A family restaurant in the Serpong area was doing what looked like a healthy delivery business: around Rp 90 million per month through the big food delivery apps. The owner was still stressed about money, and when we sat down with the numbers, the reason was obvious. Roughly 20 percent of that revenue, about Rp 18 million a month, was going out the door in aggregator commissions.
This restaurant online ordering case study is about what happened next, and it is not the story you might expect. The answer was not "quit the apps," and it was not a fancy custom app either. It was a boring, deliberate split: aggregators for discovery, a direct channel for regulars. The tech that made it work cost less than one month of the commissions it saved.
Details are anonymized, numbers are rounded, and the owner reviewed this account. The pattern, though, applies to almost any F&B business doing meaningful delivery volume in Indonesia.
The Starting Point
The restaurant: a single location, mid-range Indonesian food, strong lunch trade from nearby offices and housing clusters. Pre-pandemic it was 90 percent dine-in. By 2021, delivery through GoFood and GrabFood had grown to roughly 40 percent of total revenue.
The monthly delivery picture looked like this:
| Metric | Value |
|---|---|
| Delivery revenue via apps | ~Rp 90 million |
| Aggregator commission (~20%) | ~Rp 18 million |
| Delivery orders per month | ~1,800 |
| Estimated repeat customers | 60-70% of orders |
That last row was the insight. When the owner and her cashier went through order names and patterns, they estimated that well over half of delivery orders came from people who ordered every week, sometimes several times a week. Offices with standing lunch habits. Families with a Friday routine.
The aggregator was doing its job, discovery, brilliantly for the 30 to 40 percent of orders that were new or occasional customers. For the regulars, it was doing nothing except collecting 20 percent for introducing the restaurant to people who already knew it. Paying a finder's fee, every week, for customers already found.
What We Did Not Do
The first instinct, and the pitch this owner had already received from a software vendor, was a custom ordering app with online payment, loyalty points, and push notifications. Quoted at around Rp 120 million plus monthly maintenance.
We killed that idea in one conversation. Nobody installs an app for a single restaurant, especially when GoFood is already on their phone. The bar for switching a regular customer to a direct channel is not "match the aggregator's app." It is "be no more annoying than the aggregator, and give them a reason."
We also did not leave the aggregators. Their discovery value was real and measurable. New customers kept arriving through them every week, and that pipeline feeds the regulars pool. The strategy was never anti-aggregator. It was pay for discovery, stop paying for loyalty.
The Actual Solution: WhatsApp Plus a Menu Page
The direct channel that worked was almost embarrassingly low-tech:
- A single menu page on the web. One mobile page with photos, prices, and an "Order via WhatsApp" button on every item that opens a chat with a pre-filled message. Built in days, hosted for almost nothing.
- WhatsApp Business on a dedicated number. Catalog filled in, quick replies for the ten most common questions, a saved order format the staff pastes for new direct customers.
- Payment by QRIS or transfer, confirmed with a screenshot in the chat. No payment gateway integration in version one.
- Delivery via on-demand couriers (the same instant courier services everyone uses) for far orders, and the restaurant's own motorbike for orders within roughly 3 km, which covered most regulars.
Total setup cost, including the menu page, photography done on a phone, and staff training: under Rp 8 million. Ongoing cost: courier fees passed to the customer at cost, plus a cashier who now spends part of her shift managing the WhatsApp number.
The migration mechanics mattered more than the tech. Every aggregator order went out with a small printed card: order direct via WhatsApp, get 10 percent off your next order and free delivery within 3 km. Ten percent sounds generous until you remember the aggregator was taking twenty. The offer was funded entirely by the commission it replaced, and the customer got cheaper food. Both sides won; only the middleman's cut shrank.
The Numbers After Four Months
The shift was gradual, which is what you want, since a sudden drop in aggregator volume can hurt your ranking inside those apps.
| Metric | Before | After 4 months |
|---|---|---|
| Aggregator delivery revenue | ~Rp 90 million/mo | ~Rp 55 million/mo |
| Direct (WhatsApp) revenue | ~0 | ~Rp 42 million/mo |
| Total delivery revenue | ~Rp 90 million | ~Rp 97 million |
| Commission paid | ~Rp 18 million | ~Rp 11 million |
| Discount cost on direct channel | 0 | ~Rp 3.5 million |
| Net margin recovered | ~Rp 3.5-4 million/mo, growing |
Two things worth noting honestly. First, the savings are real but not magic: the 10 percent discount, courier costs, and staff time eat into the recovered 20 percent. The owner nets roughly Rp 4 million a month more than before, on slightly higher total volume, and the direct share is still climbing as more regulars convert. Second, there was operational pain. Peak-hour WhatsApp chaos in week two forced them to formalize the order format and add a second staff phone. Direct channels are not free of effort; they trade commission for workload.
The less measurable win might be the biggest: the restaurant now owns its customer list. Several hundred WhatsApp contacts of proven repeat buyers, reachable for free with a new menu announcement, versus zero customer data from the aggregator era. When they launched a Ramadan hampers pre-order the following season, it sold out through that list alone.
What Makes This Repeatable
This case worked because three conditions held, and you should check them before copying it:
- A real base of repeat customers. If most of your delivery orders are one-time discovery orders, the aggregator is earning its commission and this play does not apply yet.
- Order volume worth the workload. Below roughly 300 delivery orders a month, the staff effort of running a direct channel may cost more than the commission saved.
- Someone accountable for the channel. A WhatsApp number that goes unanswered for 40 minutes at lunch will burn regulars faster than any commission ever did.
Notice also what the solution was not: it was not the most software the budget allowed. It was the least software the problem required. That instinct, buying or building only where it changes your economics, is the same one I unpack in SaaS vs Custom Software: The Decision Framework. And the sequencing, payments and presence before anything custom, follows the exact order I recommend in UMKM Go Digital: A Realistic 90-Day Roadmap.
The Practical Takeaway
If you run an F&B business doing serious delivery volume, pull three numbers this week: monthly aggregator revenue, commission paid, and your best estimate of what share of orders are regulars. If regulars are more than half, you are paying discovery prices for loyalty, and a direct channel funded by that very commission will almost certainly pay for itself in a quarter.
Keep the aggregators for what they are genuinely good at: being found. Move the people who have already found you to a channel you own. If you want a second pair of eyes on the numbers before you commit, that is exactly the kind of problem I like working through with owners, and you can reach me via the partnership page.