The Industry's Growth Double Helix
Why we say efficiency is the only competitive advantage for industrial internet companies.
"Why we say efficiency is the only competitive advantage for industrial internet companies.
Too Many People Involved
"Ten days." Wang Peng, founder of new-home e-commerce platform Juli Xinfang, did the math for Beijing's real estate agency industry.
"Beijing has roughly 100,000 real estate agents. Each needs to make at least 20 effective calls per day. In about ten days, they could contact all 21 million or so permanent residents of Beijing. Yet only about one million people view properties annually, with total yearly transactions hovering between 300,000 and 400,000." Average per-capita efficiency: roughly 0.2–0.3 deals per agent per month.
Wang's calculation captures the shared reality of all traditional industries today — abysmal efficiency.
During China's high-speed economic growth, efficiency was never an existential threat for companies operating in a strong seller's market. Capital reserves and "human wave tactics" alone could fuel relentless expansion and growth.
After four decades of breakneck growth, China's economy entered a "new normal." Material and cultural products became abundant, even oversupplied. The "produce first, sell later" approach saddled companies with high inventory and long payment cycles. With people serving as the core engine of business growth, human efficiency became the bottleneck. Once companies approached the tipping point of people managing people and people managing operations, further expansion actually drove up management costs, dragged down per-capita efficiency, and degraded customer service.
How to deliver better products and services at lower cost and higher human efficiency has become the most urgent question across industries.
"In the hierarchy of value creation for companies, it's always revenue growth > cost cutting, efficiency gains > cost reduction," analyzed Xingshi Wang, MD at Source Code Capital who leads industrial internet investments. "Human efficiency naturally becomes the North Star metric for traditional industries seeking to improve."
Over the past decade, internet technology first reacted with communications, information, entertainment, transportation, and lifestyle services — dramatically raising industry-wide human efficiency and birthing internet giants from "BAT" to "TMDJ."
As these sectors stabilized, people looked to the next decade for internet technology to fuse with more traditional industries, sparking a new wave of development. The buzzword of the moment: "industrial internet."
In this sense, the term that Pony Ma popularized through his late-night Zhihu visit describes essentially a dynamic evolutionary process — using internet and other technologies to raise personnel efficiency and service quality in traditional industries. Alibaba in retail, Meituan in food service, ByteDance in media.
Thus it cannot be simply categorized as to B or to C. It encompasses all traditional industries not yet penetrated by digitalization and intelligent systems, including the still-evolving supply chains and production systems behind retail and food service — except these commercial activities mainly occur between rational economic actors: government and enterprises, enterprise and enterprise.
This is undoubtedly worth anticipating. Not merely because of the lure of "trillion-RMB market opportunities," but because the visible, tangible results of internet companies improving traditional industry efficiency speak for themselves.
Take Juli Xinfang. Since its 2014 founding, the platform has maintained average human efficiency of 2.63 deals per person per month — far exceeding industry averages. Even as its sales team expanded from a dozen people to nearly a thousand, efficiency never declined. Rare indeed, given the housing market climate in recent years.
How has Juli Xinfang sustained this high efficiency over time? When explaining the logic of Juli Xinfang's entire product system, Weng Feng, head of innovative business, gave the answer: "Effective connections oriented toward the end goal."
This is entirely different from the activity, retention, and other traffic metrics we grew familiar with over the past decade. But even then, these app data metrics we knew so well were always merely surface-level indicators in domains connected to offline economies.
Former Meituan COO Jiawei Gan, discussing how Meituan launched in new cities and subsidized growth, bluntly stated that "revenue bought with money has no value." Meituan's strategy was subsidizing the supply side, building differentiation through supply-side dominance. Even while burning cash to generate transaction volume, Meituan still put "efficiency" first. In the early group-buying wars, Meituan's business volume in Beijing, Shanghai, Guangzhou, and Shenzhen long remained second place — further evidence that scale is not the primary determinant of victory. Efficient scale is.
Recognizing the importance of human efficiency for industrial internet companies is easy. Yet since 2015, the so-called "B2B breakout year," roughly 200 industrial internet companies received funding annually. Everyone waved banners of information systems, online platforms, and data middle offices to reduce costs and raise efficiency for industries. Why didn't everyone become Alibaba or Meituan? What is the real driving force that can genuinely improve human efficiency?
With assistance from Source Code Capital, pioneers in industrial internet investment, 36Kr conducted in-depth visits to six leading industrial internet companies — IT equipment leasing platform Edianzu, new-home e-commerce platform Juli Xinfang, one-stop auto parts trading platform CassTime, used-car transaction and financial SaaS provider Car300, textile trading platform Baibu, and cross-border logistics service platform YQN Logistics. Most received their first investment in 2015, with funding rounds at Series C or beyond. Their investor rosters include top-tier funds like HSG, Source Code Capital, and DCM. Interview subjects span CEOs, CTOs, business VPs, product managers, and frontline sales staff.
Through these interviews, we distilled a development path for industrial internet companies: First build a multi-role, real-time interactive digital network for collaborative operations. Then, among the many connections in product, business, and organization that serve the end goal, use data intelligence to identify and strengthen the effective links.
Network synergy and data intelligence are the twin spirals of industrial growth.
Finding and Strengthening Effective Links in Collaborative Networks
"Without fulfillment capability, rapidly scaling users through subsidies is meaningless. To fulfill efficiently, YQN Logistics had to get into transactions." For founder Zhou Shihao, this conclusion has been validated through YQN Logistics' several transformations.
If we rewind to 2014, YQN Logistics' main products were shipping social tools. User scale, activity, and retention were metrics Zhou closely tracked, yet the company fell into crisis. Zhou contacted Source Code Capital founding partner Yi Cao through Weibo direct messages. Cao said if YQN Logistics chose to get into transactions and tackle fulfillment — "the hardest but most defensible mountain" — building standards and processes through transactions, he would invest.
Zhou then received Source Code Capital's first institutional investment. The company has since fully transformed into a trading platform with efficient fulfillment capability for every order, with revenue mainly from shipping fees and service charges. In June 2019, YQN Logistics completed a $70 million Series C led by HSG and Coatue, with Source Code Capital participating. A frontline logistics fund investor surnamed Qin believes YQN Logistics has now been "resurrected."
Along with YQN Logistics, Baibu, Edianzu, and Juli Xinfang also handle transactions. Early B2B 2.0 representative Zhaogang Group, which started with order matching, similarly completed this transformation. It seems industrial internet must become a trading platform. Yet Car300 and CassTime are exceptions — the former provides valuation and risk-control data for car owners, dealers, and financial institutions; the latter is a transaction matching platform.
Why this difference? Why did Cao insist YQN Logistics get into transactions?
Xingshi Wang analyzed: "If you understand industrial internet from a business model perspective, it's easy to get trapped in tactical questions like whether to handle transactions. In fact, tactics constantly shift based on industry characteristics — you'll always find counterexamples. The core reason YQN Logistics needed to get into transactions was to build a real-time interactive, multi-role collaborative network. In cross-border logistics, which is primarily local-service-based, information matching's connectivity and extensibility fall far short of transactions. But in CassTime's full-vehicle-parts industry, demand is mostly temporary and nationwide, so information matching alone can build a collaborative network — naturally leading to different tactical manifestations."
The story and steps of building an ecosystem network are familiar: key data nodes going online, real-time online interaction between roles, ultimately forming a digital web. What we care more about is: what kind of collaborative network can sustainably improve human efficiency?
"Everything Juli does is for effective connections." In Weng Feng's view, "effective" means ultimately achieving a transaction. Thus at every node — customer, employee, property listing — and throughout the entire internet-based home buying process from online customer acquisition to offline viewing, subscription, and online contract signing, all of Juli Xinfang's performance metrics directly tie to customer satisfaction and final contract signing.
This requires Juli Xinfang not merely to digitize customer and employee behavior, but to make customer-system and customer-employee interactions traceable and controllable, identifying and continuously strengthening effective links. The two most important links are "lead quality" and "employee behavior."
To improve lead quality, the company must control lead sources and customer intent. Juli Xinfang categorizes leads from different traffic channels and contact points by purchase intent level; uses NLP to capture keywords in conversations between customers and AI customer service or sales staff; even uses stereo channels to count conversation rounds to assess call quality... The ultimate goal is to continuously score and rate the same lead through all associated metrics, selecting higher-probability leads for sales to follow up on, improving transaction probability and personnel efficiency.
For employee behavior, rather than top-down, experience-based behavioral guidelines, Juli Xinfang uses bottom-up "machine learning." Specifically, from pre-hiring info sessions to later business execution, Juli Xinfang repeatedly conducts A/B tests and performance comparisons to distill a complete set of hard execution standards. For example: contact customers within 15 minutes of receiving a sales lead; follow up three times within one week, with at least one effective call. To verify execution and capture more precise, comprehensive data, Juli Xinfang launched proprietary phones and work numbers to aggregate and analyze business behavior data, replacing traditional location check-ins and call screenshot records.
This entire system gives Juli Xinfang not only far-above-industry per-capita transaction volume, but also a latest transaction cycle of just 15 days — faster than the industry average of 21–30 days.
For CassTime and Baibu, where industry upstream and downstream are fragmented and mainly serve long-tail demand, organizing real-time interactive multi-role collaborative networks and achieving effective connections are equally important.
Full-vehicle parts are the prescription drugs of auto parts — long-tail categories, mostly temporary purchases, with high timeliness requirements. In traditional procurement, repair shops first issue full-vehicle-parts work orders, then purchase from dispersed channels. Connection breakpoints in this process include not just supply and demand sides, but also 5–10% matching errors from non-standardized part names.
"The digital foundation for auto parts is good. It's just that among thousands of car brands and models, and between demand and supply sides, there's a lack of unified language and an efficient dispatch hub." CassTime founder & CEO Jiang Yongxing said CassTime aims to be the dispatch hub in this dispersed industry.
On one hand, CassTime connects repair shops and suppliers through its online platform, obtaining real-time quotes and inventory information. On the other, it has built an auto parts database with "Google search functionality" for precise part finding. The platform now achieves over 98.5% fulfillment rate across all categories.
Baibu, entering from textile trading connections, has gone further by extending upstream to integrate excess production capacity from textile factories. This equals接入 a new batch of roles into its collaborative network, further breaking traditional industries' linear structures and evolving into an industry ecosystem moving from "line to surface to volume."
Who Decides Which Links Are Effective?
Floodgate Capital founding partner Mike Maples once said that in a highly connected world, "things that accelerate connections become more valuable." The same holds in industrial internet. Industrial internet companies replace middlemen who profited from information asymmetry, becoming middlemen who benefit both buyers and sellers, thereby accelerating effective connections.
But we must ask: wouldn't direct buyer-seller transactions be fastest? In reality, because of supply and demand diversity, society will never become a binary structure of just buyers and sellers. And compared to direct buyer-seller relationships, middlemen communicate more frequently with both sides, better building trust. This has been fully validated in bilateral platforms like Taobao and Meituan.
In networks with at least three roles, possible links increase exponentially. Which are the effective links that accelerate connections?
How does YQN Logistics rapidly generate recommended solutions among numerous shipping routes? Edianzu has 500,000 devices in circulation — why did it dare offer deposit-free leasing to SMEs from day one? Who handles complex lease plan design and rent calculations? How can Car300's remote intelligent used-car pricing service judge true vehicle condition, provide pricing and loan recommendation amounts, with 97% accuracy?
Who makes these judgments and decisions? If still relying on people to manage people and business in the network, companies descend into chaos — because the data volume for network synergy in industry far exceeds human "computing power." This brings in industrial internet companies' other growth driver: data intelligence, where machines make decisions directly.
For Edianzu and Car300, data intelligence isn't just a bonus — it's foundational to survival.
"Enterprise production equipment leases run at least six months. During the lease, management dimensions include rent calculation, device management and maintenance, residual value assessment and recovery, and changing enterprise needs during the lease. Compared to purchasing, the management difficulty of leasing rises exponentially." Edianzu founder Pengcheng Ji said without intelligent management systems, no one could do this.
Edianzu CTO Zheng Tao gave a seemingly simple example: during a lease, if a company wants to upgrade one computer's RAM from 4GB to 8GB, this requires arranging 8GB RAM outbound, dispatching someone to install it in that computer, and returning the 4GB RAM to inventory — with device parameters, rent, and residual value calculations all changing simultaneously.
But this initially looks more like a process management system. Where data intelligence truly shines is in Edianzu's SME credit risk-control model and AI supply chain.
To surpass the service experience of selling, Edianzu adopted "deposit-free" model from launch. Ji said, "Edianzu's current customer approval rate is 90%, but bad debt rate is under 1%." After five years' development, Edianzu has achieved "6-hour delivery, 2-hour on-site repair" service standards — still improving, powered by AI product selection, AI warehouse allocation, service site selection, and intelligent detection equipment. Every link doesn't merely improve efficiency of certain human behaviors, but does what "humans" cannot do.
To make data scattered across parallel subsystems come alive, truly mining inter-data relationships to guide business decisions, Edianzu built an entire system from front-end order flow to back-end workflow, attempting to connect underlying data to build its own data middle office, making the company truly run on systems and data.
Car300 CTO Zuo Qiangxiang explained that low human efficiency in used-car assessment mainly manifests in two ways: first, cultivating mature used-car appraisers takes roughly 2–3 years; second, offline assessment requires appraisers to travel back and forth, taking at least 1–2 hours per assessment. Meanwhile, however skilled, no appraiser can master all vehicle models. And once appraisers form interest ties with dealers, enormous gray-area manipulation becomes possible.
Compared to traditional assessment relying on experience, Car300's basis comes entirely from real, real-time data and repeatedly optimized models. Car300 synchronizes data in real-time from used-car trading platforms: a vehicle's listing and delisting cycle (reflecting sales cycle and market heat), model information, price data, transaction and listing data, user purchase behavior data — ultimately arriving at a transaction price conforming to market patterns.
In the first two to three years, CEO Xu Wei and CTO Zuo Qiangxiang spent every weekend in used-car markets, figuring how to continuously iterate and optimize their comprehensive accuracy from then-80% to 97%. Car300 gradually became a big data service provider for Guazi and other auto trading platforms and financial institutions.
What Juli Xinfang big data research institute head Zhang Weishi is currently proudest of is using data insights to analyze lead quality — in short, among the same batch of leads, who is more likely to sign contracts, thereby helping sales staff improve contract signing efficiency.
The advanced form of data intelligence improving human efficiency is automated processes.
YQN Logistics CTO Wang Yu explained that YQN Logistics internally breaks cargo transport into three dimensions for process optimization: cost flow, order flow, and logistics. In order flow, document preparation takes the most time, accounting for 40–50% of single-order time; next are route selection and order tracking. Thus, digitizing and automating document flow alone can improve efficiency by at least 40%. Through system interconnection with relevant parties, using OCR for automated document entry, and system-generated intelligent route plans and order tracking, YQN Logistics has raised human efficiency to several times industry levels.
Where Will People Go?
"You've made the system so powerful — won't you not need us someday." In one interviewed employee's joke lies people's underlying concern about technological progress: will people be replaced by AI?
This article doesn't aim to predict this outcome. If we recall carefully, the entire collaborative network is essentially a network composed of people. Activated by buyers' and sellers' transaction needs, with middlemen building trust bridges and sharing potential risks — remove the people, and the network disappears too.
From the steam age to the electrical age, to the internet age and AI age, countless historical facts have verified that technological progress indeed replaces people within certain scopes, liberating most from repetitive, prolonged physical labor. But more importantly, it continuously spawns new jobs and industries.
Just as Taobao, JD.com and other e-commerce platforms not only spawned entirely new jobs like shop decorators, online store models, and e-commerce sellers, but also birthed the e-commerce agency operations industry, typified by Baozun E-commerce which listed on NASDAQ in 2015 with latest market cap around $2 billion. And the courier logistics and food delivery industries now fully integrated into daily life have become symbiotic industries with e-commerce and food delivery platforms.
What new species industrial internet development will create remains unpredictable. But what's certain is that people's responsibilities and roles in the network will change. At minimum, we've already seen that within these pioneers, employees' scope of responsibilities and work forms have changed.
YQN Logistics business VP Maggie analogizes YQN Logistics's customer service to "bank tellers" — complex calculations don't require tellers with abacuses, reducing the complexity of work content itself. More importantly, they must shoulder "emotional labor," building trust bonds with customers.
Fear of potential risk is human survival instinct; thus resisting unknown new things is people's subconscious first reaction. But the wheel of history never slowed for anyone's fear. Rather than worrying, better to consider this question: "In the current environment, what value can you create?"
Postscript
The year 2020 opened with an era's speck of dust falling on every company and every person. For industrial internet companies within collaborative networks, "sharing breath, sharing fate" feels especially real. Thus beyond "self-rescue," "mutual aid" stands out prominently among industrial internet companies.
Car300 directly participated in the "epidemic battle," developing and launching a COVID-19 confirmed patient same-itinerary query tool. The tool synchronizes and updates query information nationwide in real-time; users input date, flight/train number, and region to check if they traveled with confirmed novel coronavirus patients.
To ride out the storm with all auto industry practitioners, Car300 planned the auto industry's "Epidemic Battle Public Welfare Cloud Classroom," inviting renowned industry experts to decode post-epidemic industry trends and launch the auto industry's first comprehensive resumption class.
During this epidemic, on one hand, YQN Logistics founder & CEO Zhou Shihao led formation of an epidemic guarantee task force. Under its operation, the platform undertook multiple batches of donated materials' international logistics needs, including masks, protective suits, and other relief materials from the United States, Japan, Brazil, Russia, and other countries. On the other hand, the platform opened international logistics tools including port information, container/cargo information, shipping schedules, and truck tracking to clients free of charge, helping clients better achieve online operation and fulfillment of logistics orders.
Juli Xinfang was among the first to proactively suspend offline property viewing services, launching "online sales offices" connecting back-end systems with front-end online stores, helping developers genuinely improve online transaction capability. Home buyers can use VR, three-party video, and other technologies to view, select, and subscribe to properties online, completing the entire home buying process through one phone without leaving home. As of February 24, Juli Xinfang's online sales offices had achieved over 400 subscriptions.
Additionally, Juli Xinfang partnered with hundreds of brand developers to launch the real estate industry's first pure-online-transaction "2.22 Pick Properties Festival," offering thousands of quality developments and millions of hot-listing properties, aiming to create real estate's "Singles' Day."
Before resuming work, CassTime had already conducted real-time dynamic matching of supply and demand nationwide based on gradually reopening auto parts markets and logistics, supporting repair shops and suppliers' normal operations post-resumption. Meanwhile, CassTime arranged for finance staff to work overtime to settle all accounts with suppliers, alleviating merchants' capital pressure.
From February 25 to March 20, CassTime launched exclusive financial service policies, providing epidemic-period special financial service support for closely cooperating, creditworthy partners with excellent annual comprehensive scores. Qualified auto parts merchants could obtain monthly interest as low as 1%, credit limits up to 1 million yuan; qualified repair shops could further enjoy 3-period interest-free installment repayment on existing credit products.
For Edianzu, whose second headquarters is in Wuhan, this epidemic was undoubtedly a major test. But in CEO Pengcheng Ji's view, the more extraordinary the period, the more it tests a startup's mission, vision, and values. Internally, Edianzu urgently "redirected refurbishment center capacity to other regions," using such emergency plans to sustain front-end supply chain until work resumption. Externally, even while being hit by the epidemic, Edianzu assumed responsibility toward industry partners and society, providing free IT support services to all industry clients during the epidemic, and actively assisting a certain makeshift hospital with fixed asset management and distribution system setup.
Collaborative networks make industry upstream and downstream a tighter community. As living organisms, companies are affected by the epidemic — they cough and catch colds, and instinctively fight the virus. But as hubs in the network, they undertook social and industry responsibility in this severe challenge because they believe that moving forward together can overcome hardships and avoid greater losses.
These startups are but a microcosm of China's economy. We believe Chinese companies will defeat the epidemic, and China's economy will defeat the epidemic. After the epidemic, more great companies will grow stronger.
Learn More
