How a Bali F&B Chain Cut Inventory Waste by 22% Through Dashboards

The operations manager at a Bali-based F&B chain with seven locations made the call after a particularly bad month. Their inventory waste had hit Rp 95 juta — produce going off, prepared dishes binned at end of service, ingredients ordered for menu items that didn’t sell. The pattern wasn’t new; the magnitude was.

Three months later, monthly waste was down to Rp 62 juta. The change wasn’t dramatic kitchen reform — it was visibility. Here’s what we built and what it surfaced.

The starting picture

A two-week diagnostic showed the cost more precisely:

• Average monthly inventory waste: Rp 78 juta across all seven locations.
• Pattern: Two locations accounted for 51% of total waste. Nobody had noticed because each location reported separately.
• Reason for invisibility: Each location’s manager kept their own records. The chain’s central office only saw monthly aggregates.
• Decision-making rhythm: Inventory orders for each location were made by that location’s manager based on intuition. No data feedback loop.

The waste wasn’t a kitchen problem. It was a coordination problem disguised as a kitchen problem.

What we built

A dashboard with three views, fed by a small data layer pulling from their POS, their ordering system, and a manual daily inventory count app we built for the kitchens.

View 1 — Daily waste tracker, by location

Each location’s daily waste, split by category (proteins, produce, prepared dishes). The chain manager opens this every morning. The two outlier locations became immediately visible — both consistently 2–3x the chain average.

View 2 — Menu item velocity

Which menu items sell at which locations, and how that varies day-of-week and season. Location managers had been ordering the same ingredient quantities for items that sold very differently across locations.

View 3 — Inventory forecast

Using the velocity data, suggested order quantities for each location’s main ingredients. The location managers could override (and frequently did at first), but the suggestions provided a baseline they hadn’t had.

Total build: Rp 85 juta over 9 weeks.

What changed in the first three months

The patterns the dashboard surfaced changed behaviour:

Location 4 was making the wrong protein decision

The dashboard showed Location 4 was ordering 18kg of beef per week against velocity that suggested 11kg. The location manager had been over-ordering for nearly a year because they didn’t want to risk running out on a weekend. They didn’t realise how consistent the over-ordering had been.

After the data conversation, weekly beef orders dropped to 13kg. Waste from beef alone dropped Rp 4 juta/month.

Location 7 had a menu mix problem

Location 7’s customers ordered different dishes than the other locations — heavier on plant-based, lighter on seafood. Their inventory ordering didn’t reflect this; they were stocking proportionally to chain average. The data made it visible. Their menu rotation also adjusted, dropping two seafood items that consistently sold poorly there.

The chain-level “promo of the day” was over-stocked

The chain had a rotating promotion. Each location ordered for the promotion based on a chain-wide guideline. The data showed adoption varied by 4x between locations. Locations with low adoption were ordering 4x what they needed and binning the rest.

The fix was per-location promotion ordering targets based on historical adoption. Roughly Rp 7 juta/month saved.

The numbers at month 3

• Average monthly waste: down from Rp 78 juta to Rp 61 juta. A 22% reduction, or roughly Rp 200 juta annualised savings.
• The two outlier locations brought their waste to within 30% of the chain average (down from 2–3x).
• Order accuracy: location managers reported they spent less time second-guessing orders because the dashboard provided baseline recommendations.

What we got wrong

Two things, worth being honest about.

The first was the daily inventory count app. We assumed kitchen staff would log waste in real-time during service. They didn’t — they were busy. We had to redesign so logging happened at the end of each shift, in a 5-minute batch process. This delayed go-live by three weeks.

The second was treating location managers as a uniform group. Some embraced the data; others felt monitored. We had to spend time individually with the resistant ones, framing the data as a tool for them rather than a check on them. The framing matters more than we initially budgeted for.

What this kind of project costs

For an Indonesian F&B chain with 5–15 locations:

• Build: Rp 60–150 juta, 8–14 weeks
• Ongoing: Rp 2–5 juta/month (hosting + occasional maintenance)
• Payback: typically 3–6 months from waste reduction alone

The upfront cost feels significant. The waste cost was bigger and recurring.

What we’d recommend to similar operations

Three patterns from this case that generalise:

1. Don’t trust averages across locations. Aggregate numbers hide where the problems actually are. Always look at per-location detail.
2. Make the data flow back to the people whose decisions it should inform. A dashboard the location manager can’t access doesn’t change behaviour.
3. Frame visibility as a tool for the team, not a check on them. Same data, different framing, dramatically different adoption.

If you’re running multi-location operations and inventory waste or coordination is biting, an hour of conversation usually clarifies whether a similar dashboard project would fit. We do those at no cost.