---
title: "If Polls Are So Unreliable, What Can Marketers Trust?"
description: "Despite Biden's victory, polls proved unreliable, prompting a debate on whether market research and data can be trusted. The article argues that data and insight serve different roles: data provides information, while insight interprets it, and they must be used together, with insight leading and data validating."
author: "响马老苗"
publisher: "New Distribution"
email: "zhaobo258@gmail.com"
telephone: "+8615854817671"
published: "2020-12-09"
language: "en"
canonical: "https://xinjignxiao.com/en/articles/if-polls-are-so-unreliable-what-can-marketers-trust-3ebc1418/"
markdown: "https://xinjignxiao.com/en/articles/if-polls-are-so-unreliable-what-can-marketers-trust-3ebc1418.md"
original_source: "https://mp.weixin.qq.com/s/sLro8l7N7Eq46cM3aNL4PA"
translation: "https://xinjignxiao.com/zh/articles/%E6%B0%91%E8%B0%83%E9%83%BD%E8%BF%99%E4%B9%88%E4%B8%8D%E9%9D%A0%E8%B0%B1-%E6%88%91%E4%BB%AC%E5%81%9A%E8%90%A5%E9%94%80%E7%9A%84%E8%BF%98%E8%83%BD%E7%9B%B8%E4%BF%A1%E5%95%A5-3ebc1418.md"
attribution: "New Distribution — https://xinjignxiao.com/en/articles/if-polls-are-so-unreliable-what-can-marketers-trust-3ebc1418/"
usage_policy: "https://xinjignxiao.com/ai-policy.txt"
---

# If Polls Are So Unreliable, What Can Marketers Trust?

> Despite Biden's victory, polls proved unreliable, prompting a debate on whether market research and data can be trusted. The article argues that data and insight serve different roles: data provides information, while insight interprets it, and they must be used together, with insight leading and data validating.

**-01-**
Although 77-year-old comrade Biden won the election, polls have made people feel even more unreliable.
The promised landslide victory, the promised crushing defeat, turned into the most contentious election in American history. The candidate with the most votes in history barely defeated the one with the second most. Comrade Jianguo (Trump) is still cursing the other side for "not playing by the rules."
It is said that major U.S. polling agencies, after being slapped in the face in 2016, worked hard for four years. This year's polls used unprecedented methods and sample sizes, with various parameter corrections and high-tech means.
The result was an unprecedented slap in the face.
Some mocked that spending so much money on massive data was less accurate than simply using the candidates' ages: 77-year-old got 77 million votes, 74-year-old got 74 million votes. That's closer to the actual result than the polls. Well, this method is also our ancient high-tech: throwing shoes.
**-02-**
Similar controversies have always existed in the business world: Should we trust market research and market data?
**The mainstream view is to let data speak and be market-oriented.** Chairman Mao said, "No investigation, no right to speak."
But saying is one thing; doing is another.
Some people are openly against market research, like Jobs and Ford. The most famous quote is: "If I asked customers what they wanted, they would say a faster horse." **In China, "using data" and "using market research" is politically correct. The only big entrepreneur bold enough to say he doesn't do research and doesn't trust data is probably Zong Qinghou.**
In practice, most bosses and marketers do this: if data supports their judgment, they believe it; if not, they "investigate further" or trust their "insights." This year, Genki Forest became popular because the team came from the internet industry. It's said that Genki Forest makes decisions "purely based on data," avoiding subjective judgment. But insiders say Boss Tang is very strong-willed, and data is just a decision-support tool.
Perfect Diary became popular, and some say data-driven decisions were key to their rise, while others say that's nonsense and it's a model project pushed by Ma (Jack Ma) for "new manufacturing."
Now, informatization has evolved into the data age and AI age. Computers can learn autonomously; AlphaGo beat Ke Jie; cars are self-driving.
But the "research school" and "insight school" are still fighting.
**The myth of big data is that future customers won't need to speak; external systems will know what you need, whether you can pay, and the best matching product.** What risks you face, whether you'll lose your job or go bankrupt, whether you have a mistress or your wife will cheat. The most persuasive business case is Toutiao, which pushes content based on your interests.
In short, in the future, big data will know you better than you know yourself.
The insight school says, "Bullshit! Even the U.S. election, where shoe-throwing could be 50% accurate, was so wrong. Predicting more complex consumer behavior is nonsense; it's just spending money to tell you what you already know."
Shouting "car" at your phone gets car ads; talking about houses gets house ads; saying you're hungry gets food delivery. These are myths created by platforms and data companies, and myths are deceptive.
For this topic where everyone pays lip service to political correctness but actually fights, Old Miao, with his consistent principle of not being objective or neutral but speaking freely, will dig deep.
**-03-**
First, let's understand a common sense about market research: What is the purpose of research?
It can be summarized in two keywords: **trend and trajectory.**
Trend includes industry policy, scale, growth, profitability, consumer trends, technological changes, upstream changes, channel changes, category development, and major moves of leading brands. By aggregating these phenomena, you get an overall trend judgment, often called "finding the wind."
**A common mistake here is that many are misled by the single parameter of industry growth, thinking fast growth means more opportunities and a wind. In the end, they all rush in and fall into pits.**
The reason is that fast growth often means more entrants; opportunities everyone sees are not opportunities. Joining the fray without real competitiveness is a recipe for being cannon fodder.
Understanding trends (note: "trends," not "market conditions") helps check whether your current strategy, direction, and operations match the trend, whether you can find next-step opportunities, leverage market trends, or spot future risks.
Before informatization, companies also learned these things, but through exhibitions, forums, media, interviews, and spying, or buying business information. Many bosses liked to drink and chat with industry veterans; a single drink or a tip could spawn a product or even build a company. This is called "small talk leads to big business; drinking boosts productivity."
But big data allows us to understand more comprehensively and specifically; this is a "quantitative improvement."
**-04-**
The other keyword is trajectory, which has three types: **behavioral trajectory, product trajectory, and psychological trajectory.**
Behavioral trajectory consists of static and dynamic parameters. In consumer goods, static parameters are mainly demographic: age, residence, gender, income, education, family status, occupation, beliefs, social class, etc. For business users, static parameters are industry, size, revenue, operational characteristics, organizational processes, reputation, etc. The "user profile" that platforms like to talk about is this part.
Dynamic parameters of behavioral trajectory revolve around consumption behavior: user identification, information search, decision-making, purchase frequency, purchase quantity, etc.
Product trajectory is the path from production to distribution, to purchase and use.
The hardest to grasp but most important is psychological trajectory, which also has static and dynamic parameters.
**Static parameters refer to user personality traits, behavioral preferences, consumption preferences, and value orientations; dynamic parameters refer to needs, motivations, attitudes, cognitive characteristics, risk perception, experience, expectations, and satisfaction.**
In the big data era, mastery of trajectories has improved qualitatively.
**What makes big data big?**
**First, sample size.** Previously, our data research was based on sampling. For example, in traditional questionnaire surveys, in first-tier cities like Beijing and Shanghai, an effective sample of over 300 was considered representative of the entire market.
But in the big data era, theoretically you can scan all users. Larger sample sizes make results more accurate.
**Second, from partial to panoramic.** Previously, we observed consumer behavior at the point of sale or got feedback from terminal staff, usually focusing on a single behavior. But with big data tracking, we can understand customers across the entire process: problem recognition, search, evaluation, decision, and after-sales.
**Third, from subjective to relatively objective.** Previously, predicting consumer behavior relied more on consumers' self-judgment, described to us in words. For example, asking consumers if they would buy a product with certain features or what price they'd accept. Big data's logic is to infer future actions based on past behavior, which is undoubtedly more reliable.
There's a saying: Don't watch what they say; watch what they do.
Big data also has an advantage in tracking product trajectories. Previously, companies managed goods flow, tracking distributors, secondary wholesalers, and terminal inventory, but it was rough. Now, technology can track each product in detail.
Research through big data is still research; it still collects and organizes information, acting like eyes and ears, not the brain.
Big data can greatly compensate for traditional research's shortcomings, but what traditional research can't do, big data can't either. The most prominent is grasping users' psychological paths.
Like in this U.S. election, there were many loyal fans of Comrade Jianguo who didn't follow the rules: they stayed quiet but voted enthusiastically. Although pollsters noticed this phenomenon and adjusted predictions, the sheer number of such people was still unexpected.
**-05-**
We often call grasping users' psychological paths "human insight." It always leads to the most effective decisions.
Excellent entrepreneurs are masters of human insight. From Bill Gates to Steve Jobs, from Bezos to Musk, from Ren Zhengfei to Ma Yun, from Shi Yuzhu to Ma Huateng, from Duan Yongping to Zhang Yiming, all are top experts.
We say insight is the ability to see the essence through phenomena. A person's business insight is a comprehensive reflection of their experience, exposure, understanding of society, business, industry, products, and users. In *The Godfather*, old Vito Corleone says to his son:
**"Those who can see through the essence of things in half a second, and those who can't see it in a lifetime, are destined to have different fates."**
Of course, insight isn't mysticism; even experts' insights aren't always reliable. This ability can be traced, cultivated, and learned. In the future, I'll dedicate an article to "business insight."
Today, we'll discuss the relationship between market data and insight.
**First, market data is an important object of insight. We can discover market opportunities, growth points, and predict risks from a pile of data.**
Those who work with market data know that a data report can be hundreds of pages. Often, it tells us what we already know or take for granted. How do we gain insight?
Here are two common methods.
**The first method is to look for obvious "errors."**
Consumers have severe information asymmetry with industry insiders. Things we consider common sense may be wrong in consumers' minds, and that's a huge opportunity.
Yake V9 pioneered the vitamin candy category in China. A crucial piece of information prompted Boss Chen of Yake to launch the project: research found that many consumers thought they had eaten vitamin candy, and many said they had eaten Alpine candy.
In fact, at that time, vitamin candy was almost nonexistent in China, and Alpine didn't have vitamin candy either. The market didn't have it, but consumers thought it did, so Boss Chen saw the business opportunity.
If you study the market enough, you'll encounter similar situations: most people think Nongfu Spring water is all mineral water, many think MSG is harmful, and strong-flavor liquor is made from alcohol, etc.
These hide huge business opportunities.
**The second method is to dig deeper logic from ordinary phenomena.**
A bakery company did data research and found that savory products grew faster than sweet ones in recent years.
A typical company would adjust product mix: make more savory, less sweet. But an insightful boss thinks differently.
They dug deeper into why savory products grew faster: meal replacement scenarios increased. People on diets, working overtime, staying up late, gaming, or having breakfast prefer savory food to fill up. So the company developed a series of products around meal replacement scenarios and achieved great success.
This is a more typical "seeing the essence through phenomena." In fact, we can gain deeper insights from many common phenomena.
**-06-**
**The second relationship between data and insight is: using data to validate and deepen insight.**
Traditional research has questionnaire design; big data research has algorithm design. This is the soul of research.
**Whether it's qualitative market visits or quantitative data research, you must go in with questions. Don't research for the sake of research; otherwise, you'll get a pile of information garbage or even be led astray.**
In recent years, domestic entertainment's top traffic has been dominated by young idols. From Weibo topics, popularity, fan numbers, and fan club activity, they dominate the industry. This led to many female-oriented products like cosmetics and luxury bags using male celebrity endorsers, especially international brands.
The results were mostly unsatisfactory, and this year, brands have switched back to female endorsers.
This is a classic case of lacking algorithm design and being led astray by data. Male traffic stars have higher fan loyalty and are more willing to vote for their idols, so they crush female stars in data. But choosing an endorser isn't just about traffic data; it also involves brand-product fit, associations, and mass appeal. If you only consider traffic data, the chance of a good outcome is lower than "throwing shoes."
**Marketers must remember: insight comes first, data second; don't reverse the order.** Data reflects phenomena; it can't directly give insight, but it can corroborate and refine insight to some extent.
**-07-**
Summary:
**1. "No investigation, no right to speak," but how much say research should have in final decisions has always been controversial.** The common understanding is that research collects information, while insight processes it and sees the essence. Insight without research is blind; research without insight is mindless and formalistic.
**2. In the information age, companies get information more easily, and people's behavioral trajectories are easier to track.** This allows companies to obtain more accurate and effective information, making it more important to build market intelligence systems through big data.
**3. Big data intelligence replaces some previous information-gathering methods, like reading media reports or digging for tips.** But big data cannot replace insight, nor can it automate decision-making. Research and insight still have their own roles. Use big data for insight, and use big data to validate and deepen insight conclusions. This is the essential relationship and won't change.
**4. Big data's value to marketing isn't just in intelligence systems; it also plays important roles in sales management, promotion management, product and brand management, and supply chain management, but the foundation is the intelligence system.**
**5. As always, "where you sit determines what you see." Platforms with raw data and tech service providers with big data capabilities like to mythologize big data to create panic, seize discourse power, or harvest the masses. Companies should be wary. But on the other hand, rejecting new things and failing to recognize the value of new technology can also harm companies. Teacher Ma Baoguo taught us to "behave well."**
Source: Old Miao Tears Marketing (ID: yiheyingxiao) Author: Xiangma Laomiao
Payment will be made 400-2000 yuan for tips adopted.


---

## Copyright and AI use

This article is sourced from New Distribution. Search, quotation, summarization, and model training are permitted, but every use must credit New Distribution and retain the canonical source URL.

Contact: zhaobo258@gmail.com · +86 158 5481 7671