Interview with Gridsum: understanding one of China’s most understated software companies
A conversation with Qi Guosheng, the founder and CEO of Gridsum
by Ha Qiang and Hima Tammineedi
Gridsum is a leading provider of enterprise-grade cloud-based big-data analytics and artificial intelligence (AI) solutions in China. Gridsum’s client list spans multinational enterprises and large local companies operating in China, including a substantial number of Fortune 500 and China 1000 companies, as well as government departments and agencies.
The company was founded by CEO Qi Guosheng in his college dorm room with friends while he was studying computer science at Tsinghua University in 2005. The company started as a search optimization company, but soon evolved to providing Big Data and AI solutions. For the past 12 years, the company focused on developing data analytics and AI technologies and gradually established itself as one of the market leaders in China. Gridsum has been listed on the Nasdaq Stock Market since Sep 2016.
The blog had the great opportunity to sit down with the company’s founder and CEO Mr. Qi Guosheng for an interview to discuss the company’s operations and competitors. This interview has been lightly edited for clarity.
Q: In your own words, could you tell us what Gridsum does?
A: Gridsum is a big data enterprise software provider, focusing on 3 sectors:
The first is marketing automation. This is the first sector that Gridsum started with. We help big enterprises make more efficient, data-driven marketing decisions. By doing this, we help them to achieve a higher ROI (Return on Investment) and increase their business KPIs (Key Performance Indicators) when they spend money on marketing channels.
The second is professional services automation. This includes helping both legal services and financial services to automate their solutions. One of our big projects in the legal area is helping all of China’s judicial system to build a “Smart Court.” And we are building finance automation as well. Basically, any area that has paperwork — if your job is about discovering useful patterns in documents, or generating some new documents from the existing documentation, then we can help you. We will train the computers to become an expert in that sector and automate the task for you.
The third is the Internet of Things (IoT). In this area, Gridsum is helping enterprises create a “digital twin” for the energy industry such as oil and gas. A digital twin is, at its core, a simulator. If you have watched the movie The Matrix, it’s similar to the simulator of the world. In our case, we aren’t creating a fictional world for people, but are rather creating a fictional world for machines. By doing so, we aim to help facilities manage energy and resource consumption more efficiently. Let us take the AC in a building as an example. Gridsum will build a model that perfectly reflects the AC working in the real world. Once you build that, you can input the real-time status and perform predictive maintenance — forecasting how it will behave in the future. You can also play with this digital twin and see how the AC will behave when the outside environment changes. Thus, beyond predictive maintenance, you can also tune the digital twin to measure and optimize performance.
In terms of Gridsum’s competitive edge, our global best-of-breed approach to unstructured data analysis is a key differentiator. This includes our approach to building “language graphs” and “knowledge graphs” to digitally map deep and comprehensive real-world domain knowledge. This also includes our natural language processing technologies which are among the best in the world for enterprise-grade, domain focused Chinese language applications. These are the biggest differences between Gridsum and other AI companies in China.
Q: In the marketing automation sector, how are you different from traditional marketing consulting firms?
A: Ad agencies provide consulting services. Gridsum does not do consulting, we offer products and solutions that cover the entire customer journey. Let us talk about the customer journey. There are five stages. Imagine you are Nike. The first step is media planning, during which your marketing department decides how much you want to spend on marketing and on which channels. Second, you spend your money and generate awareness/traffic. The third step is about converting traffic into business KPIs. If sales is your primary KPI, then you want to convert your traffic into achieving sales targets. If engagement is the most important then you want to convert traffic into KPIs like how much time customers and potential customers spend watching a video or playing a game related to that product, or how many people take a survey. The fourth step is loyalty — you want them to be your client the second time and repeatedly thereafter. And the last step is converting loyal customers to be active ambassadors for your brands. These customers tell their friends about your brand — becoming a powerful advocate for you. For example, when talking about sports shoes they would recommend Nike to their friends over other brands.
Gridsum is the only company in China that covers all five stages. Some companies might be good at ad analytics, media planning, how you spend your money, and detecting if your advertisement is being viewed, but they are not doing traffic conversion to KPIs and do not have solutions on loyalty. There are also companies good at building loyalty programs, such as Convert Lab. But we are the only ones in the space with an integrated solution for the entire customer journey.
Q: We found the digital twin concept quite interesting. Can you give us a detailed example of how you construct this digital twin?
A: Let’s use a real example and look at oil pumping machines. To build a digital twin, you first place sensors on the pump and collect many different types of data: pressure, workload, temperature outside, and many others. Then you can also incorporate knowledge data into building the digital twin. Knowledge is understanding how to read, interpret, analyze, and mine key insights from the data. Knowledge is another piece of data, but you cannot just generate it or collect it. It must be acquired from the experts who know the machines very well. So the experts will tell you how they read the data and what the data means, etc. Gridsum will start modeling based on sensor data to produce the digital twin. Once we have the model, we will also input the knowledge data from the human experts. Once we combine time-stamped sensor operating data and knowledge (unstructured data) together, we then have an AI expert digital twin. With the digital twin, we can figure out what a strategy for producing more oil and gas in a certain time period should be.
Q: You mentioned you operate in three sectors: marketing, professional services, and IoT. Why did you choose these three sectors?
A: Our sector selection philosophy is to choose a sector where there are a lot of data and a lot of problems that can be solved by data. Phase 1, we enter the sector, try to solve the problem and accumulate expertise about the sector. After a while, we enter phase 2, where we begin automating the problem-solving process, utilizing the expertise we have accumulated.
We chose marketing because everybody needs it. We chose professional services because there are a lot of lawyers, bankers and accountants. We chose energy production and consumption because it is a big industry. And also because of the impact we can make there. A 1% or 2% enhancement of production will make a big difference.
Q: Got it. Who are your competitors?
A: We have different competitors in the 3 different sectors.
Sector 1: In the marketing automation sector, there is Adobe Marketing Cloud, IBM Marketing Cloud and Google Analytics. In China, we don’t really have one strong competitor, but have competitors doing pieces of what Gridsum does. Instead, what Gridsum offers is a comprehensive, one-stop solution.
Sector 2: In the professional services sector, Gridsum is a pioneer (in China). IBM Watson is working on similar things and they are big, but they aren’t doing as well in China. A big reason lies in the difference between Chinese and English. In English, there are spaces between words, but in Chinese, there are no spaces. So the NLP algorithms required for English and Chinese are very different. So really, in this sector, Gridsum is a pioneer. It’s a new market with new tech, and everything is very new. There are some indirect competitors working in the same field, but they don’t offer similar solutions. You might think traditional IT providers for courts and offices might pose a challenge for Gridsum. But in fact, what Gridsum offers sits on top of those previous solutions. We build AI solutions on top of the traditional IT platform.
Sector 3: In IoT and in oil and gas, Gridsum is pretty unique as well. We do have competitors, but we also cooperate with them, like traditional oil and gas consulting firms, and strategy firms. These companies provide all the consulting and produce the big equipment. But Gridsum is very focused on creating the digital twin to enhance all these companies’ results. There are also some companies that sell IT integration solutions, but it’s not exactly the same as we focus more on the analytics side.
Q: How did you develop the competitive advantages of covering the entire customer journey and the ability to analyze unstructured data?
A: First, Gridsum has a long history. We are a big company now and have a lot of people. We have more people than competitors so we can have more people covering the customer journey.
Second, Gridsum is not like many tech or applications companies. We started as a company building “soft infrastructure,” not applications. So we have a very strong infrastructure layer for data and analytics. Our infrastructure and development platform enhances our capability to develop other data analytics on top of it.
Third, even outside of marketing automation, Gridsum can expand our landscape to professional services and IoT. Some might say these things are not connected, but we think they are because the key challenge for delivering tangible results for clients in this area is about having a powerful soft-infrastructure AI platform layer to process structured and unstructured data. This is our strength.
I want to expand my answer to the competitors to give you a bit more information. If you’re in an industry that has no competitors in 5 years, you are probably in the wrong industry.
In marketing automation, there were a lot of foreign competitors. But Gridsum could (and has) won significant share because the China environment is very different and we understand it better. It is not just the operating environment that is different, such as the Great Firewall in China, but the way of thinking and principles of spending time on the Internet is different. China has a very different consumer and enterprise technology culture. Even without the firewall, Chinese companies and foreign companies would have developed different online applications. For example, let’s take a look at QQ and MSN messenger. The Chinese government never banned MSN messenger. 20 years ago, MSN messenger was the best, everyone used it as a desktop chatting tool. But later when QQ came out, MSN lost significant market share. QQ just did a better job, to make sure that Chinese people like it, so that more and more people switched to it. Initially, Chinese people used both MSN and QQ at the same time, but then after 2–3 years, no one used MSN messenger anymore. We, like QQ, understand Chinese customers better. That is why Gridsum has been very successful when competing with Western companies in the marketing automation sector.
Now let us turn our attention to the difference between the Data Analytics/ AI business and traditional IT businesses. In the traditional IT industry, new companies always have a stronger position when they compete with existing companies because they do not have the historical burden of legacy technology and can adapt the latest technologies easily. But in the Data Analytics / AI sector, things are different. Accumulated data is a treasure. The longer you have been in the industry, the more data you have. You can learn from the abundant data and sharpen your expertise. You can then use your expertise to provide better service and accumulate more data. And this creates a self-reinforcing cycle. Even though new companies have newer technologies, they do not have the data. This is something I like about my business. If you do the right thing at the beginning, then your competitive advantage will get bigger and bigger.
Q: Is the Chinese government supporting this industry?
A: Yes, definitely. Speeches from the leaders show they not only support, but are also trying their best to promote this industry. This is an industry where China could have fallen behind the US. But now we have a lot of data, a much bigger population and hence bigger consumer market and at the same time, we do not have the legacy system problem for using this data. I am certain we can lead here, globally. In addition, in China, almost everything is already in the cloud, so you don’t have to worry about moving from on-premise IT to cloud-based environments like in the US with very old companies. So data collection here (in China) is much easier.
Q: Now let’s switch gears to talent and innovation. Where do you find the best talent?
A: I am sad to say this, but in the IT (Data/AI) industry, we look, to some degree to the west. Technology-wise, they took an early lead in areas like microchips and algorithms, etc. They have great universities and enterprises such as CMU, Microsoft, and innovation hubs like Silicon Valley. So that’s why Gridsum wants a presence overseas (in the States). In the R&D division of my company, many people have either overseas working experience or have studied in a foreign country. We also hire some of the most talented graduates from the best universities in China such as Tsinghua, my alma mater, and others.
Q: Lastly, innovation is a key element of success in this industry. How do you encourage real innovation in your company?
A: Great question! We spend a lot of time thinking about this and have striven successfully to incorporate innovation into our company’s DNA! We organize a lot of activities and encourage people to try out new ideas. We host hackathons every year and also have an internal incubator, helping people to build new products that generate real revenues. If a product is successfully launched, the team that made it will receive a certain amount of bonus and accolades. In addition, we encourage our people to apply for patents. We are now ranked #1 among all AI companies in China for IP applications.
—
In conclusion, we learned quite some interesting things from Qi, about his company, and about how to think about Chinese tech companies in general.
This story was originally published on December 22, 2019 at https://www.chinatechblog.org/blog/gridsum.
Disclaimer: The company received a Nasdaq Notice due to being unable to timely file its annual report on Form 20-F in 2018 and was requested for a hearing in front of the Nasdaq Hearing Panel. The company has regained compliance with Nasdaq rules on January 11, 2019.
