Original · Unique Research · 2026-05-15
Editor's note: The first-person report and its judgments belong to the original Chinese author. This English rendition retains the opening essay, the full interview narrative, and the complete Q1–Q10 selected Q&A. Company scale, store-count, case-study performance, and future-projection figures are source or speaker claims, not independently audited findings. Product names (Qiyu Xinlian, Xinlian Video, Xinlian Streamer, Xinlian Steward, Xinlian DataHub, Xinlian SmartBid, Xinlian SmartCheck) and company names (Qiyu Digital / 企域数科) are working English renderings where official English forms remain unverified. Platform names (Douyin, Kuaishou, Video Account, Xiaohongshu) are retained in common English usage. The term "4S店" is rendered as "4S store," the standard Chinese dealership format. Monetary figures without a currency in the original remain without an inferred currency. The source is dated May 15, 2026.
Unique Awards · Guest Interview
Selling Cars Is Being Redone by AI
In conversation with Shen Yang, founder of Qiyu Digital: Auto marketing doesn't lack tools—it lacks a new-media customer-acquisition system that actually runs
"
In the past, selling cars relied on ad spend, on channels, on waiting for customers to walk in;
now it's about managing people—on intake, on response, on retaining users for the long term.
Over the past few years, many auto dealerships have shared a common anxiety.
The store is still there, the cars are still there, the salespeople are still there—but the foot traffic isn't what it used to be.
In the past, when a user wanted to buy a car, they might first come to the store, look around, be received by a sales consultant, take a test drive, negotiate price, and move forward gradually. Now the user's first stop is often not the 4S store, but Douyin, Xiaohongshu, Kuaishou, or Video Account.
They might scroll past a short video and pause for a few seconds; might enter a live stream and ask, "What's the discount right now?"; might leave a comment saying, "Is this car suitable for family use?"; or might click into private messages and tentatively ask, "Do you have any cars in stock?"
The problem is, many dealerships haven't truly caught these actions.
Short videos get made, but posting often goes dark.
Live streams get broadcast, but it's hard to keep them stable long-term.
Comments and private messages get replies, but at night, on weekends, during peak hours, things always slip through.
When leads come in, who follows up? When do they follow up? How far do they take it? More often than not, it still depends on the salesperson's initiative, WeChat group reminders, and manual spreadsheet tracking.
So the hardest part of auto marketing today is not simply "traffic got expensive" or "users stopped buying cars."
The real change is: users have already migrated to new-media scenarios, but the organizational structure of many automakers and dealers is still stuck in the past.
Shen Yang, founder of Qiyu Digital, put this change very directly in an interview with Unique Research:
In the past, selling cars relied on ad spend, on channels, on waiting for customers to walk in; now it's about managing people—on intake, on response, on retaining users for the long term.
The auto industry is not talking about digitalization for the first time.
In the past, when many automakers did digitalization, the core actions were building systems, entering data, and defining processes. Systems certainly have value, but a system itself doesn't proactively do work. It needs people to click, people to fill in, people to run the process.
At the AI stage, Shen Yang believes the most critical change is that systems are shifting from "people-driven" to "AI-automatically-driven."
In the past, people carried the system to get work done.
Now AI carries the system and the people together to deliver results.
This is also the position Qiyu Xinlian wants to enter.
It's not adding another video-editing tool, another live-streaming tool, another customer-service bot to the dealership. Instead, around "AI + new media," it's building a matrix of AI digital employees for automotive new-media customer acquisition.
This set of digital employees covers six links: content production, live-stream customer acquisition, lead intake, data management, ad-optimization, and compliance risk control.
In other words, what it does is not point-to-point efficiency improvement for a single tool, but reconnecting the segments of the automotive new-media customer-acquisition chain that are most likely to break.
Why does auto marketing need to be redone by AI?
If he didn't use the official introduction, how would Shen Yang explain what Qiyu is doing now?
His answer is very plain.
Qiyu Xinlian mainly focuses on the vertical track of auto marketing, helping traditional OEMs, 4S stores, and new-energy vehicle brands use AI agents to do customer-acquisition marketing on new-media platforms such as Douyin, Kuaishou, Video Account, and Xiaohongshu.
According to Shen Yang, Qiyu currently cooperates with nearly 100 auto brands, and nearly 20,000 4S stores nationwide are working with it. The main functional scenarios are helping OEMs and dealers manage new-media accounts, or operating them and obtaining sales leads on a performance-delivery basis.
There are a few industry backgrounds that must be clarified.
The first change is that offline natural traffic has decreased.
In the past, if a store had a good location, a strong brand, and diligent salespeople, the business had a baseline. But now the user's decision-making entry point has moved forward. Browsing cars, comparing cars, asking prices, leaving contact information—many of these actions happen on content platforms and in live streams.
The second change is that automakers and 4S stores generally lack new-media operations capability.
It's not that people don't know short videos and live streams are important; it's that many stores genuinely can't get them off the ground.
Hire someone who understands short videos, and they may not understand cars.
Hire someone who understands cars, and they may not understand content.
Even if you find the right person, they're likely to leave because the compensation system, work pace, and growth space don't match.
The third change is that the traditional staffing model doesn't fit the new-media rhythm.
New media is not a nine-to-six business.
Users might ask about cars in a live stream at 10 PM, might consult in the comment section on weekends, might scroll past a short video at midnight and click into private messages. If a store relies only on manual intake, leads will inevitably be missed.
Shen Yang said that what should be most reconstructed in AI-era auto marketing is not the exposure at the very front, nor the final transaction at the very end, but the full-chain intake and conversion from when new-media traffic comes in to before the final deal.
In the past, this chain relied entirely on people, so there were missed leads, slow responses, chaotic follow-up, and lack of control.
What AI needs to do is rerun response, screening, nurturing, distribution, and follow-up, so that after traffic comes in, it is not wasted as much as possible.
This is also where many automakers most easily go off track when talking about AI marketing.
Some treat AI as a content generator, thinking that posting more videos will solve the growth problem. The result: views come in, but leads don't.
Some treat AI as an auto-reply tool, thinking that being able to answer messages is enough. The result: after the user asks one question, no one follows through.
Others treat AI as an efficiency tool for a single department—faster editing, faster reporting, faster scripts—but the closed loop from customer acquisition to transaction isn't connected.
Shen Yang's judgment is that true AI marketing is not a small tool, but using AI to fill staffing gaps and build a replicable, scalable new-media customer-acquisition and conversion system that can be directly accountable for results.
The most important word in that sentence is not "AI" but "system."
Not one more tool, but one more set of collaborative digital employees
The problems of auto marketing in the past are much like the pitfalls many enterprises stepped into when doing digitalization:
Lots of systems, lots of data, lots of tools, but the business still doesn't run smoothly.
Video editing is one set of tools.
Live streaming is one set of tools.
Comment and private-message intake is one set of tools.
Ad placement is one set of tools.
Reporting is yet another set of tools.
Each tool can solve a little bit of the problem, but what truly determines conversion is precisely the connection between tools.
A short video brings exposure; the user clicks into the homepage, watches a live stream, asks a price in the comments, enters private messages, and leaves a purchase intention.
At that point, who catches it?
After catching it, how do you judge the intention?
Who gets the high-intent leads?
Did the salesperson follow up promptly?
Did they come to the store afterward?
Did this lead ultimately convert?
If not, at which step did it drop off?
This is not something a video-editing software or auto-reply tool can solve.
Xinlian AI's approach is to break the key links of automotive new-media customer acquisition into six categories of AI digital employees: Xinlian Video, Xinlian Streamer, Xinlian Steward, Xinlian DataHub, Xinlian SmartBid, Xinlian SmartCheck.
They correspond respectively to content production, live-stream customer acquisition, lead intake, data management, ad-optimization, and compliance risk control.
Xinlian Video solves the content-capacity problem.
The most common problems for dealerships are "don't know what to shoot," "shoot too slowly," and "what we shoot nobody watches." Xinlian Video can automatically complete scripts, materials, voiceover, narration, and editing around vehicle models, store policies, thematic directions, and volume requirements, letting dealerships more stably produce the short-video content platforms need.
Xinlian Streamer solves the live-stream normalization problem.
The difficulty of auto live streaming is not running one session, but broadcasting long-term, stably, and even during off-peak hours. AI streamers can automatically start and end streams, conduct voice explanations, interactive Q&A, and lead-capture guidance. After a human streamer goes offline, AI can take over; for stores without a streamer, AI can also handle basic live-stream customer acquisition.
Xinlian Steward solves the lead-intake problem. It is a customer-acquisition chain for new-media platforms such as Douyin, Kuaishou, Video Account, and Xiaohongshu, catching user inquiries and intentions from touchpoints including live streams, short videos, comments, bullet screens, and private messages.
Many stores don't lack traffic; what they lack is someone promptly catching traffic after it comes in.
Xinlian Steward can catch user inquiries on new-media platforms 7×24 hours, identify customer intention, guide lead capture, and grade leads, letting salespeople prioritize those more likely to convert.
Xinlian DataHub solves the data-fragmentation problem.
For OEMs and dealer groups, the most painful thing is not whether a single store is doing new media, but that headquarters can't see the whole picture. Which region is doing well, which store is falling behind, which account is effective, which live stream has poor conversion—without unified data, management easily becomes gut-feeling-based.
Xinlian DataHub adapts to multi-level architectures such as headquarters, brand, war zone, store, and account, aggregating data from platforms including Douyin, Kuaishou, Video Account, and Xiaohongshu, making processes trackable and results reviewable.
Xinlian SmartBid solves the ad-efficiency problem.
Automotive new-media ad placement is not simply adding budget. Where the budget goes, which plan should get more, which plan should be stopped, which creative is effective, which account converts better—all require constant judgment. Xinlian SmartBid can help teams batch-manage ad plans, monitor budget and bidding, shut down low-efficiency plans, and turn ad placement from an "experience job" into a more controllable efficiency job.
Xinlian SmartCheck solves the compliance problem.
Auto marketing can't only look at growth; it must also hold the bottom line. Random price quoting, prohibited words, fake streaming, vulgar content—all can bring brand risk. Xinlian SmartCheck can detect live-stream footage, scripts, price quotes, and fake streaming, output quality-inspection reports, and help OEMs and dealers control content compliance.
Looking at these six things together, you'll find that what Qiyu Xinlian truly wants to do is not replace a single position, but automate as much as possible the repetitive, standard, and high-frequency actions in the automotive new-media customer-acquisition chain.
What do people do?
People do judgment, strategy, deep communication with high-intent customers, and offline conversion and service.
What does AI do?
AI does content, live streaming, response, grading, data, ad optimization, and compliance monitoring.
This division of labor is closer to real business than simply talking about "AI replacing people."
What dealerships need most is not flashy tech, but missing a little fewer leads
Many AI products like to talk about capabilities, models, and technical architecture.
But for auto dealerships, the problems are often very specific.
Is there a video to post today?
Is there a live stream tonight?
Is someone replying when users ask about prices?
Are high-intent leads given to sales the first moment?
Is sales following up?
Can headquarters see the process?
Is there compliance risk in the live stream?
These questions don't sound sexy, but they determine whether there are ultimately leads, store visits, and deals.
In the interview, Shen Yang mentioned that after AI enters auto marketing, automakers shouldn't only look at exposure, followers, and view counts—they should look at closed-loop metrics closer to monetization.
The first tier is lead effectiveness rate, first-response speed, and lead conversion rate.
The second tier is store-visit rate, test-drive rate, and deal-closing rate.
Long-term, you also need to look at customer lifetime value, repurchase, and referrals.
This is actually a very important reminder.
In the past, many new-media operators were led by the nose by view counts.
A video goes viral, and everyone gets excited.
A live stream has high viewership, and the team feels it's working.
But autos are a high-ticket, long-decision-chain industry; view count is not the endpoint, and may even be just the beginning.
What truly matters is whether the user goes from "a glance" to "a question," from "a question" to "leaving contact info," from "leaving contact info" to "visiting the store," from "visiting the store" to "closing the deal."
So Qiyu Xinlian emphasizes the full chain not to inflate the concept, but because auto-marketing results are inherently not determined by a single point.
In one automotive new-media operations project, Qiyu connected more than 1,400 matrix new-media accounts,integrate data from platforms including Douyin, Video Account, and Xiaohongshu; matrix lead-capture rate rose to over 70%, AI achieved 7×24 real-time response, 3-minute rapid-response rate reached over 97%; live-stream sessions increased 115%, live-stream duration grew 193%, lead cost dropped 22%; low-quality violation session share fell to 1.05%, suspected fake-streaming behavior was cleared to zero in a single month; and each month it additionally contributed over 10,000 purely incremental high-quality leads.
These numbers shouldn't be simply understood as "AI is amazing."
More accurately, they illustrate one thing:
When content, live streaming, lead intake, data, ad placement, and compliance are connected, the losses originally scattered across the chain genuinely have a chance to be pressed down little by little.
Missing one fewer lead, responding a few minutes faster, cutting one ineffective ad placement, adding one nighttime intake, reducing one compliance risk—at a single store these may be small changes; across a matrix of hundreds or thousands of accounts, they become very large operating differences.
OEMs, dealers, and stores don't face the same problem
The complexity of auto marketing lies in the fact that OEMs, dealer groups, and individual stores don't have exactly the same goals.
OEMs care more about standards, efficiency, risk, and global management.
They need to ensure that information communicated externally by different regions, stores, and accounts doesn't go off track; they need to know where the budget is spent, where leads come from, which region is effective; and they need to prevent risks such as random price quoting, non-compliant scripts, and fake streaming.
Dealer groups care more about people-efficiency and capacity.
They need to get more stores moving, make live streaming and short videos no longer rely on a few outstanding employees, make lead distribution more reasonable, and let weak stores also achieve a baseline.
Individual stores care most about things more direct:
Can it save people?
Can it save money?
Can it save trouble?
Can it close more deals?
If an AI system ultimately still requires the store to invest heavily in learning costs, configuration costs, and operations costs, it likely won't land.
What stores want is: I provide basic information, materials, and account permissions—can you help me get videos, live streams, leads, and reports running?
This is also why Shen Yang repeatedly emphasizes that what AI brings is not "tool increment" but "people-efficiency reconstruction."
In Qiyu Xinlian's vision, headquarters sets standards, AI handles batch generation and automatic execution, stores provide basic business information and offline conversion capability, and salespeople devote more energy to high-intent customers.
This logic is not novel, but it's very pragmatic.
Because what the auto industry truly lacks is never another pretty backend, but a way of working that different levels are all willing to use, can afford, and can keep using.
From digitalization to intelligence: where does auto marketing change next?
Shen Yang's judgment for the next three years is relatively clear.
First, AI sales assistants will become standard equipment in more stores.
Not every store can afford a complete new-media team, but every store needs someone to respond to users, identify intentions, and remind follow-ups. The value of an AI sales assistant is to connect these high-frequency, repetitive, standardized actions first.
Second, store operations will become more intelligent.
Short videos, live streams, lead intake, follow-up, and review will no longer be disconnected actions. In the future, stores may no longer ask "what to shoot today," "who's streaming tonight," "who's replying to private messages," but instead AI automatically generates tasks based on store policies, vehicle information, user feedback, and historical data.
Third, user data will truly become assetized.
In the past, what many stores called "leads" was just a phone number. After leads wereaccumulated, there was no continuous operation or re-engagement. After AI enters, user behavior, intention grades, interaction records, and follow-up节奏 can all beaccumulated and reused.
Fourth, content matrices will become more automated.
Headquarters sets the tone, AI batch-produces, stores distribute and execute, data flows back for optimization. Once this process runs through, content no longer depends entirely on someone who "can shoot videos," but becomes part of organizational capability.
But this doesn't mean people are no longer important.
On the contrary, the deeper AI goes into the business, the more people need to pull out of repetitive labor to do things that require more judgment.
For example, what audience group the brand should focus on, how to design regional promotions, how to close high-intent customers, how to get old customers to repurchase, how to improve store service.
These are not problems AI can solve alone.
What AI changes is the underlying execution method.
What people need to change is the operating perspective.
Auto marketing competition is shifting from "can you do new media" to "can you run the new chain"
If you compress Shen Yang's interview into one sentence, it's probably:
AI reconstructs auto marketing; what truly changes is not the tools, but turning the old car-selling model of "waiting for customers, relying on people, going by experience" into a user-operation model that is "proactive, standardized, replicable, and full-chain automated."
This sentence is not that flashy, but it's very real.
For a long time, the auto industry pursued traffic. Whoever spent more on ads, whoever had bigger voice, whoever had a better store location—whoever was more likely to win.
But now, traffic is of course still important, it just no longer alone determines the result.
What truly determines the result is whether, after traffic comes in, you can catch it.
The user asks a question—is someone replying?
The user hesitates—is someone following up?
The user isn't buying right now—is someone continuously operating?
Headquarters spent budget—can you see where the money went?
The store ran a live stream—can you know which sentence brought a lead?
Sales followed up with a customer—can you know what the next step should be?
These small, trivial problems used to rely on人海, experience, and sales initiative.
In the future, they will increasingly be handed to AI digital employees.
This may be the sign that auto marketing has truly entered the agent era.
Not AI writing copy that sounds more human.
Not AI generating a streamer that looks more like a real person.
But AI beginning to enter the new-media customer-acquisition process, catching those links that in the past nobody caught, couldn't catch, or didn't catch well.
For the auto industry, this is not romantic, but it's critical.
Because in存量 competition, growth often doesn't come from a single explosion, but from wasting a little less at every link.
Miss a little fewer leads.
Be a little less slow to respond.
Burn a little less ineffective budget.
Have one fewer compliance risk.
Add one more effective follow-up.
When these actions are continuously amplified, so-called AI marketing truly turns from concept into result.
Selected Interview Q&A
Q1: If you didn't use the official introduction, how would you explain to someone who doesn't know Qiyu Digital what you're doing now?
Shen Yang: Qiyu Xinlian mainly focuses on the vertical track of auto marketing, helping traditional OEMs, 4S stores, and new-energy vehicle brands use AI agents to do new-media customer-acquisition marketing on platforms such as Douyin, Kuaishou, Video Account, and Xiaohongshu. We mainly help OEMs and dealers manage new-media accounts, or operate them and obtain sales leads on a performance-delivery basis.
Q2: Why does the auto industry suddenly need AI so much?
Shen Yang: It's not that it suddenly needs AI; it's that the problems have accumulated to today, and they're increasingly hard to solve with traditional staffing.
Offline store natural traffic has decreased, and users have migrated heavily to new-media platforms; but many automakers and 4S stores lack new-media teams and continuous operations capability. Short videos, live streams, comments, private messages, lead distribution—these actions all require high frequency, timeliness, and standardization, and it's hard for people alone to stay stable long-term.
What AI can fill is precisely these repetitive, high-frequency, easy-to-miss links.
Q3: Over the past few years, what has been the biggest change for automakers and dealers?
Shen Yang: The biggest change is not expensive traffic, difficult conversion, or slow decision-making—it's that the operating logic has changed.
In the past, selling cars relied on ad spend, on channels, on waiting for customers to walk in; now it's about managing people—on intake, on response, on retaining users for the long term.
Expensive traffic, difficult conversion, long chains—these are all surface phenomena brought by this change.
Q4: If auto marketing in the past relied on ad spend, channels, and in-store sales, which link should be most reconstructed in the AI era?
Shen Yang: What should be most reconstructed is the full-chain intake and conversion link from when new-media traffic comes in to the final transaction.
In the past this area relied entirely on people, prone to missed leads, slow responses, chaotic follow-up, and uncontrollable processes. AI needs to redo the actions of response, screening, nurturing, distribution, and follow-up, so that after traffic comes in, it's not wasted as much as possible.
Q5: When automakers talk about AI marketing today, what's the easiest misconception to fall into?
Shen Yang: The easiest misconception is treating AI only as a tool.
Some companies only use AI to generate content, ending up with views but no leads; some only treat AI as auto-reply, able to answer but unable to convert; others only pursue local efficiency improvement without connecting the closed loop from customer acquisition to transaction.
True AI marketing is using AI to fill staffing gaps and build a replicable, scalable customer-acquisition and conversion system that can directly deliver results.
Q6: What is Qiyu Xinlian's core capability right now?
Shen Yang: We mainly build an AI digital-employee matrix around "AI + new media," covering six links: content production, live-stream customer acquisition, lead intake, data management, ad-optimization, and compliance risk control.
In terms of products, that's Xinlian Video, Xinlian Streamer, Xinlian Steward, Xinlian DataHub, Xinlian SmartBid, Xinlian SmartCheck.
They are not disconnected tools, but work collaboratively around the automotive new-media customer-acquisition chain.
Q7: For dealers, what is the most direct value of AI?
Shen Yang: For dealers, what AI brings is not "one more tool" but people-efficiency reconstruction.
For example, content can be produced more stably, live streaming can be more normalized, leads from comments, private messages, and bullet screens on new-media platforms can be caught more promptly, data can be automatically aggregated, ad placement can be continuously optimized, and compliance risk can also be monitored.
This way salespeople don't have to spend a lot of time on repetitive actions, but can focus their energy on high-intent customers and offline conversion.
Q8: After AI enters auto marketing, what metrics should automakers pay most attention to?
Shen Yang: In the AI era, you can't only look at exposure, followers, and view counts—you need to look at metrics that can directly affect operating results.
The first tier is lead effectiveness rate, first-response speed, and lead conversion rate.
The second tier is store-visit rate, test-drive rate, and deal-closing rate.
Long-term, you also need to pay attention to customer lifetime value, repurchase, and referrals.
The value of AI is to raise the earlier response, conversion, and store-visit metrics, ultimately lifting deal-closing and long-term customer value.
Q9: In the past, when automakers did digitalization, much of it was building systems; entering the AI stage, what is the most important change?
Shen Yang: In the past, doing digitalization, the core was building systems, entering data, defining processes. Systems needed people to click, people to use, people to run.
At the AI stage, the most essential change is: systems shift from "people-driven" to "AI-automatically-driven"; business actions shift from "manual execution" to "AI continuous optimization."
Simply put, in the past people carried the system to get work done; now AI carries the system and the people together to deliver results.
Q10: Please summarize in one sentence: what does AI reconstructing auto marketing truly change?
Shen Yang: AI reconstructs auto marketing; what truly changes is not the tools, but turning the old car-selling model of "waiting for customers, relying on people, going by experience" into a user-operation model that is "proactive, standardized, replicable, and full-chain automated."