Farm Management and Data

The Spreadsheet Problem

Most coffee farms in Colombia run on spreadsheets, paper notebooks, and the administrator's memory. The Diario de Labores (daily labor diary) is a massive Excel file. The budget lives in another. Pest monitoring is noted on paper. Soil tests sit in PDF reports from the lab. Fertilizer applications are recorded somewhere else. None of these systems talk to each other, and the farmer makes decisions based on intuition and experience rather than integrated data.

This is not a criticism -- it is reality. I know because I lived it for years. Our farms ran on exactly this setup: the administrator tracking labor in one Excel, the agronomist sending field reports by WhatsApp, soil test results in PDFs from the lab, and me trying to make sense of all of it by looking at disconnected pieces of information.

The turning point came when I realized I was making fertilization decisions without knowing the current phenological stage of my plots. I was approving labor schedules without checking whether the budgeted tasks matched the actual agronomic needs. I was buying inputs without knowing what we had in inventory. The information existed -- it was just scattered.

What We Track

Across our farms, we maintain integrated data on:

• Labor -- every task assignment, every day, every plot. Who did what, where, when, and at what cost per jornal (daily labor unit)
• Budgets -- planned vs actual spending by farm, plot, crop, and category
• Fertilization -- every fertilizer application tracked by product, quantity, date, and plot, with nutrient compositions for every product we use
• Production -- wet mill lots, conversion factors, quality scores
• Pest and disease -- broca counts, roya incidence, weed assessments from field monitoring
• Soil and foliar -- lab results linked to specific plots for nutrient planning
• Phenology -- the current physiological stage of every plantation, updated regularly
• Photos -- geotagged field photos linked to plots and dates

From Data to Decisions

Raw data is noise. The value comes from cross-referencing:

Plot Intelligence

By combining soil tests, fertilization records, production data, pest monitoring, photos, and phenological stage for a single plot, we can generate a comprehensive health assessment. A plot showing low potassium in soil tests, below-average yield, and high broca pressure tells a different story than any single data point would suggest.

I had a Caturra block that was underperforming for two years. Looking at production numbers alone, it seemed like the variety was just tired. But when we cross-referenced the soil tests (low pH, aluminum toxicity), fertilization records (no lime application in 18 months), and pest data (elevated roya), the real picture emerged. The soil problem was driving everything else. We applied a lime amendment, adjusted the fertilization plan, and the plot recovered.

Budget vs. Execution

Comparing planned spending against actual execution by category reveals where resources are being misallocated. When the fertilization budget is underspent but the labor budget is over, it signals a management issue -- not just a financial one. It usually means the team is doing manual labor (desyerba, deschuponar) but skipping the technical tasks (fertilization, drench applications) that require more planning.

Phenological Alignment

Knowing that a plot is in the grain fill stage (high potassium and calcium demand) while the fertilization records show no recent K application creates an actionable alert. Timing nutrient delivery to match the plant's physiological needs is more effective than calendar-based scheduling.

This is one of the biggest wins from data integration. When our Geisha block entered grain fill last season, the system flagged that we had not applied potassium in 45 days. We corrected immediately. That lot scored 88 at cupping. Would it have scored 88 without the timely potassium? Maybe. But I would rather not gamble.

Labor Productivity

Tracking task completion rates, yields per jornal, and contractor performance across plots identifies both efficiency opportunities and quality risks. A contractor with consistently high picking speed but high conversion factor (poor cherry selection) is costing the farm in quality, not saving it in labor.

Technology in Traditional Farming

The challenge is not building the technology -- it is integrating it into farm operations without disrupting the human relationships and institutional knowledge that make farms work. Our approach:

• Data enters through existing workflows -- the administrator's Excel diary, the agronomist's field reports, the team's WhatsApp messages
• Analysis happens in the background -- automated ingestion, parsing, cross-referencing
• Insights are delivered in the language the team speaks -- Spanish-language reports with familiar metrics (jornales, arrobas, lotes)
• Decisions remain human -- the system recommends, the administrator decides

I could build the most sophisticated data system in the world, but if Carlos (my administrator) does not trust it or cannot use it, it is worthless. The technology has to serve the people, not the other way around. That means reports in Spanish, metrics they already understand, and recommendations that sound like a knowledgeable colleague -- not a computer.

The goal is not to replace the experienced farmer's judgment. It is to give that judgment better inputs.

---

Data is the new fertilizer -- it feeds better decisions. Want to see how data-driven management works on a real specialty coffee operation? Join the community at skool.com/particular-3064 for discussions on technology, decision-making, and farm management systems.