news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

As the hotel industry recovers in the post-epidemic era, changes in customer consumption behavior and the impact of local economic hotspots are constantly affecting the nerves of hotel operators and even every store manager. From the core of customer acquisition to improving revenue efficiency and digital online collaborative management, hotels need to actively seek change.

BTG Homeinns Hotel Group (hereinafter referred to as "BTG Homeinns") is known as one of the "Big Three" hotels in China. According to the latest public data, it owns 6,300 hotels and 480,000 rooms across the country. In the past few years, with its strong operational capabilities, BTG Homeinns has entered a very rapid development stage. It currently covers the fields of luxury high-end, taste mid-to-high-end, quality mid-range, classic comfort, diversified franchise, leisure and vacation.

Starting from July 2023, BTG Home Inn has promoted the launch of "AI Digital Store Manager", actively explored AI big models, and initiated AI performance competitions by the Group's Technology Center to continuously form scenarios for implementation.

Recently, in the "Digital Value Observation Room", Wan Ning, co-founder of Titanium Media Group, initiator of ITValue, and dean of Titanium Media Research Institute, talked with Wang Bo, general manager of BTG Hotels Technology Center. During the interview, Wang Bo revealed that BTG Home Inns is building a passenger flow prediction model and establishing a "revenue encyclopedia" knowledge base, and using large models to summarize rules and output optimization suggestions, so as to accurately price 6,300 hotel rooms.

Click to view live replay

"Revenue Encyclopedia" has also grown into the most important functional module for BTG Home Inn to create "AI Digital Store Manager". At the same time, Wang Bo also pointed out that exploring the application of AI in business scenarios is a key link in the digital development of the entire enterprise. He also put forward a bold hypothesis: the future organization will be an organization based on the twin of human and AI digital technology. If AI can replace 40% to 60% of work capabilities, it will bring a huge leap in organizational productivity. This is also the driving force behind BTG Home Inn's firm embrace of AI.

The business logic of hotels has changed

Based on market feedback, especially since the second half of 2023, continued holiday consumption is gradually affecting the recovery of the hotel economy, while also showing some new trends.

The 2024 China Hotel Industry Development Report released in April this year shows that by the end of 2023, there will be about 90,600 chain hotels in China, an increase of 28% from 71,000 in 2022. The number of facilities in the four hotel industry grades of economy (two-star and below), mid-range (three-star), high-end (four-star), and luxury (five-star and above) accounted for 78.51%, 13.91%, 6.02% and 1.56% respectively. Calculated by the number of chain rooms, the hotel chain rate has increased from 38.75% last year to 40.95%, of which the chain rates of mid-range and luxury hotels are both over 55%.

The report also predicts that this year, after more than a decade of rapid development, the growth rate of the economy hotel industry will slow down and competition in the mid-range hotel industry will intensify.

It is not difficult to find that economy and mid-range hotels are still the basic plate of the industry. These hotels are often aimed at tourists who are sensitive to price, as well as guests who have a certain pursuit of grade but cannot afford too high a price. In addition, with the further increase in the hotel chain rate, the competitive pressure of different brand hotels is also further increasing.

Wang Bo has his own judgment on the changes in the market structure: First, the structure of the travel population has changed, from business travel in the past to a very rapid growth trend in leisure and tourism travel after the recovery of the epidemic market. Second, the game between stocks is becoming more and more obvious. There are fewer and fewer opportunities for growth. Compared with developed countries, the number of hotel rooms in China for every 10,000 people is still relatively low. But at the same time, everyone feels that it will become more and more difficult to open new hotels, and there are differences in demand and the number of travelers. Third, the industry is gradually concentrating on the head. The challenges of development, operation, talent, digitalization or other related investments for the following hotels will become greater and greater. This is basically consistent with some of the trends presented in the above report.

Intuitively, as market demand changes, hotels need to quickly adapt to changes in the market environment and actively seek innovation and transformation to meet challenges. Different brands of hotels have given different answers to this question.

Hotels should start with the revenue model when implementing AI

BTG Home Inn has been exploring hotel digitalization for a long time. Whether it is attracting customers online or through new channels, or using digital means to improve hotel management efficiency, hotel service level and quality management, as well as digital investment in high-end and high-star hotels, it has made many arrangements and initiatives.

Relying on relatively mature AI technology or co-creation of the industry is also a natural consideration for BTG Home Inn. In the current chain hotel industry, the core revenue management scenario actually has the value and advantages of AI technology implementation, that is, high input and high output, and it is feasible.

In Wang Bo's view, the hotel industry has accumulated a wealth of data, with a high degree of coverage of market insights and city hot spots. "With coverage of 580 cities across the country, it is easy to know where the hot spots are, where the hot spots are changing rapidly, and how to seize short-term hot spots to do business. This requires us to provide an agile revenue model for these revenue points."

What is a revenue model? Why do we need revenue management?

The purpose of revenue management is to maximize room revenue and obtain the most profitable guests. This concept is often used in service industries such as aviation, hotels, and cinemas, which have the characteristics of high fixed costs, fixed production capacity, and non-storable products.

For example, the number of hotel rooms is fixed, hotel rooms can be pre-sold, and the value of hotel rooms is time-dependent. If a room is not sold on a certain day, the value of the room on that day disappears. When the demand for rooms is high, the hotel manager needs to sell the rooms at the highest price possible; conversely, when the demand for rooms is low, the room sales volume should be maximized.

To this end, hotels often classify customer needs, such as business customers and leisure tourists, price-sensitive customers and those pursuing quality. At the same time, the customer needs faced by hotels will fluctuate with time, season, city, and hot events.

In each time period, there are certain cycles and rules. Experienced store managers often need to make scientific and reasonable price recommendations based on the rules and data analysis summarized over a long period of time and their own experience. Once some sudden hot events occur, the response ability of each store manager is often tested.

BTG Home Inn hopes to improve its digital analysis capabilities and AI large model learning capabilities, define the basic revenue model for each hotel based on big data and the store managers' rich experience, formulate room price forecasts more quickly, and provide specific reason analysis, which will directly affect each hotel's room revenue and seize competitive opportunities in a rapidly changing market.

To this end, Home Inns & Hotels.com first conducted a logical breakdown of revenue management in two dimensions.

The first is passenger flow. Since 2023, BTG Home Inn has started to predict passenger flow based on a large model, and has included historical comparisons of the same store, comparisons of sister stores in the business district, comparisons of competing products in the business district, hot events, weather and other factors into analysis variables. Currently, based on the passenger flow prediction model, it is possible to achieve flow predictions for the same day, 3 days, 7 days, and 14 days. For example, the accuracy rate for 14 days can reach more than 80%, and the accuracy rate for the same day can reach more than 90%.

The second is price. Wang Bo pointed out that the team first wanted to build a rule model, that is, to set different price rules under specified conditions to simulate price adjustments. However, due to the particularity of each hotel, it is difficult to verify whether the white box model is correct after it is completed.

In comparison, the black box model, that is, the price prediction method based on AI, is more applicable to different scenarios. The logic of BTG Home Inn is: first, let the model learn from the price adjustment rules of 500 or 1,000 excellent hotel general managers and find variable factors; secondly, connect to the big model capabilities provided by external manufacturers to achieve a revenue encyclopedia model combination based on "data model + big language model".

"First, we make data model predictions based on massive amounts of data, and then we make a revenue encyclopedia. After collecting a lot of revenue experience, we summarize the rules based on the big model and tell the hotel general manager why we came up with such rules, including the source of prices, price adjustment logic, suggestion logic, etc." Wang Bo pointed out.

Training private domain models is difficult

However, the difficulty of building a private domain big model for a vertical industry will be different from that of a public domain big model trained based on Internet data. Due to the lack of data sets, learning private domain knowledge will bring new challenges. Can these challenges be solved by using transfer learning, external knowledge base, or adjusting the applicability of scenarios?

Regarding the challenges in the process, Wang Bo emphasized, "In the private domain (B-side), the path of AI training/learning is very complicated. Therefore, we began to conduct research in collaboration with famous universities on how to make digital capabilities or AI twin capabilities closer and closer to human best practices."

Take the Revenue Encyclopedia as an example, it is a typical growth path. If experienced store managers share their experience in the Revenue Encyclopedia, and then let the big model learn based on these samples, the growth rate of the model will be greatly enhanced.

In his opinion, unlike other AI scenarios that have been implemented, such as AI smart customer reviews and the Little Prince of Answering Questions that provides answering training, the core of which is to generate answers based on a corpus or integrated into the workflow. Income Encyclopedia is a process of self-learning and self-growth.

"Being able to generate a new answer, instead of explicitly telling him the options, the new answer can be guided. This ability is what particularly attracts me to the big model. At the same time, we are also looking for such a learning path, because learning the private domain big model is inevitable. Maybe in the future we can find some particularly suitable scenarios, and through continuous learning, the model will become more and more vital." Wang Bo pointed out.

Through the above practices, BTG Home Inns finally built its own "AI Revenue Assistant", which is just the first step. Behind the AI ​​Revenue Assistant, there is a bigger vision - AI digital store manager.

According to Wang Bo's sharing, the "AI Digital Store Manager" currently has three main capabilities: First, 60% of the repetitive functions of the hotel's daily operations are handed over to the AI ​​digital store manager to complete, such as daily business review, market hotspot judgment, hotel revenue management, guest service feedback, risk and emergency alarm processing, to achieve operational assistance; second, especially in terms of revenue, feedback is provided on some sudden events and other overlooked revenue opportunities; third, the standardized management level of the entire hotel chain can be improved.

At present, AI digital store manager has explored many scenario models, including hotel management indicator model, hotel order query model, hotel operation knowledge base model, AI questionnaire interactive model, AI question bank and simulated answering model. In the future, it is also planned to build an AI hotel management data analysis model, which can output decision-making information in the form of chart analysis, which can be used for industry data comparison, KPI prediction, etc.

It can be seen that BTG Home Inn has made in-depth exploration and preparation in shaping AI and digital transformation capabilities. The business results will in turn boost its confidence in investment.

Wang Bo also pointed out that before this, it was very difficult and challenging for hotels to achieve standardization in terms of efficiency, management, revenue, and daily management of employees. BTG Home Inns has taken some measures, such as improving the transparency and penetration of management through mobile and online tools. However, the implementation of these measures still requires a lot of manpower to establish rules, models, and scenarios, and then promote them. (Generative) AI provides a new path that changes execution efficiency and management transparency, and also helps to achieve management standardization.

"AI has opened a new window for us. But in the process of embracing AI, we must also maintain a calm mind. Only a big scenario can create a big industry," said Wang Bo.(This article was first published on Titanium Media APP, author | Yang Li, editor | Gai Hongda)

From September 11th to 14th, the 2024 ITValue Summit Digital Value Annual Conference, co-organized by Titanium Media Group and ITValue, was themed "Ready For AI". We reunited with old and new friends in Sanya to exchange experiences and lessons, cross-industry thinking, promote innovative transactions, and explore new opportunities in the digital economy era driven by AI based on innovative scenarios, and jointly create a feast of AI innovation and exploration in the digital economy era. Click to enter the 2024 ITValue Summit Digital Value Annual Conference for more information

Titanium Media has launched the "Innovation Scenarios 50" selection, selecting and interpreting 50 innovative scenarios and solutions that deeply integrate the entire industry and business, and grandly awarding and in-depth exchanges at the 2024 ITValue Summit Digital Value Annual Conference. Scenarios are currently being collected, with more accurate interpretations, wider exposure, and stronger brand potential. You are welcome to ask questions and leave solutions and tools. Click to sign up for the selection of Innovation Scenarios 50