Using Data for Decision-Making - Social Enterprise junction

Kabelo’s agricultural input distribution company in Botswana was bleeding money. Every month for six months, costs exceeded revenue. He knew something was wrong but couldn’t pinpoint it.

His investors were getting nervous. His team was working harder than ever. But the losses continued. Finally, desperate, Kabelo did something he’d been avoiding: he spent a weekend building a simple spreadsheet tracking every transaction. Not for donors or investors — just for himself. He wanted to understand where money was actually going.

What he discovered shocked him.

The problem wasn’t what he thought. He’d assumed transport costs were killing him (they weren’t — only 12% of expenses). He’d worried about theft (minimal — less than 2%). He’d blamed low margins (actually decent at 28%).

The real killer? Inventory sitting too long. Products were taking 45 days average to sell instead of the 15 days he’d assumed. That meant:

Cash tied up for 3x longer than planned.
Storage costs tripling.
Some products expiring before sale.
Constant cash flow crunch requiring expensive short-term loans.

Once Kabelo saw the data, the solution was obvious: tighter inventory management, faster stock turnover, smaller but more frequent orders. Within three months, average inventory time dropped to 18 days. Cash flow transformed. The business became profitable.

The tool that saved his business? A basic Excel spreadsheet he built himself in one weekend.

Not sophisticated data science. Not expensive analytics platforms. Not consultants or external analysis. Just the right data, tracked consistently, used to make one critical decision.

Let me show you how.

Why Data Matters More When Resources Are Scarce

Most entrepreneurs think data is a luxury — something for well-funded companies with data teams and fancy software. This is completely backwards.

When you have abundant resources, you can afford bad decisions. Make a mistake? Pivot. Launch a product that fails? Try another. Waste money on ineffective marketing? Increase the budget.

When resources are constrained, every decision matters. One bad hire can cripple your team. One wrong product investment can drain your runway. One inefficient process can kill your margins.

This is why data-driven decision making is more critical for resource-constrained enterprises than for well-funded ones.

Yet most social enterprises make decisions based on:

Gut feeling and assumptions.
Anecdotal evidence (“I talked to a few customers…”).
What donors want to hear.
What worked for someone else somewhere else.
Hope and optimism.

The result? Expensive mistakes that could have been avoided with basic data.

The Five Data Mistakes That Kill Enterprises

Before we talk about what works, let’s understand what doesn’t.

Mistake 1: Collecting Data for Donors, Not Decisions

The most common data mistake: enterprises meticulously track impact metrics for donor reports while ignoring operational data that would actually improve their business.

Example: A Ugandan agricultural training organisation tracks:

Number of farmers trained (for donors).
Demographics of participants (for donors).
Pre/post knowledge test scores (for donors).
Self-reported behaviour change 6 months later (for donors).

But they don’t track:

Cost per farmer trained (profitability).
Training completion rates (quality/effectiveness).
Which topics farmers find most valuable (product development).
Which training methods work best (efficiency).
Farmer acquisition costs by channel (marketing optimization).

The result: Beautiful donor reports, but no idea how to run the business better.

The fix: Track donor metrics because you must, but prioritise operational metrics because they determine survival.

Mistake 2: Over-Engineering Data Systems

Many enterprises try to build sophisticated data infrastructure they can’t maintain.

Example: A Ghanaian health products distributor invests in:

Custom-built inventory management software.
Tablet-based sales tracking app.
Cloud-based analytics dashboard.
Integration with multiple external systems.

Within six months:

Software has bugs nobody can fix (developer moved on).
Tablets break, aren’t replaced (no budget).
Dashboard shows old data (sync issues).
Staff revert to paper and WhatsApp (easier, more reliable).
Thousands of dollars wasted.

The lesson: Simple systems you actually use beat sophisticated systems that fail.

The fix: Start with spreadsheets. Only upgrade when you’ve maxed out simpler tools.

Mistake 3: Collecting Data You Never Use

Enterprises often collect mountains of data “just in case” without clear purpose.

Example: A Zambian microfinance organisation collects 47 data points per customer loan application. Loan officers spend 30 minutes per application on data entry.

When asked what they use the data for: “We generate monthly reports.” When asked what decisions those reports inform: Long silence.

Half the data is never looked at. The data collection burden slows operations and frustrates staff.

The fix: Only collect data that will inform specific decisions. For everything else, ask: “What would we do differently if we had this data? If the answer is ‘nothing,’ stop collecting it.

Mistake 4: Ignoring Data Quality

Garbage in, garbage out. Bad data is worse than no data because it leads to confidently wrong decisions.

Common data quality issues affecting enterprises:

Staff entering fake data to hit targets.
Inconsistent definitions (what counts as “trained farmer” varies by region).
Data entry errors accumulating.
Old data never cleaned or updated.
Multiple systems with conflicting numbers.

Example: A Nigerian solar distribution company’s sales dashboard showed strong growth. Management celebrated, made hiring plans, ordered more inventory.

Then they ran out of cash.

Turns out sales data included “commitments to buy” not actual sales. Real sales were 40% of reported numbers. The data was systematically inflated.

The fix:

Regular data quality audits.
Spot-check samples against reality.
Cross-reference different data sources.
Clear definitions consistently applied.
Consequences for data falsification.

Mistake 5: Analysis Paralysis

Some enterprises collect data obsessively but never make decisions from it.

They’re waiting for:

More data (“we need at least 12 months…”).
Perfect data (“we should fix these quality issues first…”).
Advanced analysis (“we need a data scientist to interpret…”).
Complete certainty (“we’re not sure what this means…”).

Meanwhile, competitors make decisions faster and learn through action.

The fix: Use data to inform decisions, not to achieve certainty. Make the best decision with available data, then monitor results and adjust.

What Data Actually Matters (The Essential Metrics Framework)

You don’t need 50 dashboards. You need the right 5-10 metrics that drive your business.

Category 1: Financial Health Metrics

These tell you if you’re surviving.

Essential metrics:

Cash runway: How many months can you operate with current cash?

Why it matters: Runway = survival.
How to calculate: Current cash ÷ average monthly burn.
How often to check: Weekly.
What triggers action: <3 months runway

Revenue vs. costs (by product/service): Is each offering profitable?

Why it matters: You might be profitable overall but specific products are killing you.
How to calculate: Track revenue and direct costs separately for each offering.
How often to check: Monthly.
What triggers action: Any product with negative margin for 3+ months.

Customer acquisition cost (CAC) vs. lifetime value (LTV): Does it cost more to acquire customers than they’re worth?

Why it matters: Unsustainable growth is worse than no growth.
How to calculate: CAC = total marketing/sales costs ÷ new customers; LTV = average customer value × expected retention.
How often to check: Monthly.
What triggers action: CAC > LTV.

Example — Sunculture (Kenya, elsewhere):

SunCulture sells solar irrigation to smallholder farmers. They obsessively track CAC by acquisition channel (field agents, radio ads, farmer referrals, demo days).

They discovered:

Radio ads: High CAC (12,000 KES per customer).
Demo days: Medium CAC (4,500 KES).
Farmer referrals: Low CAC (800 KES).

Decision: Shifted budget from radio to referral programmes. CAC dropped 60% while sales increased.

Category 2: Operational Efficiency Metrics

These tell you if you’re using resources well.

Essential metrics:

Productivity per employee/resource: What value does each resource generate?

Why it matters: Identifies underperforming areas.
Examples: Sales per agent, farmers served per extension officer, deliveries per vehicle.
How often to check: Monthly.
What triggers action: Significant variance between similar resources.

Process completion time: How long do key processes take?

Why it matters: Time is money; slow processes kill capacity.
Examples: Order-to-delivery time, onboarding time, service response time.
How often to check: Weekly.
What triggers action: Times increasing or exceeding targets.

Error/return/failure rates: How often do things go wrong?

Why it matters: Rework costs money and erodes trust.
Examples: Product return rate, payment failures, delivery failures.
How often to check: Weekly.
What triggers action: Rates above 5% or increasing trends.

Example — Twiga Foods (Kenya):

Twiga tracks “perfect order rate” — orders delivered on time, complete, correct quality, no damage.

When perfect order rate dropped from 94% to 87%, they investigated:

Root cause: New sorting facility had training gaps.
Solution: Additional training + process adjustments.
Result: Rate back to 95% within two weeks.

Without tracking this metric, the quality decline would have continued unnoticed until customer complaints escalated.

Category 3: Customer Behaviour Metrics

These tell you what customers actually do (vs. what they say).

Essential metrics:

Retention/churn rate: How many customers stay vs. leave?

Why it matters: Acquisition is expensive; retention is profitable.
How to calculate: % of customers from last period still active this period.
How often to check: Monthly.
What triggers action: Churn rate >10% or increasing.

Repeat purchase rate: How many customers buy more than once?

Why it matters: One-time customers = you’re not creating real value.
How to calculate: Customers with 2+ purchases ÷ total customers.
How often to check: Monthly.
What triggers action: <40% repeat rate.

Product/feature usage: What do customers actually use?

Why it matters: Effort on unused features is wasted.
How to track: For physical products: which features drive repurchase. For services: usage frequency by feature.
How often to check: Monthly.
What triggers action: Features with <20% usage.

Example — M-KOPA (Kenya, Uganda, Nigeria):

M-KOPA tracks daily usage of their solar systems. When usage patterns drop (customer not turning on lights as often), it’s an early warning of potential payment default.

This allows them to:

Proactively contact customer before payment is missed.
Understand issues (broken component, economic hardship, dissatisfaction).
Take action (repairs, payment flexibility, customer service).

Result: Default rates 30% lower than if they only reacted to missed payments.

Category 4: Impact Metrics (That Actually Matter)

Not donor metrics — operational impact metrics that validate your theory of change.

Essential metrics:

Intended outcome achievement: Are you actually creating the change you intended?

Why it matters: Impact is why you exist; if you’re not achieving it, what’s the point?
Examples: Income increase for farmers, health improvement for patients, time saved for customers.
How often to check: Quarterly (outcomes take time).
What triggers action: Not achieving intended outcomes or declining impact.

Impact per dollar spent: Are you achieving impact efficiently?

Why it matters: Resources are limited; efficiency maximizes total impact.
How to calculate: Outcome achieved ÷ cost to achieve it.
How often to check: Quarterly.
What triggers action: Efficiency declining or below peer benchmarks.

Example — One Acre Fund (Rwanda, Kenya, Tanzania, others):

One Acre tracks “net income increase per farmer” after accounting for programme costs.

They discovered:

Some crops generated 60% higher income gains per dollar invested.
Some training modules had minimal impact on farmer income.
Geographic areas varied widely in cost-to-impact ratios.

Decisions made:

Shifted crop mix toward higher-impact options.
Cut low-impact training modules.
Optimized resource allocation by geography.

Result: 40% improvement in impact per dollar over three years.

The Low-Cost Data Collection Toolkit

You don’t need expensive systems. Here are proven low-cost approaches:

Tool 1: Structured Google Sheets

What it’s good for: Financial tracking, inventory, customer data, sales pipeline, operational metrics

How to use it well:

Create templates with consistent structure.
Use data validation to prevent errors (dropdown lists, number formats).
Build simple formulas for automatic calculations.
Use conditional formatting to highlight issues (red if inventory low, etc.).
Create simple charts for visual trends.
Share with team for collaborative updating.

Example template — Sales tracking:

Cost: Free

Learning curve: 1-2 days for basic proficiency

Limitations: Doesn’t scale beyond ~10,000 rows; requires manual updates

Tool 2: Mobile-Based Data Collection

What it’s good for: Field data collection, customer surveys, delivery confirmations, attendance tracking

Options:

Google Forms: Free, simple, works offline with app.
KoBoToolbox: Free, designed for development sector, offline capable.
ODK (Open Data Kit): Free, powerful, offline capable.
CommCare: Free tier available, designed for health/social services.

How to use it well:

Design forms on computer, use on phone.
Include GPS coordinates for location verification.
Use photos for proof of delivery/quality verification.
Build in validation rules to prevent bad data.
Set up automatic data sync when online.
Export to spreadsheets for analysis.

Example — BRAC Uganda:

BRAC uses mobile forms for their agricultural extension officers to record:

Farmer visits (GPS verified).
Training topics covered.
Farmer questions/issues.
Photos of farm conditions.
Follow-up needed.

Data automatically syncs to central database. Managers can see real-time activity and identify gaps.

Cost: Free – $50/month depending on volume

Learning curve: 2-3 days to build forms, 1 hour to train users

Limitations: Requires smartphones; offline mode has limits

Tool 3: SMS/USSD Data Collection

What it’s good for: Real-time updates, customer feedback, simple surveys, transaction confirmations

Options:

Africa’s Talking: Pan-African SMS platform.
Telerivet: Good for two-way SMS workflows.
RapidPro: Free, designed for social programmes.
FrontlineSMS: Free, basic SMS workflows

How to use it well:

Keep questions simple (yes/no, numbers, single word).
Use structured formats (#code-value).
Send automatic confirmations.
Build workflows that route responses correctly.
Export data regularly for analysis.

Example — Ignitia (Ghana, others):

Ignitia collects weather data from farmers via SMS:

Automated daily prompts: “Did it rain today? Reply 1=Yes 2=No”
Farmers reply with single digit.
Data feeds machine learning weather models.
Farmers get improved forecasts in return.

Cost: $0.01-0.05 per SMS.

Learning curve: 1-2 days for basic setup

Limitations: Limited data complexity; SMS costs add up at scale

Tool 4: WhatsApp for Rapid Data Sharing

What it’s good for: Photos, quick updates, coordination, informal data sharing

How to use it well:

Create specific groups for specific data types.
Establish clear formats (e.g., “DELIVERY: CustomerName, Product, Time, Photo”).
Extract data to spreadsheets weekly.
Use status updates for daily metrics.
Save important images to organized folders.

Tool 5: Simple Dashboard Tools

What it’s good for: Visualizing data, sharing metrics with team, spotting trends

Options:

Google Data Studio: Free, connects to Sheets.
Microsoft Power BI: Free tier available.
Tableau Public: Free, powerful visualizations.
Simple spreadsheet dashboards: Build in Excel/Sheets

How to use it well:

Focus on 5-10 key metrics, not 50.
Update weekly/monthly on schedule.
Use colors to highlight issues (red/yellow/green).
Include trends over time, not just current numbers.
Make it visible to whole team.
Review together regularly.

Example — Living Goods (Uganda, Kenya):

Living Goods built simple Sheets-based dashboards for community health workers showing:

Sales this month vs. target.
Product stock levels.
Customers served.
Top selling products.

Updated weekly. Health workers check their performance, managers identify who needs support.

Cost: Free – $100/month depending on tool

Learning curve: 1-3 days for basic dashboards

Limitations: Visualizations only as good as underlying data quality

How to Build a Data-Driven Culture (Without Data Scientists)

Tools are useless without culture. Here’s how to make data central to decision-making:

Practice 1: Weekly Data Reviews

What: 30-minute meeting every week reviewing key metrics

How:

Same day/time every week.
Review same core metrics.
Each metric: current status, trend, what changed, why, action needed.
Everyone comes prepared.
Focus on decisions, not just reporting numbers.

Example — Zola Electric:

Every Monday morning, leadership reviews:

Sales vs. target (by region).
Payment collection rates.
Customer service response times.
Inventory levels.
Cash position.

If any metric is red, team discusses why and assigns action owner.

Practice 2: Hypothesis-Driven Experiments

What: Test assumptions with data rather than debating opinions

How:

State hypothesis clearly: “We believe [action] will cause [result] because [reason]”
Define success metrics and timeline.
Run controlled experiment.
Measure results.
Decide based on data, not opinions.

Practice 3: Make Data Visible

What: Display key metrics where everyone sees them

How:

Physical dashboards in office.
TV screens showing real-time data.
Regular team emails with key numbers.
Individual scorecards for team members.
Celebrate data-driven wins publicly.

Why it works: Visibility creates accountability and aligns team around what matters.

Practice 4: Train Everyone in Basic Data Literacy

What: Ensure all staff can read, interpret, and use basic data

How:

Monthly “data literacy” training sessions.
Teach basic spreadsheet skills.
Show how to create simple charts.
Practice interpreting trends.
Connect data to daily decisions.

Example — Copia (Kenya):

Copia trains all field agents on:

How to read their performance dashboards.
What metrics matter and why.
How to improve their numbers.
Basic Excel for tracking customers.

Result: Agents make better decisions independently, managers spend less time explaining.

Real Decisions that Data Helped Some Enterprises Make

Decision 1: Which Markets to Enter (Tugende, Uganda)

Situation: Tugende provides motorcycle financing to boda boda drivers. They wanted to expand to new regions but had limited capital.

Data they collected:

Motorcycle registration data by district.
Population density.
Average daily boda earnings (via surveys).
Competitor presence.
Road infrastructure quality.
Payment default rates in pilot districts.

Analysis: Ranked potential markets by:

Market size (registered motorcycles).
Customer ability to pay (earnings).
Competition intensity.
Infrastructure (affects default risk).

Decision: Entered top 3 ranked markets, avoided markets that looked attractive but had high-risk factors.

Result: 75% lower default rates in data-driven market choices vs. markets chosen by “gut feel.”

Decision 2: What Products to Stock (Sokowatch, Kenya, Tanzania, Rwanda)

Situation: Sokowatch supplies products to informal retailers. Initially stocked 200+ SKUs. Inventory management was chaotic.

Data they tracked:

Sales velocity by product.
Margin by product.
Inventory turnover rate.
Storage costs.
Customer reorder rates.
Seasonal patterns.

Analysis:

80% of revenue came from 35 products.
40% of products sold <5 units/month.
Slow-moving products tied up cash and warehouse space.

Decision:

Cut to 50 core products (focused on top movers).
Added products only when data showed sustained demand.
Optimized inventory levels based on turnover data.

Result: Inventory costs down 60%, stockouts down 40%, margins up 15%.

Decision 3: How to Price Services (Pula, Multiple Countries)

Situation: Pula provides agricultural insurance. Needed to price products that were both affordable and sustainable.

Data they collected:

Historical crop yield data.
Weather patterns.
Farmer income levels.
Willingness-to-pay surveys.
Claim rates in pilot programmes.
Administrative costs.

Analysis:

Modelled risk levels by region and crop.
Calculated break-even pricing.
Tested price sensitivity with customer data.
Identified optimal price points by segment.

Decision: Tiered pricing by risk level, bundled with inputs to increase affordability, dynamic pricing based on weather forecasts.

Result: Pricing that’s both affordable for farmers and sustainable for Pula.

Your Data Action Plan

Ready to become data-driven? Here’s your roadmap:

Week 1: Identify Your Critical Metrics

Choose 5-10 metrics that matter most:

2-3 financial health metrics.
2-3 operational efficiency metrics.
2-3 customer behaviour metrics.
1-2 impact metrics.

Write down for each:

What it measures.
Why it matters.
How you’ll calculate it.
How often you’ll check it.
What number triggers action.

Week 2: Build Simple Tracking Systems

Start with what you have:

Create Google Sheet templates.
Set up mobile forms if needed.
Establish data collection protocols.
Assign responsibilities.
Test with one week of data.

Week 3: Establish Review Rituals

Create rhythms:

Weekly team data review meeting.
Monthly deep-dive analysis.
Quarterly strategic review.
Share dashboards with team.

Week 4: Make Your First Data-Driven Decision

Pick one decision you need to make:

Look at relevant data.
Analyze objectively.
Choose based on evidence.
Implement.
Monitor results.

Ongoing: Build the Habit

Reinforce data culture:

Celebrate data-driven wins.
Share learnings from data.
Train team on data skills.
Improve data quality continuously.
Expand tracking as you grow.

The Truth About Data in Resource-Constrained Enterprises

Data doesn’t need to be sophisticated to be useful. It needs to be relevant, accurate, and actually used.

The enterprises that succeed aren’t those with the fanciest analytics. They’re the ones that:

Track the right metrics consistently.
Use simple tools they can maintain.
Make decisions based on evidence, not opinions.
Build data into their culture and rituals.
Focus on action, not perfect analysis.

So stop waiting for:

More resources to invest in data systems.
Data scientists to hire.
Perfect data quality.
Sophisticated analytics tools.
Complete certainty before making decisions.

Start now with:

Spreadsheets and mobile forms.
The metrics that drive your business.
Weekly reviews that inform decisions.
Experiments that test assumptions.
Learning from data you already have.

Every day you operate without data is a day you’re making expensive decisions blind.

Build your data muscle. Start simple. Use what you learn. Get better over time.

Your mission deserves evidence-based decisions, not gut-feel gambles.

The One Question That Changes Everything

Before making your next major decision, ask:

“What data would help me make this decision better — and can I get that data in the next week?”

If yes, get the data. Then decide.

If no, make the best decision with available data, but commit to collecting better data going forward.

Never let lack of data be an excuse for indefinite delays. But never let lack of data be an excuse for preventably bad decisions either.

Use data to decide better, faster, with less waste.

That’s the competitive advantage hiding in your spreadsheets.

What’s the one decision you’re facing right now? What data would help you make it better? And what’s stopping you from collecting that data this week?