Three Lessons the World Cup Teaches Pharmaceutical Companies: Build Relationships, Learn to Translate, Deliver Answers
Article images may contain Chinese text.

During a World Cup match, a commentator casually says:
“Messi has covered 9.8 kilometers in this match, with a dribble success rate of 67%—”
Three seconds. The data is there.
And it is not just a table, but an entire network of relationships.
Now turn back to your office.
Since last week, your team has wanted to see which products saw declining sell-through in East China last month.
IT says to pull it from CRM, sales says to reconcile it against the distribution data system, and marketing says to add a layer of retail outlet coverage… Three different definitions, and by the time they are reconciled, a week has passed.
Where Is the Difference?
Not in computing power or database size.
The difference lies in how the data is organized.
Almost every client asks the same question: why is there more and more data in our systems, yet using it when we actually need it gets slower and slower?
The World Cup has already provided the answer.
It comes down to three things


World Cup data is inherently a network of relationships.
Player A passes to Player B → the relationship is recorded automatically. A player’s movement hotspots → location data is linked automatically.
There is no need to manually “reconcile tables,” because the relationships are part of the data itself.
This relationship network is exactly what pharmaceutical companies’ data is missing
Retail outlet data is in CRM, pharmaceutical master data is in ERP, and staff coverage is in the office automation system—they are all “in the company’s databases,” but disconnected from one another.
Question: Which retail outlets did a particular representative visit last month? Answer: Look up visit records in CRM → reconcile retail outlet master data in ERP → check purchase records in the distribution data system.
Every cross-system question becomes an exercise in “manually building relationships.”
The First Thing Weiming Penguin Data Genius Does: Ontology Modeling
We no longer treat data as a series of two-dimensional tables. Instead, we first identify business entities—retail outlets, medicines, and personnel—each an independent entity with attributes and relationships to the others.
Once this knowledge graph of business entities has been built
Question: “Which products saw declining sell-through in East China last month?” System: Follow the relationship network straight to the answer, without searching through three tables.
Building relationships is the first step.


When a World Cup commentator says “defensive statistics,” the system automatically knows they mean:
Tackles + interceptions + duel success rate + recovery runs.
The same term can mean completely different things in different contexts. The system must learn to “translate.”
The “Translation” Challenge in Business Is 10 Times Harder Than at the World Cup
Your business user asks, “How were sales last month?”
IT has to ask at least 5 questions in return—does “sales” mean shipments or distributor sales to retail outlets? Does “last month” mean a calendar month or a rolling month? Which business unit?
Because the system does not understand everyday language.
The Second Thing Weiming Penguin Data Genius Does: A Semantic Modeling Layer
We build a semantic modeling layer on top of the knowledge graph.
Translate business users’ natural language into data language the system can understand
“Sales performance” → automatically map to different combinations of metrics based on the questioner’s role “Declining sell-through” → automatically identify as “month-over-month growth < 0” “East China” → automatically map to the corresponding list of provinces/retail outlets
Once this translation is in place, business users can truly query data in natural language, without learning SQL.
Learning to translate is the second step.


World Cup broadcasts do not just throw data at you.
They tell you: given the current score, this team will most likely make substitutions and adjust its tactics.
What they provide is not data, but a judgment based on data.
Traditional BI Stops at “Giving You the Data”
Your BI report is ready. You open it and see: East China sales last month were 12 million, down 8% month over month.
Now what?
What does that decline mean? Should you adjust your strategy?
BI gives you the data. Making decisions is still up to you.
The Third Thing Weiming Penguin Data Genius Does: From Answers to Recommendations
After you ask a question and receive an answer, the system automatically triggers the next stage of analysis
You ask: “How were East China sales last month?” System answers: “12 million, down 8% month over month.” Automatically triggered: Which products contributed most to the decline? What are competitors doing? How much room remains to increase sales volume at retail outlets?
The final output is not just data, but decision recommendations.
Delivering answers, not data, is the third step.


The World Cup comes around once every four years.
But pharmaceutical companies face data problems every day
Data is scattered everywhere, and answering one question takes three days of reconciling definitions.
Buying a few more BI tools will not solve these problems.
The fundamental solution is to rethink how data is organized.
From table-centric to entity-centric. From displaying data to recommending decisions.
World Cup commentary teams were already using this logic thirty years ago.
It is not too late for pharmaceutical companies to start their data intelligence journey now.