Why Tokopedia and Shopee Are Not Your AI Search Strategy

I sell on Tokopedia. Through Hibrkraft, my leather goods company. We have reviews. We have ratings. We have repeat customers who buy our handmade journals and leather covers without even checking the listing description anymore. By marketplace standards, we are doing fine.

And none of it matters for AI search. Not one review. Not one five-star rating. Not one "Produk terbaik, pengiriman cepat" comment.

When someone asks ChatGPT "Who makes quality handmade leather journals in Indonesia?", Hibrkraft does not show up. Neither do most brands that sell exclusively through marketplaces. It does not matter if you have 10,000 reviews on Shopee or a Power Merchant badge on Tokopedia. AI does not see those signals. They exist inside walled gardens that AI crawlers cannot access, and even if they could, marketplace data does not carry the kind of signals that AI systems use to determine authority.

This is not a failure of AI. This is a structural mismatch between what marketplaces provide and what AI search needs.

Two different infrastructure problems

Most businesses treat online presence as a single problem. Get visible. Sell stuff. Whether that visibility comes from Tokopedia, Shopee, Google, or ChatGPT feels like a detail. It is not. These are fundamentally different infrastructure problems with different inputs, different outputs, and different requirements.

Marketplace visibility runs on one set of signals. AI search visibility runs on a completely different set. And the overlap between them is almost zero.

Here is what I mean.

Count the check marks. For marketplace signals, almost nothing transfers to AI. For AI signals, almost nothing comes from marketplaces. These are not overlapping problems. They are parallel infrastructure that share a surface-level goal (visibility) but require completely different investments.

Why marketplace data stays trapped

Tokopedia, Shopee, Lazada, Bukalapak. All of them are walled gardens. Their review data, transaction data, and seller metrics exist inside closed systems designed to keep users on the platform. This is by design. Shopee does not want you to build independent authority. Shopee wants you to depend on Shopee.

When Perplexity or ChatGPT tries to answer "Who makes the best handmade journals in Indonesia?", it cannot access any of the following:

• Your 4.9-star rating on Tokopedia
• Your 2,000+ completed transactions
• Your customer photos showing the actual product quality
• Your chat response time metrics
• Your Power Merchant or Star Seller badges

All of that data is locked behind login walls, rendered via JavaScript, and served dynamically in ways that AI training pipelines do not capture well. Some of it may appear in Google's traditional web index, but AI training data and AI retrieval augmentation are different things. The marketplace product page that shows up in Google Search result #47 is not the same as being cited by an AI system as a recommended source.

As I wrote in The Difference Between a Website and a Verified Digital Entity, the distinction between having an online presence and being a verifiable entity is the whole game. A Tokopedia store is an online presence. But it is not a verified entity. It is a listing inside someone else's entity.

The million-review blind spot

I tested this systematically. I picked five Indonesian brands that dominate their Tokopedia categories, each with over 50,000 reviews, and asked ChatGPT, Gemini, and Perplexity about them using unbranded queries.

Example queries: "Best [product category] brands in Indonesia." "Where to buy quality [product] in Jakarta." "Who are the leading [industry] companies in Indonesia?"

The results were stark.

Brands with their own website and structured schema markup were cited in 67% of relevant AI responses. Brands with only marketplace stores? Under 8%. The gap is not marginal. It is an order of magnitude.

A brand can have a million reviews on Shopee and be completely invisible when someone asks an AI assistant "Who are the best [product] brands in Indonesia?" The reviews are not the problem. The platform is. AI cannot see inside Shopee. It does not know your reviews exist.

What marketplaces are actually good for

This is not an argument against marketplaces. I still sell on Tokopedia. I would be stupid not to. Indonesia's e-commerce market is projected to reach $94.5 billion in 2025 and nearly $194 billion by 2030 ^[1]. Shopee and Tokopedia together handle the majority of that volume. If you sell physical products in Indonesia and you are not on marketplaces, you are leaving money on the table.

Marketplaces are excellent transaction infrastructure. They handle payment processing, logistics integration, buyer protection, and dispute resolution. They give you access to millions of active shoppers who are already in buying mode. They let you test products, run promotions, and build a customer base without building your own commerce stack.

But they are not identity infrastructure. They are not authority infrastructure. And increasingly, they are not discovery infrastructure for the fastest-growing search channel in the world.

The framing from a recent analysis puts it well: "Shopee grabs their attention, while Tokopedia reassures their buying confidence" ^[2]. Both are valuable. Neither builds the kind of entity signal that makes AI systems cite you.

The entity gap in Indonesian e-commerce

Here is what I see happening across Indonesian SMEs. They build a Tokopedia store. They optimize their listings. They get reviews. They run ads inside the marketplace ecosystem. Traffic comes. Sales happen. They think they are visible.

Then AI search grows from 0% to 5% to 15% of how buyers research products. And these businesses are simply not in the conversation. Not because they are bad. Not because their products are inferior. Because they never built the infrastructure that AI can see.

As Kathleen Marrero wrote in a LinkedIn analysis: "Brands that appear in trusted industry publications, expert commentary, and authoritative content sources are more likely to be referenced by AI systems. If your brand rarely appears outside your own website, AI may have limited signals to rely on" ^[3]. Now imagine a brand that does not even have its own website. That only exists inside Tokopedia and Shopee.

That brand is invisible to AI. Not partially. Completely.

I covered a similar pattern in What ChatGPT Knows About Your Industry (And What It Gets Wrong). ChatGPT's knowledge of Indonesian industries is thin because authoritative English-language content barely exists. Now stack another layer on top: even the Indonesian brands that are successful, that have real market share and real customers, are invisible to AI because their entire presence lives inside marketplace walled gardens.

Two layers of invisibility. The industry itself is underdocumented. And the brands within it exist only inside platforms AI cannot access.

What AI search actually needs from you

If you want to be cited when someone asks an AI assistant about your industry, you need infrastructure that AI can read. Not marketplace listings. Not Instagram posts. Infrastructure on properties you own, structured in ways machines understand.

The minimum stack:

1. Your own domain with structured data. JSON-LD schema that tells AI who you are, what you do, where you operate. This is your entity anchor. Tokopedia cannot give you this.
2. Published expertise. Essays, guides, documentation on your domain that demonstrate topical authority. AI needs to see that you know your field, not just that you sell products in it.
3. Cross-platform corroboration. The same entity information consistent across your website, LinkedIn, Google Business Profile, and industry directories. AI triangulates. If your LinkedIn says one thing and your website says another, confidence drops.
4. Third-party mentions. Press coverage, guest contributions, directory listings, industry association mentions. AI weights external validation heavily. A mention on an industry publication is worth more than 10,000 marketplace reviews for AI citation purposes.

None of this replaces your Tokopedia store. This sits alongside it. You sell on marketplaces because that is where transactions happen in Indonesia. You build entity infrastructure because that is where authority lives in AI search.

As I discussed in The LinkedIn Article vs. The Website Essay, even publishing on LinkedIn, the most-cited platform in AI search, has limitations. LinkedIn gets the entity credit, not you. The same logic applies ten times over for marketplaces. Tokopedia gets zero entity credit from your store. And you get zero AI visibility from Tokopedia.

The practical gap: what this costs you

This matters because buyer behavior is shifting. Research from Passionfruit found that e-commerce sites have already lost up to 40% of traffic to AI-driven search ^[4]. Not marketplace traffic specifically, but the broader pattern is clear. When someone researches a purchase through AI instead of through a search engine, the brands that get mentioned are the ones AI can see. The brands inside walled gardens are simply absent from the conversation.

For a small Tokopedia seller doing Rp 50 million a month, this might not matter yet. But for a brand trying to move into B2B, enterprise contracts, or export markets, AI invisibility is a real problem. The purchasing manager at a multinational does not browse Tokopedia to find a leather goods supplier. They ask their AI assistant. Or they ask a colleague, who checks with their AI assistant. And if your brand does not show up in that answer, you were never in the running.

I know this because I have been on both sides. Hibrkraft sells journals on Tokopedia to individual consumers. But when a corporate client in Singapore wanted 200 custom leather portfolios for a conference, they did not find us on Tokopedia. They found us because we had a website with structured data, published content about our craft, and enough cross-platform corroboration that AI could confirm we are real and competent.

Two channels. Two infrastructure stacks. Both necessary. Neither substitutes for the other.

What I actually did about it

I will not pretend I had this figured out from the start. For the first two years of Hibrkraft, we only sold through Tokopedia and Instagram. Reviews were good. Sales were steady. I thought that was enough.

Then I started noticing that when people asked ChatGPT about handmade leather goods in Indonesia, we were nowhere. Brands with less skill, fewer reviews, and worse products showed up because they happened to have a website with some content on it. Or because a blogger mentioned them once. Or because they had a Google Business Profile with photos.

So I built the parallel infrastructure. Not instead of Tokopedia. In addition to it. Own domain. Structured schema. Published content about our craft, our conservation work, our process. Made sure the entity information was consistent across every platform. Got listed in relevant directories.

It took about three months of part-time work. Not a major investment. But the difference is that now when someone asks an AI "Who does leather bookbinding in Indonesia?", we actually have a chance of showing up. Before, we had zero chance. Not because of quality. Because of infrastructure.

The uncomfortable math

Indonesian e-commerce is dominated by marketplace transactions. Shopee holds roughly 21% market share, Tokopedia around 18%, with TikTok Shop growing fast after acquiring Tokopedia ^[5]. The total addressable market is massive. If you sell consumer goods, being on these platforms is non-negotiable.

But AI search is growing at a rate that makes these numbers feel temporary. The percentage of product research that starts with an AI query instead of a Google search or a marketplace browse is increasing every quarter. Within two to three years, AI-mediated discovery will be a primary channel for how businesses find suppliers, how consumers research purchases, and how institutional buyers shortlist vendors.

When that happens, the brands that exist only inside marketplace walled gardens will face a problem they did not see coming. Their sales may still be fine inside the marketplace. But the growing pool of AI-mediated buyers will never know they exist. That pool will buy from the brands AI can see. The ones that built entity infrastructure alongside their marketplace stores.

It is not either/or. It is both/and. But if you are only doing one, and you think your Tokopedia store is your digital strategy, you are building on rented land while the discovery landscape shifts under your feet.