left to right. At every gate, those who don't pass drop below it, with their number and the reason. You see where the money leaks without reading a table.

Simulation · fictitious data

All channels: Arrived 120, Contacted 84, Replied 51, Still talking 23, Bought 6.

  • Above: leads moving forward
  • Below: where they stopped and why
  • The biggest leak

Lead monitor

Simulation · fictitious data

  1. Arrived120
  2. Contacted84
  3. Replied51
  4. Still talking23
  5. Bought6
Observed
Classifieds portal: 14 leads, 2 contacted (14%)
Threshold
10+ leads and 20% or fewer contacted
Suggested action
Add the channel to assignment so each salesperson gets the alert. Target: over 80% contacted.

Thresholds and actions match the rules defined in ZoDesk. The numbers are invented and do not represent any client.

06

Tools and what they do

FastAPI and Jinja2
Dashboards served by ZoDesk, without external BI tools.
SQLite and a Postgres mirror
Production data and a copy for queries.
Dealership website API
Chat events and form leads.
Automated tests
Rules and indicators covered by tests before they are shown.

Available evidence

  • Chat dashboard with 28 tests and journey dashboard with 23 tests.
  • Lead monitor with 40 written tests, not yet published.
  • A team decisions table with author and date.

07

Current status

  • Financing chat funnel

    Built November 11, 2026; the open count is a floor from that date.

    Built and tested
  • Eight-stage lead journey

    Built November 13, 2026.

    Built and tested
  • Lead monitor with rules and decisions

    In development as of November 14, 2026; not yet published.

    In progress
  • Touch tracking (call, WhatsApp, open)

    Started November 14, 2026; no earlier series.

    In progress
  • Ad spend by audience inside ZoDesk

    Meta Ads and Google Ads read-only connections still needed.

    Pending

“In production” is only used when the project’s documentation confirms it. “Built and tested” means automated tests exist, not that it is in use. “Proposal” is a documented idea, not something built.

Next steps

  • Publish the lead monitor and gather weeks of data before judging rules.
  • Connect ad platforms read-only to measure cost per lead inside ZoDesk.
  • Ask the website for per-session journeys with their campaign.

08

Lessons learned

Rule targets are goals, not results, and I show them as such.
Phone-based lead-to-sale matching is useful but limited; someone who buys with another number isn't counted. Saying so on screen prevents bad decisions.
An alert that fires every day is ignored within two weeks: rules must stay quiet when everything is fine.

Related services

Connected cases