Chapter Two: The Foundation — What I Created During Eleven Years at Dun'an

Chapter Two: The Foundation — What I Created During Eleven Years at Dun'an In the spring of 2007, the tax was refunded, and the nation revised its law. In that moment, I was certain I had embarked on the right path. But conviction alone is not enough — you must have the capability, the methodology, and the skill to turn conviction into reality. Dun'an provided me with a scenario sufficiently complex to test and run the systemic models I had in my mind, piece by piece. Those eleven years at Dun'an were the best experimental field of my life. It was not that I learned something from someone else; rather, I already had the system, the method, and the model, and then, on the vast field that was Dun'an, I repeatedly sowed, validated, adjusted, and harvested. I. A Single Excel Sheet That Transformed Real Estate Decisions from Guesswork into Calculation In the mid-to-late 2000s, the real estate market was at its hottest. Dun'an had acquired a plot of land in Shenyang to develop a courtyard housing project. The land was obtained through public auction. I still remember the day of the auction — the competition was incredibly fierce, with over a dozen rounds of bidding, even making the newspapers. We ultimately won the land, but at a very high premium, nearly double the starting bid. This incident plunged me into deep thought: why did the auction spin out of control? Because before the auction, we had no clear model to tell us — what was this land truly worth? Beyond what price would our profits be entirely consumed? In that era, most real estate companies bought land based on experience and intuition, with no quantitative basis for decision-making. I decided to build this model myself. My logic was simple, captured in four words: begin with the end in mind. The endpoint was certain — how much could the houses built on this land sell for? Assuming a sales price of five thousand yuan per square meter, a figure already validated by the surrounding market, the future sales revenue became a certain endpoint. Working backward from this endpoint, I began building the model. What was the plot ratio? How many square meters could be built? How long was the development cycle? What were the pre-sale conditions? When could the sales permit be obtained? I quantified every link with data. Then, I introduced the biggest variable — land cost. I discovered that for every 1% increase in land price, several percentage points of net profit would be eroded. If the premium exceeded a critical threshold, the project would flip from profit to loss. No one had ever presented this relationship so clearly with data before. This model was built in Excel, yet it became the core decision-making tool for company management. Every time a land purchase was considered, the model was run — input the land area, plot ratio, surrounding housing prices, and it would instantly calculate: at what land bid price would net profit be what amount; beyond which price point would risk become uncontrollable. From ambiguity to clarity, from gut feeling to data-driven decision-making — this was the first step toward certainty. But this was only the budgeting level. The real challenge lay at the execution level. Real estate projects have long cycles and numerous stages — from land acquisition to construction, to pre-sale, to delivery — each link tightly connected. A problem in any single link would trigger a chain reaction. I spent three years deconstructing the entire project execution process from start to finish, assigning responsibilities down to every department, every person, every single day. Take the most direct example: the pre-sale permit. Without it, the houses could not be sold. And to obtain the pre-sale permit, the rigid criterion of 25% of total project investment must be met. This meant that from land acquisition to groundbreaking, from foundation work to the main structure emerging above ground, no time node in any link could slip. Securing the permit one day earlier meant opening sales one day earlier and recouping funds one day earlier. Delaying it by one day meant one more day of capital costs, eating further into net profit. I precisely calculated the time nodes for every link down to the day. This was the embryonic form of my