XPENG Turns Crisis PR Into Influencer Content
A breakdown of how the Chinese EV brand XPENG turned customer complaint into content—and what real crisis and safety comms would look like
Since I wrote this, XPENG has deleted both the LinkedIn posts and videos I mentioned below. That, in itself, is a case study in what not to do. When a CEO has already posted something publicly, pulling it down later doesn’t erase the message — it just makes you look like you’re trying to walk it back. If it’s out in public, deleting it from your own channels is about as effective as hiding a screenshot-era mistake.
When I first saw Chinese EV brand XPENG CEO He Xiaopeng’s LinkedIn post about flying an unhappy customer from Australia to Guangzhou, I thought: this is exactly what happens when crisis PR is reimagined as influencer content.
“When I learned about Vanessa’s unpleasant experience at one of our stores, I was determined to invite her to Guangzhou, China from Australia. This time, I stepped in as her sales consultant to create a brand-new, exclusive experience for her personally.”
On paper, it sounds heartfelt and almost humbling. But the video tells a different story.
A “dissatisfied” customer from Australia, Vanessa, is flown all the way to Guangzhou. The CEO personally greets her. Cameras are already rolling. She’s all smiles. There is an entourage that looks like press, XPENG employees, and a polished production setup.
Vanessa also happens to be a fashion blogger, which means she fits neatly into the “influencer for hire” template. On camera, she looks slightly nervous, but goes along in that eager-to-please way you see when someone knows they are part of a brand shoot. She mentions that she owns a Tesla, but cheerfully adds that XPENG’s full self-driving feature is included, “which is great!”
She then asks what should have been a queue-up question: “How many stores do you have globally?”
That is where the mask slips a little. Instead of answering directly, the edit cuts to b-roll of the XPENG model gliding down the road. The CEO’s voiceover in Chinese (he had been speaking English up to this point) comes in, boasting that XPENG is number six in the global sales ranking and that the G6 accounts for most of those sales. The shot snaps back to the CEO as he speaks to (probably) his colleague.
From there, the “apology tour” unravels and drifts over to flying cars, XPENG’s IRON robot (like, as in Iron Man?), and the high-tech showroom XPENG really wants the audience to see. The CEO never actually answers the question about store count.
The Comms Playbook that wasn’t
The original “unpleasant experience” barely exists in the video narrative. It functions as a pretext, a hook. The story is no longer “we heard you and fixed it,” but “look how impressive our ecosystem is.”
Founder-led communications are all the rage these days, but in the XPENG CEO’s case, this is almost a textbook example of how not to do it:
The hero of the story is the CEO, not the customer.
The arc is not “here’s what went wrong and how we fixed it,” which is how a brand builds trust with the most critical audience: potential customers watching.
The tone is not one of accountability, but a self-promotion show-and-tell with a human prop.
XPENG in Australia
All of this would be strange enough if it were just a one-off human interest story. But the context matters. Why a customer from Australia specifically?
Australia is not a random sideshow for Chinese EV brands. It’s a right-hand-drive, high-income, safety-conscious market where Chinese brands like BYD and MG are already gaining share and trying to prove they can match or surpass Japanese and European incumbents. For XPENG, Australia is supposed to be a credibility market: a place to demonstrate that the brand can deliver reliable cars, functioning after-sales service, and basic service in an English-speaking environment, to gear up for other right-hand-drive markets like the UK and Singapore.
On Whirlpool, one of Australia’s big tech and EV forums, the comments on XPENG’s local rollout are not about flying cars. They are about basic execution: long waits for cars, confusion about delivery timelines, lack of clarity on parts and servicing, and a messy website.
One comment calls out the problem:
“XPENG has made a lot of announcements, but doesn’t seem to be too good at following through on the basics.”
It’s a brutal line because it captures the asymmetry so cleanly. On LinkedIn, XPENG is showing off a CEO flying customers across continents, humanoid robots, and dramatic robot demos. On forums, real customers are asking whether spare parts are available in-country and why their car hasn’t arrived yet. And XPENG has been courting Australian consumers with incentives like lowering deposits, and offering zero-percent financing or cashback.
Which leads to an obvious question that the LinkedIn post never acknowledges: Out of all the customers in Australia who have complained, how did XPENG pick Vanessa? By the end of the video, I still don’t know what her complaint is. How was she chosen? To be the one flown to Guangzhou, given the CEO treatment, and turned into content.
Was she picked because her case was the most serious, or because she was the most on-brand—a camera-friendly, English-speaking fashion blogger whose presence reinforces XPENG’s aspirational image in Western markets?
Good crisis work, when something actually goes wrong, is boring on camera. It is about listening, acknowledging the problem, and explaining in concrete terms what broke in the system. The company shows what has changed for everyone, not just for one person flown in on a sponsored trip. It empowers frontline staff, changes incentives and processes, and monitors whether the fixes work.
“Blind driving” without a real third party
Less than a day after the Guangzhou–Vanessa reel, XPENG’s CEO posted a second video, this time to showcase XPENG’s AI driving capabilities.
The video shows a car with its windows fully covered, running what is described as a “blind driving” test. A host—clearly not Chinese, not once named, positioned as an international voice—narrates the experience and jokes about his “dangerous” job riding along. There’s a playful dig at Tesla, and he says on camera that Xpeng’s system is “probably better than Tesla’s FSD.”
On the surface, this looks like a third-party endorsement: look, a foreign host is saying our system is safer and better than Tesla’s. But again, the fundamental questions any serious viewer would ask are never addressed:
Who is this host? What is his background?
Why should we care about his opinion?
This “international host explainer video” format only works if the person comes in with independent credibility: a CNBC anchor, a BBC correspondent, a well-known safety tester, or a YouTuber serving as a trusted third-party voice.
The test itself is conducted in a closed, controlled area on a track Xpeng has built, with prearranged obstacles laid out in front of the vehicle. As a proof-of-concept demo, that setup is reasonable; companies have to start somewhere. But the way it is framed (“100% safety,” “full confidence”) asks the audience to make a leap from controlled environment to real-world safety without evidence.
Even the “dangerous job” joke undermines the seriousness of what they are supposedly proving. If the whole point is that the system is trustworthy, why is everyone still half-treating it like a stunt?
Earned influence vs. paid influence
Both posts share the same underlying confusion: mistaking paid or controlled influence for earned authority.
Hiring influencers to do a segment about your AI driving safety is, structurally, an advertisement. Putting a fashion blogger who had a bad experience into a glossy “VIP redemption” storyline is also, structurally, an advertisement. An ad is not PR.
There is nothing wrong with paid or staged content. Brands do it every day. The problem comes when companies expect these pieces to be received as fair and objective third-party opinions—as if the audience will treat the influencer like a neutral journalist or a safety authority.
They won’t. People see the “hard sell” and recognize when a “host” is there to deliver talking points rather than to interrogate claims. They notice when, at the end of the video, he is essentially reading out a recap that sounds like the brand deck.
Earned influence works differently:
In crisis: you earn trust by being specific about what went wrong and what you changed.
In product and safety: you earn trust by working with credible third parties—engineers, testing agencies, researchers, skeptical reviewers—and letting them design or at least seriously interrogate the tests, even if the results are not purely flattering.
Paid influence can create awareness, reach, and a short-term boost. But it is not the strategy for building a long-term reputation. The moment the audience senses that what they are seeing is over-produced, over-managed, and under-explained, the credibility falls off a cliff.
The algorithm vs. the audience
If these two LinkedIn posts are XPENG’s way of “getting a head start” and being aggressive in external comms—one emotional, one technical—they should probably rethink how they deliver the strategy.
Looking at the comment sections reveals a familiar pattern: waves of positive but hollow support, the kind of generic praise that reads more like people trying to ingratiate themselves with the CEO than offering genuine engagement. That is not a measure of real persuasion; it is a measure of internal power dynamics and LinkedIn politeness.
The danger for any company, not just XPENG, is that leadership begins optimizing for the wrong audience. The stories are crafted to play well inside the company and within the executive’s own social graph—the visionary CEO, the fearless AI, the human-like robot—rather than to build durable trust with the people whose lives and safety are actually affected by the products. Even Chinese media 36Kr has called on Chinese EV brands, including XPENG, to stop “avoiding problems” in a piece aptly titled: “Domestic brands are still one 'recall' away from foreign brands.”
Earlier this month, Xpeng showcased a new humanoid robot and discussed its self-developed Turing AI chips. The rebrand as an “AI company” is timely. The humanoid robot’s eerily human-like gait went viral, and Xpeng even cut it up live on stage to prove there was no real person inside.
XPENG Co-President Brian Gu told CNBC that the “company has been developing some technology before Tesla but has not been as vocal in promoting it.”
It sure is now.
I'm increasingly enjoying your coverage, videos, everything.
Xpeng's service is slow. Good cars though.