Why We Built an Analytics Dashboard — and What We Learned About Reports People Actually Use

The Reports Nobody Opens

When we researched existing POS systems in the market, one pattern kept appearing: lots of reports, few actually used.

Some POS platforms offer 30+ report types — from inventory aging to customer cohort analysis. Sounds impressive in a sales pitch. But when we talked to cafe owners using these tools, most said the same thing: "I only open the daily sales report. The rest? Never."

This isn't a problem of users being "not tech-savvy enough." It's a problem of design that doesn't match operational reality. Small cafe owners don't have the time or training to interpret complex reports. They need answers to specific questions — not spreadsheets that require decoding.

The Questions They Actually Want Answered

If you ask a cafe owner "what data do you need?", the answer is often technical: "sales reports", "revenue graphs", "per-product data." But if you dig deeper — "what decision are you trying to make?" — the answers get much more concrete:

• "Is today better or worse than usual?"
• "Which menu items are most profitable — not just most popular?"
• "What time should I add staff, and when can I reduce?"
• "Is this week's sales trend up or down compared to last week?"
• "What percentage of revenue comes from dine-in vs takeaway?"

These questions are simple. But answering them from raw data — scrolling through reports, calculating manually, comparing across periods — takes time that most cafe owners don't have.

Our first insight: a good dashboard answers questions, not displays data.

What We Chose to Highlight

With the principle of "answer questions, don't display data", we chose a few core metrics:

Today's revenue vs average. Not just the absolute number, but a comparison. "Rp 2.3 million" means nothing without context. "Rp 2.3 million — 15% above average Tuesday" immediately tells you whether today is going well or not.

Top items by revenue, not just quantity. The best-selling item (by units sold) isn't necessarily the most profitable. Iced Latte might sell 80 cups but with thin margins. A croissant sells 20 but with much larger margin per piece. We show both perspectives — by quantity AND by revenue — so you see the complete picture.

Trends over time. Charts showing daily sales over 7 days, 30 days, or a custom range. Patterns invisible in daily numbers — seasonal dips, growth trends, anomalies — become visible.

Per-shift breakdown. Instead of one daily number, we break it down by shift. Because the same cafe can have very different performance in the morning shift vs evening shift. And the right decisions (staffing, prep quantity, promo timing) depend on per-shift data, not daily totals.

Service mode breakdown. After building the service mode feature (dine-in vs takeaway), the revenue breakdown by mode became one of the most-viewed insights. Owners who previously "felt" most revenue came from dine-in were often surprised that takeaway could be 40-50%.

What We Chose NOT to Build

Just as important as what we built is what we decided not to build for V1:

Customer segmentation reports. Powerful? Yes. But for small cafes that don't yet collect customer data (and most don't), these reports would be empty — and features that are always empty are demoralizing.

Predictive forecasting. "Tomorrow's predicted sales based on historical data." Sounds impressive, but accuracy requires at minimum 6-12 months of data. For a new cafe that's been using the system for 2-3 months, predictions would be misleading — potentially more dangerous than having no predictions at all.

Inventory-linked analytics. For example: "how many more portions of Fried Rice can you make from current stock." This requires real-time inventory data that's accurate — and from our experience, inventory accuracy in small cafes is a significant challenge. Reports based on inaccurate data lead to wrong decisions.

Our principle: better to have 5 metrics that are accurate and always checked than 30 metrics where 25 are never opened.

Design Decision: Clarity Over Density

One thing we learned: the amount of information on screen is inversely proportional to the likelihood it actually gets read.

A dashboard packed with numbers, charts, and tables looks "sophisticated." But if the cafe owner needs 30 seconds to find the number they want — that's too long. They check the dashboard between inspecting the kitchen and handling customers. Their attention span is 5-10 seconds, not 5-10 minutes.

Some design decisions we made:

• Big numbers, small labels. Today's revenue = large number that's immediately visible. The label "Today's Revenue" = small, because after 2-3 times opening the dashboard, you already know the layout.
• Color for trends, not decoration. Green = up. Red = down. No rainbow gradients or meaningless color variety.
• Progressive disclosure. Main page = overview. Click for details. Don't dump all data on one screen.
• Mobile-readable. Cafe owners often check from their phone, not a desktop. Numbers and charts must be readable on a 6-inch screen.

What This Data Enables

An analytics dashboard isn't just about "looking at numbers." What matters is what decisions can be made from that data:

• Menu optimization. Items that sell well but have thin margins → adjust pricing or portions. Items that don't sell but have high margins → push through cashier recommendations or placement.
• Staffing decisions. Revenue by hour → know when it's peak, when it's slow. If Tuesday afternoons are always dead but Tuesday evenings are packed — the same staffing all day is wasteful.
• Promo evaluation. Revenue on promo days vs normal days → know whether a promo genuinely increased revenue or just shifted it from other days.
• Trend spotting. Weekly revenue declining three weeks in a row? That's a signal to investigate — before the decline becomes a crisis.

What's Next

Current analytics is still basic — and that's intentional. Getting the foundation right matters more than a large but fragile feature set.

What we're exploring next:

• Period-over-period comparison. "This week vs the same week last month" — so seasonal patterns become more visible.
• Per-cashier analytics. Not for micromanagement, but to identify training needs — which cashier consistently has lower average orders (might need upselling coaching).
• Automated exports. Weekly reports automatically emailed to the owner — so they don't need to log in to check trends.

All of this is still exploration. What already exists: a dashboard that provides answers to the most important questions — with enough clarity that those answers can be read in 5 seconds. Because a sophisticated dashboard that never gets opened has the same value as having no dashboard at all.