“AI has flattened the barrier to ‘making things.’ The money and the moats have all moved somewhere else — outside production.”
Start with two numbers.
Hongguo Short Drama’s DAU is around 170 million. The figure is debatable — but even discounted, it sits at the 100–200 million scale.
The overseas market’s leading short-drama apps? Their DAU probably doesn’t even reach 10 million.
That was the read from Peter, co-founder of Firefly, thrown out during a roundtable at the Unique Awards. He has spent more than a decade in overseas ad monetization — following web novels overseas in 2020, and short dramas in 2023. Roughly 80% of overseas short-drama platforms are his partners.
His verdict was blunt: the overseas market is not as big as most people think.
The roundtable was titled “From Made-in-China to Global Revenue.” Three people on stage, each sitting at a different point of the chain: Peter handles monetization, AhaCreator’s Susan handles influencer marketing, and Qingnai Technology’s Ella handles AI-powered commerce content.
Totally different business models. But by the end, all three were saying the same thing: AI has flattened the barrier to “making things,” and the money and moats have moved outside production.
Start with Peter’s cold shower.
Why is the overseas short-drama market small? His explanation is interesting: one of the roots of overseas web novels and short dramas is the fan fiction community — Twilight and Fifty Shades of Grey both grew out of it.
That created a genre path dependency. Overseas audiences consume werewolves, vampires, and fantasy — a handful of genres. Domestically, time travel, costume dramas, farming stories — just about anything can run.
But you might ask: if the market is overestimated, where’s the opportunity?
Peter’s answer is neat: “What’s overestimated is the present. What’s underestimated is the future.”
Why do only a few genres work overseas? One reason: everything else is too expensive to produce. A normal family drama or CEO romance in North America costs $150,000–$200,000 — about RMB 1 million.
Sci-fi? Forget it. AI is bringing that cost down. Once costs fall, more genres become viable, and only then does the market’s ceiling rise.
He also dismantled a widely held bias: would short dramas starring Chinese actors flop overseas? Many assume overseas audiences only accept white actors.
But the data they see says otherwise. Dianzhong, Jiuzhou, and other leading domestic platforms translate domestic short dramas into multiple languages and distribute them overseas — and some of these titles outperform locally-shot white-actor dramas.
Why? Because short dramas are born from web-novel wish-fulfillment. The core is delivering that hit of satisfaction. If the script and story structure hold up, whether the actors are Asian or white is not decisive.
Going overseas with short dramas has been a loud story for two years, but insiders uniformly complain there’s no money in it. Peter opened the books.
A North American short drama costs around RMB 1 million to produce, with only about a 10% hit rate — and he means big hits. That doesn’t even include user acquisition: buying one North American user can cost tens of dollars.
At those numbers, ad-unlock alone can’t cover costs. That’s why the revenue mix has long been nine-to-one: 90% or more from user payments (IAP), under 10% from ads (IAA).
But starting this March, the scales moved.
Peter’s key observation: in the data they see, the paid-user-acquisition ROI of pure AI-produced short dramas is now close to that of high-cost North American live-action ones.
Content costs fell, and the chain reaction followed. The IAA share started rising — there are even pure-IAA short-drama apps on the market now. ByteDance is doing exactly that.
IAA also has a hidden advantage: users willing to pay are hard to find and expensive to acquire. Users willing to watch ads form a much larger base, with lower acquisition costs.
This shift has only been underway for six months. The top platforms still sit at 90/10 — but directionally, the old ledger no longer works.
Asked which links are overestimated and which underestimated, Qingnai Technology partner Ella answers against intuition.
What’s overestimated, she argues, is AIGC’s productivity itself.
The logic is simple: after AI became widespread, anyone can make a beautiful video. A flood of content rushes in, which paradoxically raises the bar. Brands’ ad budgets, conversion opportunities, and consumer attention all get wasted.
“A large volume of content is not itself productivity. Finding the effective content within it matters more.”
What’s underestimated is the user insight behind the content. Many material companies just put the product into a video. What’s actually valuable, she says, is dissecting why users buy and why they don’t — and creating content with hypotheses.
Qingnai walked that path itself. The company started deep in digital-human and AI technology, building up capability — but kept hitting walls when trying to land it in business scenarios.
Then they figured something out: technology has to sink into scenarios with real demand. They chose cross-border commerce content — and initially didn’t use AIGC at all. The founding team worked like “muddy-footed” operators doing traditional live-action commerce content, and reached the top of the US commerce ranking three times in a very short period.
First run commerce on the front line, then serve brands with that proven experience, then productize content capability into the LoopLab SaaS system. After those three phases, Ella’s conclusion: production is already a red ocean. Production itself is not a core competency. Content that understands users is.
One detail says it all: about 80% of Qingnai’s clients now pay by results — DAU, CPC, even GMV — rather than by material delivery.
You can only be accountable for results if you genuinely understand, not just know how to make videos.
If Ella answers “how content becomes effective,” AhaCreator co-founder and CGO Susan answers “how content gets seen.”
Overseas influencer marketing is a year or two behind China in maturity. How far behind? Building relationships, communicating, and delivering with overseas influencers is mostly done by email — a 5-in-100 reply rate counts as high in the industry.
Anyone who’s done influencer BD knows it’s like dating: you have to talk to each person specifically, get them interested in the brand, and only then will they take the order.
AhaCreator’s approach is to let AI take over 80–90% of that repetitive work: matching, outreach, negotiation, contracts, follow-ups, delivery, review, authorization — all handled by AI.
The platform now has over 5 million influencer resources, with 150,000 active influencers who have successfully taken orders on AhaCreator, and roughly 20,000 new ones added every month.
One case shows the system’s efficiency. A top-tier 3C brand wanted 1,000 UGC pieces covering overseas markets within a little over a month. But the product cost was high and they didn’t want to send samples — so they needed influencers who already owned the product.
Hand that brief to a traditional agency or MCN, and basically no one will take it.
AhaCreator’s play: the brand inputs the goal, product link, and brief. Within 48 hours, it matches three to four hundred qualified influencers, with a pass rate above 70%. Each delivers three to four pieces of content. All 1,000 pieces land in under a month, at an average CPM below $4, with no extra paid boosting.
And some influencers who rarely took orders before produced breakout hits.
But what Susan really wants to say is the part AI can’t do.
Her words: “In the AI era, content can be produced and replicated — but trust between people, and the social network behind it, cannot be replaced.”
Behind every influencer is a web of connections formed by their fans and audience. A brand doing influencer marketing first figures out its target audience, then finds the matching influencer profile, then validates the content direction — and only then scales up paid spend.
She calls her position “three blocks away from the big tech companies”: this kind of business takes six months to a year to cold-start, and even something as mundane as keeping bulk emails out of the spam folder requires trial and error. Big companies won’t prioritize it — which is exactly the gap startups can fill.
And choosing product and technology as the moat, rather than sales relationships, is deliberate: a client relationship held together by people vanishes when the person leaves. The transaction habits and data flywheel accumulated on the platform are what actually belong to you.
Near the end, the host asked Peter for the most concrete advice he could give overseas founders: which link is most easily overlooked, and which math tends to be wrong?
Peter’s answer was three words: “Stay compliant.”
Southeast Asia, North America, Africa, Latin America — every market has different rules. Global ad supply is concentrated in Google, Meta, and TikTok, which enforce policy strictly. Ad violations can escalate to account-level problems.
Content, ads, monetization, apps — every link has a compliance dimension. It’s a thousand words in three, but anyone who’s stepped on the landmine knows their weight.
Ella added another invisible cost. She and a client had cracked Vietnam; the client wanted to open Thailand the next month. The launch phase went badly.
Religion, culture, habits, aesthetics — all different. Platform rules vary by country, and some adjust weekly.
People used to assume Chinese content just works overseas, and that cracking one country means you can copy-paste to its neighbor. Reality is completely different. And Qingnai carries much of that friction cost itself, because it charges by results — no results, no payment, and it keeps investing in accounts and materials.
Her self-consolation is pragmatic: if they walk the pits with this batch of clients, the next batch pays less in trial-and-error.
Look back at the roundtable: three people at different positions of the chain, but the puzzle pieces fit.
Peter sees the monetization-end math: after AI cuts content costs, short-drama’s business model is shifting from “find people willing to pay” to “get more people willing to watch ads.”
Ella sees the content-end truth: anyone can make a video now; what’s valuable is knowing why users pay.
Susan sees the distribution-end moat: AI can compress the process to matching hundreds of influencers in 48 hours — but the web of trust between influencer and fan has to be woven thread by thread.
“From Made-in-China to Global Revenue” — what separates the two has never been production capability. AI is leveling the production end. And the flatter that end gets, the more expensive everything else becomes: monetization efficiency, user mindshare, compliance awareness, and trust between people.
Guests: Peter, co-founder of Firefly; Susan (Peng Shan), co-founder & CGO of AhaCreator; Ella, partner of Qingnai Technology.
Host: Zhao Liang, partner of Unique Capital.
Opening: introduce yourselves and your business
Zhao Liang: Welcome to the three guests. Today we’re talking about “From Made-in-China to Global Revenue,” with a focus on distribution and commercialization. The three of you come from different companies with different business models and positions in the industry chain. Please introduce yourself and your company’s main business, starting with Peter.
Peter: We mainly do commercialization — ad monetization for over a decade. Since 2020, we’ve focused on overseas web novels and short dramas. Short dramas look hot, but they haven’t been around long — about three years.
In 2020, Chinese web novels going overseas entered a rapid growth phase; in 2023, Chinese short dramas going overseas had their own explosion. We followed both waves, working with web-novel and short-drama companies to explore suitable ad-monetization models.
Global advertising has existed for decades, but the traffic patterns that web novels and short dramas create are still relatively new. We work with going-global companies to monetize while improving user experience — that’s our main business. We’ve built stable relationships with leading companies in overseas web novels and short dramas, and they’ve continued to this day.
Zhao Liang: Peter started his business right out of college, always in overseas ad monetization — a veteran of both ad monetization and going global. Looking forward to his sharing. Susan, please.
Susan: Hi everyone, I’m Susan. Let me introduce AhaCreator: we’re an AI-native bilateral trading platform for overseas influencer marketing.
When people think about overseas growth, influencer marketing is usually the first thing that comes to mind. But compared with China, overseas influencer marketing has many pain points: the commercialization maturity, willingness to take orders, and delivery performance of overseas influencers are all weaker than in China.
Many going-global AI tools and consumer brands want to do influencer marketing well, but from outreach to communication to delivery, it demands huge manpower, and scaling is hard. We’ve been polishing the product for over two years.
Our model is a bit like ride-hailing: AhaCreator helps brands find influencers and completes content delivery. The core is matching and transactions.
On one side are brands and advertisers who want to find the right influencers, validate messaging and product selling points, then amplify with influencer content and paid acquisition. On the other side are creators across platforms who want stable, reliable, clearly-specified commercial orders without wasting time. Balancing the two is what we do — mostly through AI.
Today AI handles 80–90% of the repetitive work in influencer marketing: matching, outreach, negotiation, contracts, follow-ups, content delivery, data review, and content authorization.
This has a commercial moat. We believe the biggest moat is brand collaboration data and the transaction habits both sides gradually form.
We’re happy to be at the Unique Awards today to share our experience building this matching platform, and overall experience in overseas growth and influencer marketing.
Zhao Liang: Thank you, Susan. Could you also introduce your personal background?
Susan: I have over a decade of big-tech experience. The first several years were in marketing, paid acquisition, and brand advertising. The last five years were at Pinduoduo — building the main-site commercialization, Duoduo Maicai, and Temu from 0 to 1 and then to 10.
In five years at Pinduoduo I built three teams, and I saw the shifts in platform traffic trends both at home and abroad, and in the AI industry.
I think overseas is a great market and AI is a great track. When I thought about which industries have pain points worth going deep into, my co-founder and I chose influencer marketing and built AhaCreator. It’s turned out to be a good transition.
Zhao Liang: Thank you, Susan. Ella, please.
Ella: Hi everyone, I’m Ella, partner of Qingnai Technology. Qingnai is an AI-driven growth service provider for cross-border content commerce. We’re not a traditional material supplier.
We provide AI materials, but we care more about conversion, accumulation, and batch replication. Beyond materials, we build result-oriented products and services based on client needs, including platform operations.
More and more clients now work with us on DAU-, CPC-, even GMV-oriented models. So we position ourselves as a content service provider that’s accountable for results.
We help global brands find global markets. Most of our clients today are Chinese brands, platforms, and tools going overseas. We’re also expanding to overseas brands entering other markets. We have sales colleagues in Singapore, South Korea, and the US handling global expansion.
Zhao Liang: Ella, could you briefly introduce your personal background?
Ella: I transitioned into the AI track. I’ve spent nearly 20 years in marketing; previously I led omnichannel sales at a listed company.
I think the underlying logic of cross-border AIGC is closely tied to consumer goods. What changed is how users watch and buy; the consumer pain points remain. Our team mostly comes from consumer goods, bringing that understanding of users, markets, and products into AIGC.
Q1: Why choose AI content commerce as the entry point?
Zhao Liang: My first question is for Ella. From your introduction, Qingnai looks like an AI-content-commerce company. Many AI short-drama companies are looking for a second growth curve in this direction — AI short-video commerce. There’s been a lot of experimentation domestically; going overseas is just beginning. Qingnai has been at it for a while — why did you choose this entry point?
Ella: Our journey had several phases. Early on, our team grew out of a metaverse team. We had digital-human and AI accumulation, but hit a lot of problems in the metaverse scenario.
Then we realized technology has to root itself in scenarios with real demand. So we found the going-overseas scenario — because no matter how much the medium changes, the underlying logic of supply and demand always exists. We committed to going deep on commerce content.
And we didn’t start with AIGC. We entered with a very traditional approach — live-action commerce content.
The founding team worked like “muddy-footed” operators, studying the underlying logic on the front line. Phase one of Qingnai was essentially starting as a traditional commerce merchant — and we got great results, reaching the top of the US commerce ranking three times in a short period.
Then we asked how to serve more brands with that validated experience. Starting this year, some big brands chose to work with us, and we helped them achieve verifiable, iterable results. That formed our current business model — from materials to operations to pay-by-results. That’s phase two.
Now we’re in phase three: productizing content capability. Production is already a red ocean, and we don’t believe production itself is a core competency — content is. So we built a SaaS system called LoopLab, currently in beta, that structures user mindshare and binds content to it, then iterates batch by batch to find which content works, why it works, and what to do next — a flywheel. Those are our three phases.
On your point about short dramas moving into commerce: I think the destination may be the same, but the paths differ. It’s like “people finding goods” versus “goods finding people.” Some teams have great goods, content, and positioning, and go looking for matching products and brands. We start from client demand and then form content products. The starting points differ, but the core logic is the same. What determines success underneath is always who understands users better — who can put the product and the demand in front of the user, move them, and close the deal.
Q2: How the short-drama overseas ledger changed this year
Zhao Liang: Many short-drama companies now believe “content itself is the shelf” and want to try AI content commerce — we’re seeing a lot of active attempts and expect positive market feedback in the next month or two.
Peter, you mentioned that about 80% of overseas short-drama platforms work with you. On the commercialization and ad-monetization side, you must feel the industry changes keenly. How did you get into short-drama commercialization?
Peter: The short-drama product is only a few years old and relatively new. It initially relied on user payments, and that’s still the dominant model. Another common model is watching ads to unlock: users who don’t want to pay can watch an ad to unlock episodes.
But there’s a problem: an episode is usually one to two minutes, while an ad can be up to a minute long. And many ads have poor experiences — the close button is hard to find.
So we thought: watching an ad to unlock one episode can take far more time than expected. The experience isn’t great.
We wanted to try something new: could a simple interaction — reading an article, or playing a light mini-game — supplement the standard rewarded-video ad? That improves user experience and their acceptance of the short-drama app. That was our original intent: a new way to work with short dramas and give users a new ad-monetization experience.
Zhao Liang: Overseas short-drama platform rankings keep shifting, and everyone complains it’s getting less profitable. Content companies spend aggressively on acquisition, costs rise, and user payments seem to decline. From your view, what’s the market trend? And you mentioned IAA and IAP — is there an ideal ratio for platforms?
Peter: One or two years ago, short dramas mainly monetized through IAP. I think it’s still IAP-dominated, mainly because production costs are high. A North American short drama costs about $150,000–$200,000 — roughly RMB 1 million.
With costs that high, hit rates are low — many dramas can’t even buy enough traffic. The big-hit rate is maybe 10%. I mean big hits. At those production costs, ads alone can’t cover them — you need user payments.
That’s before acquisition: buying a North American user can cost tens of dollars. User acquisition is expensive, so payments remain the main monetization. Only users who truly love short dramas and pay long-term can cover overall costs.
But this year things changed: AI sharply cut content costs. Since around March, in the data we see, pure AI-produced short dramas’ acquisition ROI has approached that of high-cost North American live-action ones.
As content costs fell, the IAA share started rising. The previous revenue structure was about nine-to-one: 90% or more IAP, under 10% IAA. But starting this year, pure-IAA short dramas have appeared — ByteDance’s is pure IAA.
After content costs fall, pure IAA can earn positive ROI. Also, when IAA is the main monetization, acquisition costs can be lower: users willing to pay are harder to find and costlier; users willing to watch ads form a much bigger base with lower acquisition costs.
This change started this year and has only been around six months. The IAA share is rising.
Zhao Liang: So there’s no new fixed ratio yet — it was about nine-to-one before, and we keep observing?
Peter: Right now, top platforms are probably still IAP-dominant, close to 90/10. But there are also brand-new apps with zero IAP, pure IAA.
Q3: How much of influencer marketing can AI take over?
Zhao Liang: AhaCreator has over 5 million influencer resources and about 120,000 active influencers — quite large for a startup. AI handles matching, outreach, negotiation, and follow-ups. It sounds complex and hard. Why did you choose this model where AI takes over so much of the process? And what’s its main value to the ecosystem?
Susan: A correction first: we now have 150,000 active influencers, adding about 20,000 a month.
Zhao Liang: That’s about 20% growth a month?
Susan: Yes. Why choose this hard path? Over the past two-plus years we’ve gotten positive feedback from the capital market, and client satisfaction feedback is above 95%.
The track itself — overseas influencer marketing — has a big problem: overall commercialization maturity is weak, maybe a year or two behind China. Whether mid-tier or top influencers, outreach, communication, and delivery are mostly done by email. When reaching out to a new influencer, 100 emails might get five replies — and that counts as a high reply rate in the industry.
Many going-global brands’ first growth step is influencer marketing. They can’t just throw paid spend around; they need influencers to validate how the product should be expressed and what its selling points are.
Another point: in the AI era, content can be produced and replicated, but trust between people and the social network behind it cannot be replaced. Behind every influencer is a social network of fans and audience.
A brand first identifies its TA — target audience. Second, finds the influencer types and profiles matching that TA. Third, decides what message to send through influencers. Only after validating that content and market direction match can they scale paid spend.
Whether scaling mid-tier influencers or working with top ones, there are clear pain points. Example: we served a top-tier 3C brand that wanted 1,000 UGC pieces published within a little over a month across overseas markets. The product cost was high and they didn’t want to send samples, so they needed influencers who already owned the product to create the content.
Traditional overseas agencies or MCNs get headaches with such briefs. Delivering over a thousand pieces of authentic content in a month is hard — a good fit for AI.
The brand inputs the goal, product link, and brief. Within 48 hours we match three to four hundred qualified influencers, with a pass rate above 70%. Each delivers three to four pieces. The full delivery lands in under a month, lifestyle-oriented, showing product selling points naturally. The core is scale — and some influencers who rarely took orders but genuinely liked the product produced breakout hits.
The whole campaign delivered in about a month at an average CPM under $4. From these numbers the delivery works well, with no extra boosting or paid spend. This case shows the pain of manual outreach and filtering: doing it person by person, finding influencers who own the product, would be extremely costly.
We call ourselves “three blocks away from the big tech companies” because big companies have higher-priority, faster-ROI businesses. I spent over a decade in big tech — I know such projects take six months to a year to cold-start. Early on, even keeping bulk emails from getting blocked or spam-foldered takes trial and error. There are also complex workflows and platform rules — how to contact influencers to raise reply rates.
Each influencer needs targeted communication. Anyone who’s done influencer BD knows it’s like dating: you have to talk to each one specifically, get them interested in the brand, before they’ll work with you.
So why this path? The core reason is growth. Influencer marketing is a key part of growth — often the first step — but it’s hard. Overseas, there wasn’t yet a particularly leading AI-native company in this track.
Second, big tech probably won’t prioritize it in the near term. Third, we want product and technology to be the biggest moat, not sales relationships. Throwing enough people might work, but it’s hard to form a data flywheel. A client relationship held together by people breaks when the person leaves. Products need continuous iteration.
So we want to operate like a platform — form transaction habits and a transaction network, and let the platform accumulate irreplaceable infrastructure capability. Influencer marketing involves a thousand small things; we try to abstract and standardize as many as possible. Some brand needs are non-standard; what can be standardized, we standardize.
Next we’ll keep iterating — stronger attribution and performance capabilities.
Q4: Which links are overestimated, which underestimated?
Zhao Liang: Next, Peter and Ella, please answer a common question related to the theme. In the chain from Made-in-China to Global Revenue, which links are overestimated, which underestimated? And what friction costs — the things not written into contracts but that consume your time and energy every day? Please speak from your business practice.
Peter: On overestimation and underestimation, a few observations. We know web novels and short dramas going overseas well. First, a few numbers. My conclusion: the current overseas market is not as big as many think.
I checked public data — Hongguo Short Drama’s DAU is around 170 million.
Zhao Liang: That number is debatable.
Peter: The exact number is debatable, but it’s at least the 100–200 million scale. Chinese web-novel users are maybe close to 700–800 million — six or seven in ten people read novels. But add up the entire non-China global market, and the user base isn’t that large.
The leading overseas short-drama app’s DAU probably doesn’t reach 10 million. Web-novel users may be even smaller. I think it relates to how overseas content developed.
It goes back to the origins of overseas web novels and short dramas — one root is the fan fiction community. Communities like that produced familiar IPs like Twilight and Fifty Shades of Grey.
That forms part of the base of overseas web novels and creates path dependency in genres. Only a handful of genres have high overseas acceptance — werewolves, vampires, some fantasy. You can see it in Twilight and Fifty Shades of Grey. In China, users accept far more genres — time travel, costume dramas, farming stories, and so on.
Currently the genres popular overseas are limited, and the market isn’t as big as imagined. That’s a problem Chinese content faces going overseas — but also, I think, a pretty big opportunity.
Zhao Liang: So you think on the content side, the overseas market is overestimated.
Peter: Yes, the current market is overestimated. But that also means future opportunity. Why do only a few genres run overseas and others haven’t taken off? One reason is they weren’t economical to produce as short dramas. Take fantasy. A normal family drama or CEO short drama already costs $150,000–$200,000 overseas; sci-fi would cost far more.
AI may solve that cost problem. So AI could expand the range of genres that can be produced and accepted overseas. The future growth space, I think, is actually underestimated.
One common question: will short dramas starring Chinese actors flop overseas? Some assume overseas audiences only watch white actors — but our data doesn’t support such a simple judgment.
Short dramas are born from wish-fulfillment novels; the core is delivering that “hit” of satisfaction. As long as the script and story structure are sound, whether actors are Asian or white may not be decisive. From the data we see, the difference isn’t as obvious as imagined. Dianzhong, Jiuzhou, and other top domestic short-drama platforms translate domestic dramas into multiple languages and distribute overseas — with solid numbers, some even exceeding white-actor dramas shot overseas. That’s what I wanted to share.
Zhao Liang: Thank you, Peter. Ella, your view?
Ella: I’ll be brief. My view differs a bit from Peter’s. I think what’s overestimated is AIGC’s productivity.
Now that AI is widespread, everyone can make decent-looking videos fairly easily. But the flood of videos also raises the bar. Brands’ ad budgets, conversion opportunities, and consumer attention can all be wasted.
So: a large volume of content is not itself productivity. Finding the effective content matters more. That’s the first point.
The second underestimated thing is content. I don’t just mean making a video and putting the product in it. I think you need to dissect why users buy and why they don’t, create content with a set of hypotheses, and think about how to move the user. Many material companies don’t invest enough time and manpower in these questions.
As for friction costs — honestly, it’s a bit painful. Frankly, we used to assume Chinese content just works overseas, and if you crack one country, the neighbor works too. Our actual experience says otherwise.
Example: we cracked Vietnam with a client. He wanted to open Thailand the next month. But when we served the Thai market, the launch phase went very badly.
Thai users’ religion, culture, habits, and aesthetics differ from Vietnam’s. Platform rules differ by country, and some adjust weekly.
So when we enter a new category or track, we have to walk the local-market adaptation together with the client. We carry a large part of that friction cost, because we’re accountable for results — about 80% of clients don’t pay by material delivery. To hit results we may invest more accounts, materials, and other costs.
And we tell ourselves: walk the pits with this batch of clients, and the next batch pays less in trial-and-error.
Final two questions: platform and country selection, and which math is easiest to get wrong
Zhao Liang: That’s the price of entrepreneurship. Next, Susan — influencer marketing is naturally cross-platform: TikTok, Instagram, YouTube, and so on. Users and rules differ by platform. For Chinese startups going overseas, which platform should they choose first? How should they adapt to platform rules and design strategy? And what are the differences between North America, Japan/Korea, and Southeast Asia?
Susan: Let’s split it into platform and country. Looking at the overseas platform ecosystem: short video is mainly TikTok and Instagram; long video mainly YouTube; there’s also X, LinkedIn, Reddit, etc.
On platforms: content-consumption habits differ a lot. TikTok leans entertainment, with a higher share of algorithmic-recommendation traffic. Its CPM or traffic cost is relatively low — good for fast volume and highly entertaining content. Instagram has existed longer and is more style-oriented — good for short video with a certain tone. YouTube suits reviews and tutorials, and can feed back into GEO effects, but collaboration costs are relatively high.
To make good content with an influencer, the brand needs to clearly communicate strategy, brief, and delivery requirements.
On countries: consumers in mature markets like the US and Europe already have some awareness. Mature markets are more competitive, but you don’t have to educate consumers from zero. Emerging niche markets also have opportunities, but need more trial and error.
An example from our clients — a key partner at this conference, SeaArt. They rolled out many markets from 0 to 1; the first to explode wasn’t a major English-speaking market but Brazil’s Portuguese-speaking market. They built the product through influencer marketing and community, taking a differentiation route — getting content creators into the community fast and locking in user retention first. It’s a good market-selection case.
In their influencer marketing, the core is making users feel immersed. On platforms, they combine multiple ones. Influencer types include reviews and tutorials — possibly better for YouTube — plus showcases, i.e. product demonstrations. Showcase content shows how to make something with the product, spread more on short-video platforms, paired with long video, and combined with hot topics in different markets and periods. Then they capture trends with reviews or showcases and form a continuously iterating market strategy. Their cooperation with us is deep — that’s my view on platform and country selection.
Zhao Liang: Thank you, Susan. For time, the last question goes to Peter. Give going-overseas founders one concrete, executable piece of advice: which link in the chain is most easily overlooked, which clauses tend to be signed wrong, which math tends to be wrong?
Peter: The first thing I want to say is policy — compliance. Southeast Asia, North America, Africa, Latin America — rules differ. It’s easy to overlook and easy to trip on, especially in ad monetization.
Global ad supply concentrates in Google, Meta, TikTok, and similar platforms, and they enforce policy strictly. Overseas rules differ from domestic ones, and ad violations can have serious consequences, including account-level problems.
Zhao Liang: A thousand words in one sentence.
Peter: Stay compliant. Yes.
Zhao Liang: Pay attention to compliance, follow the rules. Keep that firmly in mind.
Peter: Whether it’s content, ads, monetization, or apps, compliance matters.
Zhao Liang: Thank you, Peter. For time, this panel concludes here. Thanks to the three guests for sharing.