A modest fashion brand from Tangerang, selling hijabs and modest wear, was doing around Rp 180 million a month across Shopee and Tokopedia. Healthy numbers, growing steadily, and the owner was increasingly uneasy. This is a marketplace to own store case study about what she did next, told honestly, including the three slow months nobody puts in the highlight reel.
Her unease had a specific source. Marketplace fees and mandatory campaign discounts were eating 12 to 18 percent of revenue depending on the month. Worse, she had thousands of buyers she could not talk to. The marketplaces owned every customer relationship. She had repeat buyers, she was sure of it, but she could not name them, reach them, or reward them.
The obvious advice she kept hearing was "leave the marketplaces, go direct." That advice is wrong, and the reason this project worked is that we ignored it.
The Strategy: Addition, Not Migration
Marketplaces are where Indonesian shoppers discover products. Killing that channel to force traffic to your own site means paying for every visitor through ads, from zero trust. It rarely survives contact with reality.
So the plan was explicitly additive:
- Marketplaces stay as the acquisition engine. New buyers find the brand there, with the trust and free-shipping programs the platforms provide.
- The own store exists for the second purchase. Repeat buyers are moved, gradually and voluntarily, to the direct channel where margin is better and the relationship is owned.
- Nothing is deprecated. If the own store had flopped, the business would have lost some tooling cost and nothing else.
The economics that justify this are simple. On a Rp 150 thousand order, the marketplace channel netted roughly Rp 125 thousand after fees and campaign costs. The same order on the own store, after payment gateway fees of around 2 to 3 percent and shipping handled the same way, netted about Rp 145 thousand. A repeat buyer moved to direct was worth roughly 15 percent more per order, before counting the value of owning the relationship.
What We Built, and What We Deliberately Skipped
The build was intentionally small: a straightforward store on a managed e-commerce platform, local payment gateway with virtual accounts and QRIS, and instant courier rate integration. Total setup was in the range of Rp 30 to 40 million including design, plus monthly platform and gateway costs.
Just as important is what we did not build in phase one: no mobile app, no loyalty point engine, no custom checkout. Every one of those was proposed at some point, and every one was deferred until the channel proved itself. Keeping stock honest across three channels mattered more than any feature; the store, both marketplaces, and a simple stock sheet had to agree, or the first oversell on a payday weekend would poison marketplace ratings, the very channel we were protecting.
The Timeline, Including the Ugly Part
Months 1 to 3: near silence. The store went live and did Rp 4 million in its first month. Rp 4 million, against Rp 180 million on marketplaces. Month two was Rp 7 million. The owner asked, reasonably, whether this had been a mistake. This is the phase where most own-store projects get quietly abandoned, and the only reason this one survived is that we had agreed in advance to judge it at month six, not month two.
Months 3 to 4: the flywheel gets a push. Two moves changed the slope. First, every marketplace parcel started shipping with a small printed card: thank you, brand story, and a voucher code worth 10 percent on the first direct order. Second, the brand's Instagram, which had 28 thousand followers accumulated over years, finally had somewhere to send people other than a marketplace link. Bio link, story links, and product tags all pointed to the own store.
Months 5 to 6: repeat buyers show up. Direct revenue reached Rp 22 million in month five, then Rp 31 million in month six. The composition was the interesting part: over 60 percent of own-store orders came from customers whose names already appeared in past marketplace shipping labels. The insert card was doing exactly its job, converting anonymous marketplace buyers into known direct customers.
Month 9, where things stood at last review:
| Metric | Before | Month 9 |
|---|---|---|
| Marketplace revenue | Rp 180 jt/month | Rp 195 jt/month |
| Own store revenue | 0 | Rp 48 jt/month |
| Own store share of total | 0% | ~20% |
| Known customer database | 0 | 3,400 buyers with consent |
| Own store repeat rate | n/a | 38% of orders |
Note the first row. Marketplace sales did not drop; they grew slightly. The own store was not cannibalizing, it was capturing purchases that used to be one-off marketplace transactions and turning them into a relationship.
What Made the Difference
Looking back, four decisions carried the project:
- Judging the channel on repeat economics, not launch spike. The store was built for the second purchase, so the fair evaluation window was six months, not six weeks.
- The parcel insert. Cost per card was trivial, and it was the single highest-ROI marketing asset in the project. Marketplaces control the storefront, but the brand controls the box.
- Instagram as the traffic bridge. The brand already owned an audience; it had just been donating that audience to marketplaces. Redirecting existing attention is far cheaper than buying new attention.
- Watching the numbers weekly. Conversion rate, traffic source, and checkout drop-off told us which fixes mattered. The store's checkout completion improved from 41 to 63 percent after two specific changes surfaced by the data. Knowing which numbers deserve attention is its own skill, one I cover in Website Analytics: The Only Metrics Worth Your Attention.
One operational warning from the trenches: multi-channel selling makes inventory the silent killer. Before adding a channel, be sure your stock control can survive it. The selection criteria are in Inventory Management Software: A Buyer's Guide for SMEs.
The Takeaway
The lesson from this marketplace to own store case study is not "go direct." It is: keep the marketplaces for discovery, build the own store for the relationship, and give the direct channel six months and a parcel insert before you judge it. The brand now owns 3,400 customer relationships it did not own a year ago, earns roughly 15 percent more on every direct order, and lost nothing on the marketplace side to get there.
If your marketplace revenue is healthy but you cannot name a single repeat customer, you have the same unease this owner had, and the same additive path is open to you.