Your SEO agency sends a monthly report. Domain authority is up. Backlinks are growing. Three keywords moved onto page one.
Now open your Shopee seller dashboard and look at where your best product actually sits in category search.
Page four, probably. Sometimes page six.
None of that backlink work touched it. Not one link. Shopee's ranking algorithm has never once looked at your domain authority, and it never will, because it was not built to. Neither was Tokopedia's.
This is the part of Indonesian e-commerce that agencies rarely explain, and it costs brands more than any technical audit ever recovers.
Two Search Engines, One Product, Zero Shared Signals
Marketplace search is not a variation of Google search. It is a different system with different goals, built by people solving a different problem.
Google wants to answer a question. Shopee wants to sell something today. Those objectives produce algorithms that look at almost nothing in common.
What each engine looks at, and what it ignores
Google reads
Content depth, backlink profile, domain authority, structured data, page speed, internal linking, dwell time.
Shopee and Tokopedia read
Sales velocity, listing title keywords, conversion rate, review volume and score, seller response time, fulfilment speed, campaign participation.
One overlap: keywords. Everything else on the left is invisible on the right, and everything on the right is invisible on the left.
Read that again and the implication becomes uncomfortable. A brand with a beautifully engineered website, immaculate schema, and a strong backlink profile can sit on page five of Shopee search behind a seller who has none of those things but replies to customer messages within four minutes.
That is not a bug. That is Shopee doing exactly what it was designed to do.
What Actually Ranks a Product on Shopee
Shopee's search behaves like a recommendation engine wearing a search bar. It is trying to predict which listing will convert, right now, for this shopper, and it uses commercial evidence to do it.
Sales velocity, weighted toward the last fourteen days
Recent sales matter far more than lifetime sales. A listing that sold 200 units this fortnight outranks one that sold 5,000 units across three years and 40 units last month. The algorithm reads momentum, not history.
This is why new listings sometimes surge unexpectedly. Shopee gives fresh products a temporary visibility boost to gather engagement data, then locks in a rank based on how that test performed. Waste the boost and you spend months climbing back.
The 120-character title, and why yours is probably half empty
Shopee allows 120 characters in a product title. Most brands use forty, because forty looks clean and matches their website.
Forty characters on a website is professional restraint. Forty characters on Shopee is eighty characters of free keyword real estate handed to a competitor.
Look at how top sellers write titles:
"Ceramide Moisturizer 50ml Original BPOM Non-Comedogenic Kulit Berminyak Berjerawat Skincare Pelembab Wajah"
It reads badly. It is not meant to read well. It is meant to match every phrasing a shopper might type, including the Bahasa-English hybrids Indonesians actually use, and it does that job while your clean forty-character title matches three of them.
Conversion rate, measured per listing
If a hundred people view your listing and one buys, Shopee learns something. If a hundred view a competitor's and eight buy, Shopee learns something more useful, and reallocates visibility accordingly.
This creates a feedback loop that rewards good product photography and punishes bad pricing faster than any Google algorithm punishes thin content. The correction is immediate.
Seller metrics that have nothing to do with your product
Chat response time. Order fulfilment speed. Cancellation rate. Return rate.
These are operations problems, not marketing problems, which is precisely why marketing teams never fix them. A four-hour average response time will cap your visibility no matter how good your listing copy is, and no SEO tool on earth will tell you that.
Shopee ranking weight, directional
Directional weighting, not published figures. Shopee does not disclose exact weights. The last row is exact.
The Tokopedia Confusion Nobody Cleared Up
Here is where most brands have their strategy wrong, and it is not entirely their fault.
ByteDance took a controlling stake in Tokopedia in December 2023 for roughly 1.5 billion dollars. In 2024 the seller centers merged. Rumours circulated through 2025 and into 2026 that Tokopedia was being absorbed entirely, that sellers would be forced onto TikTok Shop, that the marketplace was finished.
TikTok formally denied it in February 2026. Tokopedia continues to operate as an independent marketplace. What merged was the operational tooling, not the platforms.
So what changed?
Sellers now manage listings, inventory, orders, advertising and creator affiliates for both platforms from a single dashboard, the Tokopedia and TikTok Shop Seller Center. One interface. Two entirely separate ranking systems underneath it.
One dashboard, two shopping behaviours
Tokopedia: intent-based
The shopper knows roughly what they want and searches for it. Rankings lean on specification match, seller location, product detail completeness, and trusted-seller status.
TikTok Shop: discovery-based
The shopper was not looking for anything. Rankings lean on content performance, creator affiliate activity, live stream engagement, and watch-through rate.
Same seller center. Same inventory. Two algorithms that reward almost opposite behaviours.
A product optimized for Tokopedia's specification-driven search will not automatically perform on TikTok Shop, where nobody typed a query in the first place. And a product that performs brilliantly through creator affiliates on TikTok can sit invisible on Tokopedia because its listing lacks the granular attributes Tokopedia's search depends on.
The single dashboard makes this failure mode easy to miss. Everything looks unified. It is not.
What Tokopedia weighs that Shopee does not
Seller location. Tokopedia's algorithm considers the distance between warehouse and buyer, because shipping time drives satisfaction. A Jakarta seller will outrank an equivalent Surabaya seller for a Jakarta buyer, all else equal. Multi-warehouse sellers gain an advantage that has nothing to do with content.
Specification completeness. Empty attribute fields hurt more here. Weight, dimensions, material, variant details. Tokopedia uses these to match granular queries, and a listing missing half its attributes cannot match half the queries.
Official Store status. Verified brand accounts receive a trust weighting. This is closer to Google's E-E-A-T thinking than anything in Shopee's system.
Why Your Website SEO Still Matters, Just Not Where You Think
None of this means abandon your own site. It means understanding what your site is for.
Marketplace listings rent you a customer. You pay commission, you inherit their trust signals, you compete inside their auction, and when you stop selling there the relationship evaporates with the listing. You own nothing.
Your website is the only place where you accumulate anything: domain authority, email addresses, first-party data, a brand entity that AI engines can recognise and cite.
The two feed each other, and the direction of that feeding is usually backwards from what brands assume. Marketplaces generate discovery. Your website converts the verification search that follows. A shopper finds you on Shopee, wonders whether the BPOM number is real, searches your brand on Google, and lands somewhere.
Whether that somewhere is your product page, structured and credible, or a reseller's listing on a competing marketplace, is decided by ecommerce SEO work you either did or did not do months earlier.
A Practical Sequence, in Order
Most brands attack this backwards, starting with the most visible task rather than the most consequential one.
What to fix first, and why
Seller operations
Response time, fulfilment speed, cancellation rate. These cap everything else and cost nothing but attention.
Listing titles
Fill the 120 characters. Use the hybrid Bahasa-English phrasing shoppers actually type, not the phrasing on your packaging.
Attribute completeness
Every empty field on Tokopedia is a query you cannot match. Fill them all, including the ones that seem irrelevant.
Review generation
Volume and specificity both matter. "Great product" teaches the algorithm nothing. "Lasts twelve hours on oily skin" teaches it a lot.
Campaign participation
Flash sales generate the velocity spike the algorithm reads as demand. Skipping them is a ranking decision, not just a margin decision.
Notice what is absent: backlinks, domain authority, content marketing. None of it moves marketplace rank.
The mistake nearly everyone makes with reviews
Brands chase five-star ratings. The algorithms care about review volume and recency at least as much, and increasingly about review content.
A listing with 400 reviews averaging 4.6 outperforms one with 30 reviews averaging 5.0, consistently. The higher score signals nothing because the sample proves nothing.
And when AI engines start reading those reviews to answer shopping questions, which is already happening, specificity becomes the asset. A review saying "perfect for oily skin in humid weather" is quotable. A review saying "good, fast shipping" is not.
Where Marketplace Optimization Meets AI Search
Something is changing that makes all of this more urgent rather than less.
Both Lazada and Shopee now run their own AI shopping assistants inside their apps. Shoppers ask a question, an AI reads product listings and reviews, and returns two or three recommendations.
Your listing is either in that answer or it is not.
And outside the marketplaces entirely, shoppers increasingly ask ChatGPT or Perplexity what to buy before they open any shopping app at all. Adobe Digital Insights tracked a 4,700 percent year-on-year rise in AI-referred traffic to retail sites. Adobe Analytics found that traffic converting 42 percent better than traditional organic visitors in the first quarter of 2026.
Those shoppers arrive having already narrowed their options to two or three names. Whether yours is among them depends on work that resembles neither marketplace optimization nor traditional SEO. Generative Engine Optimization for e-commerce runs on structured data, third-party consensus, and content architecture built for machine extraction.
Three search surfaces. Three algorithms. One product.
I wrote Found Before They Search because Indonesian brands kept treating these as competing budget lines rather than sequential stages, and the cost of that mistake compounds quietly. The companion volume, Cited or Silent, goes deeper on the AI citation mechanics specifically. Both are available as free gated editions, and both are also published in paperback and hardcover and listed on Google Play Books and Apple Books.
What This Costs You If You Ignore It
Nothing visible. That is what makes it dangerous.
Your Google rankings will keep improving, because that work is real and it compounds. Your agency report will look healthy. Domain authority climbs.
Meanwhile a competitor who never hired an SEO agency, but who answers customer chats in three minutes and fills all 120 title characters, quietly occupies page one of the search results where seventy percent of Indonesian e-commerce transactions actually happen.
You will notice eventually. Probably in a quarterly review, comparing marketplace revenue against last year, wondering which campaign underperformed.
None of them underperformed. You just optimized for the wrong engine.
Frequently Asked Questions
Do backlinks help Shopee or Tokopedia rankings at all?
No. Marketplace search algorithms do not read external links, domain authority, or any off-platform signal. They rank on commercial performance inside the platform: sales velocity, conversion rate, review data, listing keywords and seller operational metrics. Backlinks remain valuable for your own website, where they do influence Google rankings.
Did Tokopedia shut down after the TikTok acquisition?
No. TikTok formally denied this in February 2026. Tokopedia continues operating as an independent marketplace. What merged was the seller center, the operational dashboard sellers use to manage both platforms. The ranking algorithms remain separate, and sellers can choose to sell on one platform, the other, or both.
Should we use the same product title on our website and on Shopee?
No. Website titles serve human readers and Google, so they should read naturally and stay under about sixty characters. Shopee titles serve an internal matching algorithm and allow 120 characters. Filling those characters with keyword variants, including hybrid Bahasa-English phrasing, is standard practice among top sellers even though it reads badly.
How quickly do marketplace rankings respond to changes?
Much faster than Google. Sales velocity is weighted toward the last fourteen days, so a successful flash sale can move a listing within days. The reverse is also true: a period of slow sales or poor seller metrics degrades rank quickly. There is no equivalent of Google's slow, compounding authority.
Does seller location really affect Tokopedia rankings?
Yes. Tokopedia's algorithm considers warehouse proximity to the buyer, because shipping speed drives satisfaction. All else equal, a nearer seller ranks higher for that buyer. Multi-warehouse sellers gain a structural advantage unrelated to listing quality.
If marketplaces drive most sales, why invest in our own website?
Because marketplace listings rent you a customer and your website owns one. On a marketplace you pay commission, inherit the platform's trust signals, and lose the relationship when the listing ends. Your website accumulates domain authority, first-party data, email addresses, and a brand entity that AI engines can recognise and cite. Marketplaces generate discovery. Your website captures the verification search that follows it.
Do AI shopping assistants inside Shopee and Lazada change marketplace SEO?
They add a layer. Those assistants read listing content and review text to generate recommendations, which means specific, descriptive reviews become an asset in a way they were not before. A review describing exactly who the product suits gives an AI something to quote. A generic five-star rating gives it nothing.
Sources & References:
- TikTok Newsroom Indonesia - official clarification on Tokopedia and TikTok Shop Seller Center integration, addressing platform closure rumours.
- Kompas.com - reporting on the seller center merger timeline and Ministry of Trade oversight, June 2025.
- Tokopedia Seller Center - shop integration guidelines and product merging documentation.
- Adobe Digital Insights - AI-referred traffic growth to retail sites, year on year.
- Adobe Analytics - AI referral conversion premium against traditional organic traffic, Q1 2026.
- Arfadia Digital Indonesia - State of SEO Indonesia 2026. arfadia.com/resources
- Arfadia Digital Indonesia - Digital Marketing Benchmark Indonesia 2026. arfadia.com/resources