Discover more from Chinese Characteristics
Convo with precision farming startup on Chinese agritech
Cheng Biao of Zhiduomei talks about the challenges and rewards of the field
After predicting that agricultural tech would be big last year, and after the shout-out that agriculture got from the annual ‘two sessions’ meeting earlier this week, I felt like it was time for a follow-up on one of my favourite Chinese tech topics.
I've always wondered what frontline practitioners felt about the most pressing issues facing Chinese tech. I was lucky enough to be connected to a founder working on precision farming focusing on strawberries. I talked to Cheng Biao about key barriers to adoption for Chinese agricultural tech, the different types of farm customers, how automated systems improve yield outcomes and big tech entrance. Our conversation was edited for brevity and clarity.
Premium readers got my piece on Verdict on Meituan last week.
What’s your background, and how did you get interested in agricultural tech?
My entry into agriculture tech is highly unexpected. I graduated from Beijing University of Posts and Telecommunications in 1998 with a bachelor's degree in engineering and a master's degree in business administration. My first job was constructing and maintaining mobile communication networks, and in 2005 I became the general manager of China Mobile in Yunnan. The first half of my life had nothing to do with agriculture at all. In 2011, I wanted to start something and resigned from my job. My team’s first efforts were working with edge computing hardware and focusing on big data analysis.
My team and I tried different business models and industries before but weren’t successful. We finally found that there might be suitable applications for our technology in agriculture and were attracted by the ability to make a real difference in farmers’ lives.
Once you get over the fragmented nature of agriculture, it’s like other industries. We thought we could apply our domain knowledge in big data here. We also felt like that would be our competitive advantage against other agriculture industry practitioners whose understanding of data application was lagging about five to 10 years behind other internet industries. So we gave it a shot and focused on precision farming in the form of water irrigation and fertiliser control systems.
Right now, we’re doing well. Our company Zhiduomei (智多莓) is a startup company focusing on strawberry precision farming, which was established in October 2020 and currently has only four employees. We generated over RMB 5 million ($790k) revenues in FY 2021 and expect similar growth levels in 2022. Currently, we’re cash-flow positive.
What do you think are the critical barriers to adoption for agricultural tech in China right now? Is it technology, policy or talent?
From my perspective, China’s agriculture is shifting from lived experience mode to a large-scale best practices mode. It is often a mode of thinking or a historical way of doing things that hinders technology adoption, rather than a lack of good technology itself.
At present, most of the growers allocate resources based on labour availability. What do I mean by this? For example, a family with two working labourers responsible for three mou of land (LL: mou 亩 is a Chinese unit of land measurement that varies by location but is commonly 806.65 square yards, 0.165 acres, or 666.5 square metres). The family will farm the land according to labour capabilities — such as watering the ground once every three days. They are constrained by what each person can do each day.
Based on system automation, the West or other advanced countries can achieve the best resource allocation. I can water the land six or ten times a day based on the weather and humidity. Chinese farming is based on the doable but not the optimal condition. I joke with my clients that the land is like a child. You can also give the child water once every three days. They wouldn’t die of thirst, but they might not be happy.
In some categories in China, such as flowers or high economic value products like blueberries, automation has become more popular. But in basic product categories like sweet potatoes, this is still in the pilot phase.
Labour availability resource allocation means that land utilisation becomes lower and lower, and labour capacity costs become higher and higher, especially against a backdrop of declining demographics. This will lead to a stalemate soon. The only way to change is to turn to a new mode of operations. With the benefits of scale economies, that's how you turn inefficient agriculture into efficient agriculture.
So for me, that’s the biggest hindrance right now. Not necessarily technology. The difference in mindset was also the biggest challenge for my team in entering the agricultural tech market. I think Chinese agriculture tech is not lagging at the technology infrastructure layer but in terms of data mining and data application.
Also, a major problem facing Chinese agriculture right now is a misconception. More than 95% of growers believe that chemical fertilisers are the primary source of crop growth, not photosynthesis. Their conceptualisation is still stuck at this level, and there’s no way to communicate with them here. Their thinking process is straightforward — if my crops are growing poorly, I should add more fertiliser. This means China's biggest problem right now is too much water and fertiliser usage.
So it sounds like technology isn’t the key barrier to adoption for Chinese agriculture but rather the approach people take. How do you get farmers to think that way if they are not at this level yet?
That’s the national situation of China as a whole; we segment the overall base into four types. The specific concerns of each kind of client are also quite different.
The first tier is what we call the industrial-level customers. They look to set the industry pricing and want an established expansion mode. They are concerned with two issues: whether they can set the industry pricing if they invest 1 billion or two billion? The second is how do they expand in the future? If they acquire 1,000 mou of land, how do they expand to the next 1,500 mou of land?
The second type of customer is scale-level customers. These scale-level customers will plant 50 mou to 100 mou worth of produce. At this level, they don’t worry about finding buyers. I’ll talk more about why this is later. Fifty mou to 100 mou we call the scale level of customer service.
The third type of customers are boutique-level customers; they have a few mou or a few dozen mou of land. High economic yield produce like strawberries may not be its primary financial source, but it may be an important point of attraction.
The fourth type is small retail customers. They might be cooperatives or individual plot holders. These customers aren’t digitally savvy and represent the general labour-centric thinking we’ve been discussing. With these customers, I think the difference in the mode of operation is too big. So in communications, we never talk about how we use artificial intelligence or anything digital. We focus on solving their problems, reducing costs, and improving their income.
How big are the different segments of users? And what are their unique characteristics?
About 95% of the market is in the fourth type of customers from the area covered. They cover nearly two million mou to 2.6 million mou. We think of these customers almost like retail customers. We are still figuring out how to approach this market. We can’t go from farm to farm to convert them. We are thinking about working with co-ops and trailing some projects in Yunnan, Anhui and Liaoning. We’ll see how far they are.
The boutique-level customers emerged really in the last two years. The so-called agricultural complex is a comprehensive farm with cultural tourism functions. The boutique-level customers are our target customer set since these customers are generally younger and open to water and fertiliser. They are happy to pay the service fee, hand over the execution, and remotely control the whole process. We will also give guidance on manual planting as well. But overall, this customer set is very responsive to our proposition.
Then scale-level customers are also a small percentage. They are typically experienced local leaders who have decades of planting experience. But these aren’t an excellent match for us. Why? Because these customers think they are the experts and will not listen to anything you say. We believe boutique-level customers will replace these customers in the future as produce production becomes more specialised.
The industry-level customers are also growing due to a few factors. Some are coming from a background in real estate and want to expand into agriculture. Others are coming to agricultural companies from rapid e-commerce and digitalisation. These customers are concerned about expansion and how to get pricing power. We work well with these customers since we are very aligned in outcomes.
So it sounds like boutique-level and industry-level customers are the key focus right now?
That’s right, focusing on boutique-level customers. We are designing our first generation of digital farms. Where we have 30 mou as a base unit. I want the yield of 30 mou to be comparable to the investment of 100 mou.
Our hope is that boutique farms do not have to reach 50 mou to 100 mou in scale but can generate 50 mou to 100 mou of output. We picked the production of 100 mou because 100 mou generates about 200 kg to 300 kg of fresh strawberries per day under the traditional planting model. A single vehicle can transport this, so logistics costs are the lowest. The cost is minimised, and the benefit is maximised. So if I can reach this yield with 30 mou of land, I can operate on a smaller business scale.
So is your offering different from the current mode of labour-centred resource allocation?
I’ll paint a picture of a traditional irrigation model. So you have often two elderly folks managing two mou of land with about ten greenhouses on there. To water, they have to open the village pump, which is often high-pressure. They keep the pump open and manually water each of the greenhouses. Not only does this take up an entire day at least, but it also uses a lot of water.
With our first generation of equipment, they have a remote control that allows them to switch on the irrigation system in each greenhouse automatically, and then the next and the next. The whole effort takes about half a day. Our next generation of equipment can do this via mobile phone.
So first off, we’ve saved the farmer time. Secondly, it’s reducing fertiliser usage. When they were using the battery-operated water pump, it was very high pressure, and they were adding all the fertiliser in one go. They will typically use 30 to 40kg of fertiliser per irrigation. With our system, since you can be more precise with water usage at lower pressure levels, they use only about 15kg each time.
The reduction of fertiliser not only saves cost but also reduces soil pollution. As the current fertiliser usage, soil quality is decreasing rapidly. With our more scientific drip irrigation system and intelligent control systems, fertiliser usage is reduced by half and yield increases by 30-40%.
So our proposition is that we are promoting a reduction in water and fertiliser usage to increase yield. But this is hard to educate most farmers on. So typically, we don’t try to educate but instead say, if you let us manage the water and fertiliser control systems, we will reduce your costs, reduce labour and increase your yields.
Do you need to have an onsite observation for these efforts?
All our services are remote. Though for our larger customers, they do want us to have people on the ground on farms. The ability to do remote monitoring and control also only happened recently since we could get enough data.
Our most significant gain from the Pinduoduo smart agriculture competition was the digital model for strawberry growth under varying growing conditions. For instance, we now have data for strawberries in various stages of growth without interference from water pollution and pests. So now, we can measure new growth against this theoretical benchmark. If it converges to the theoretical model, then great. If it’s deviating, then that’s how we know we should be intervening. Either through technology or onsite. Right now, we still haven’t fully automated the decision process to intervene, so occasional in-person intervention is still needed. Once we have enough data, the process should be fully remote.
What are your plans for the future?
From a company perspective, our short-term key focus is to standardise our hardware. We had several versions out during last year, and some of them were still in prototype mode. This year, we hope to produce a standardised model suitable for low-end customers and get greater coverage.
We want to slowly move from a project model-based business to a standardised product service in the medium term. We’ll also want to explore the prospect of turning our hardware into more of a platform. Right now, we have complete strawberry growth data, but we also think that if experts have blueberry growth data, they can also load that onto our platforms and use our automated systems. There’s an exciting and underexplored area where if you can systemise experts’ horticultural knowledge and allow a monetisation model that rewards them accordingly, there’s a lot of potentials there.
What do you think about the entry of big tech into the field of agricultural tech?
I think it’s a net positive thing. From my company’s perspective, we had market validation after Pinduoduo’s smart agriculture competition, and all of our customers came from inbound sales after the event. We had concrete data to benchmark our performance to the traditional farming approach. We also got a lot of investors' interest from the publicity. Investors now see agriculture as hot given the tech giants' involvement. We wouldn’t be where we are today without Pinduoduo’s competition, and I’m very grateful for that.
The downside is the lack of patience that internet giants and traditional VCs have for conventional sectors. It will be easy to get discouraged without patience if results aren’t improving in a quarter or two. You can not rush agriculture, not only because of the weather but also because client relationships are built on long term trust. This also means the sector wouldn’t have the typical exponential growth investors and internet giants are used to. The market is a big opportunity but will require players to be in it for the long haul.
Premium subscribers can look forward to a piece about the key challenges and rewards of investing in China next week and also new product videos about Ctrip in Circle community