Better data starts with treating people like people | MarTech

Better data can make marketing more confident without making it smarter. We capture a click, score a lead, infer an intention, and place someone into a segment. The behavior may be real. The meaning we attach to it may not be.

For most of my career, I believed more data would make marketing smarter. Better tools would help us understand people. More precise targeting would make marketing more relevant. Instead, we built systems that often misunderstand people with greater confidence.

People rarely get to inspect the resulting profile, correct what’s wrong, or challenge what a company has inferred. They participate in a transaction while an unseen system creates another version of them. That imbalance is harder to ignore.

More fields don’t mean more truth

I recently requested my profile from a major data broker. It contained 132 fields describing who I am, what I buy, and what the company thought it knew about me. It looked precise. It was detailed. Much of it was wrong.

The experience exposed a basic flaw in our thinking. We treat the amount of data as a proxy for the quality of understanding. A profile containing 132 fields must be better than one containing 12. But more fields create more places to be wrong.

The person described has little opportunity to review the profile, challenge its assumptions, or explain the context behind a behavior. A company creates a version of you, sells access to it, and uses it to influence how other organizations treat you. That isn’t customer understanding. It’s guessing with infrastructure.

Better data needs participation from the person being described. People need the ability to see, correct, and control information about themselves. Organizations need to distinguish what a person disclosed from what an algorithm inferred.

Without that distinction, precision becomes theater.

McDonald’s turns lunch into a 515-page dossier

A recent McDonald’s case shows how far commercial profiling can go. WIRED journalist Reece Rogers requested the information McDonald’s had collected about him through its loyalty program. What came back wasn’t a short transaction history. It was a 515-page dossier.

McDonald’s tracked years of purchases and used them to predict that Rogers would visit 2.16 times during the following six weeks, spend an average of $13.49 per visit, and spend $29.15 in total. It ranked the products he was most likely to order and calculated his likelihood of abandoning the brand.

A person thought he was ordering lunch and collecting rewards. The company was building a predictive model of his future behavior.

McDonald’s states in its privacy policy that it may combine information provided by customers with automated collection and information obtained from other sources. It may use that information to train algorithms, personalize experiences, support targeted advertising, and create profiles that include preferences, behavior, attitudes, and psychological trends.

The problem isn’t that McDonald’s remembers an order. Customers may appreciate that convenience. The problem is the imbalance between the simplicity of the customer’s action and the complexity behind it.

Ordering a Diet Coke shouldn’t require understanding an almost-invisible predictive data system. Joining a loyalty program shouldn’t mean losing the ability to understand what a company has concluded about you.

The customer sees points and a discounted meal. The company sees a long-term behavioral asset. Those aren’t equal levels of understanding.

The giveaway is often the data collection campaign

The same imbalance appears in promotions designed to feel like entertainment.

At Fanatics Fest NYC, I examined sweepstakes and activations from brands including Fanatics, Topps, the NFL, NHL, Honda, American Express, Paris Saint-Germain, and Qatar Airways.

The prizes varied, but the data requests followed the same pattern. Visitors scanned a QR code for a chance to win tickets, merchandise, or a trip.

The forms requested combinations of full name, email address, phone number, complete birth date, gender, country, and city. Some included multiple marketing permissions. Others sent visitors via third-party platforms before they reached the entry form.

A company needs a name and one reliable contact method to notify a winner. The remaining fields help build a marketable identity.

The giveaway is often a data-collection campaign with a prize attached. The customer receives a chance to win something. The brand receives verified identity fields, marketing permission, and information it can match with other records. Even the customer’s time becomes part of the exchange as they stand in a crowded venue completing forms for multinational companies.

The fan sees a game. The marketer sees a profile-building opportunity.

A night at Madison Square Garden creates another kind of profile

Now consider what may happen when you attend a Rangers game, Knicks game, or concert at Madison Square Garden.

You enter a building equipped with facial-recognition technology. Your face may be scanned and compared with internal watchlists. MSG has defended the system as a security measure, but it has also used the technology to identify and exclude attorneys connected to firms involved in litigation against the company.

Recent reporting based on data leaked by ShinyHunters describes something broader: an internal surveillance system that includes risk scores and records of celebrities, politicians, critics, and fans.

The database reportedly included more than 39,000 entries and information such as social media activity, personal associations, race, sexual orientation, and past interactions with MSG staff.

People were reportedly classified according to the risk they posed to the company. Some were flagged because they criticized owner James Dolan, worked for the wrong law firm, or associated with someone MSG considered a problem.

The breach also reportedly exposed a customer database containing more than 10.5 million entries, including email addresses, phone numbers, and birth dates. Separate reporting covers exposed biometric tracking logs, background check information, and online threat assessments.

Think about the difference between the experience a customer believes they purchased and the system operating around it.

You thought you bought a ticket to see the Knicks, Rangers, or a concert. The venue may see an identity to verify, a face to scan, a customer record to retain, an online history to assess, and a potential risk to classify.

Security at a large venue is legitimate. The question is where security ends and corporate surveillance begins. The customer has little visibility into the answer and almost no ability to inspect or challenge the resulting profile.

The McDonald’s dossier predicts what you might order. The MSG system may influence whether you’re allowed through the door.

Each begins with ordinary activity. Each creates a version of you that the company controls.

Behavior doesn’t explain motivation

Ken Beller’s work adds another important distinction. Beller, co-author of The Consistent Consumer and the forthcoming Why They Buy, has spent years arguing that demographics and past behavior don’t explain why people make decisions.

Two customers may buy the same product for completely different reasons. One may be seeking comfort. Another may be seeking status, control, belonging, or relief. Traditional customer data records purchases and assumes that the behavior explains the person. It doesn’t.

Recent work by Beller and J.D. Pincus focuses on directly measuring emotional needs rather than inferring motivation from demographics and transactions. Their MotivationMetrics research reports that directly measured emotional needs predicted consumer behavior two to four times more accurately than traditional demographic and behavioral approaches.

The business implication is significant. Marketers have spent years collecting the “who” and “what” while making assumptions about the “why.”

Direct measurement is better than unsupported inference, but it still requires guardrails. Emotional information may be more accurate and also more sensitive. People should understand what’s being measured, why it’s being collected, and how it’ll affect the way an organization treats them.

Better data isn’t simply data that predicts more effectively. It must also come from a fair relationship with the person who provided it.

The LUMAscape maps the industry — but where is the person?

The LUMAscape has helped marketers understand the machinery of advertising technology for years. It organizes a complicated industry into categories: data providers, identity companies, customer data platforms, publishers, agencies, measurement vendors, and activation tools.

What it rarely makes visible is the person whose information keeps the machinery running.

The consumer appears as an audience to identify, a profile to enrich, a customer to score, or an impression to monetize. Nearly every company on the map has a role in collecting, connecting, interpreting, moving, or activating information about people.

But where is the category for the person?

Where is the infrastructure that allows someone to inspect the information attached to them, correct it, control access to it, and participate in the economic value it creates?

The LUMAscape map reflects how the industry works: Companies exchange data about people while the people themselves remain outside the system.

McDonald’s builds a predictive dossier. MSG creates watchlists and risk assessments. Data brokers assemble hundreds of fields. Martech companies enrich, connect, and activate those records.

The person is present everywhere as data and almost nowhere as an active participant. This is the missing category I describe as data agency: infrastructure that treats the individual as a participant with rights, context, and decision-making power.

Adding another logo to the existing LUMAscape won’t correct the imbalance. The model needs a new starting point. Instead of asking how data moves between companies, we need to ask how information moves with the knowledge and participation of the person it describes.

The next marketing data system shouldn’t begin with the audience. It should begin with the human being.

Tactics are becoming commodities

The tactical side of marketing is easy to reproduce.

Almost anyone can generate competent copy, build a website, create a video, automate an email sequence, and produce hundreds of campaign variations. AI has reduced the cost of content production. Martech platforms have made distribution available to nearly every company.

That doesn’t mean marketing has improved. It means we can make more of it.

A company can publish every day without having an original point of view. It can personalize thousands of messages without understanding one customer. It can fill a dashboard with activity while remaining unable to explain what changed in the business.

Marketing can’t continue defining its value by what it produces. Its value must come from judgment.

Marketing should help the business decide which customer problem matters, which information can be trusted, and which promise the company has earned the right to make.

The hard questions come before the campaign:

• Do we understand the customer’s actual problem?
• Is our data accurate enough to support this decision?
• Did the person knowingly provide the information?
• Does the product deliver what marketing is about to promise?
• Can people see and correct what we believe about them?

If the answers are weak, more marketing will only distribute the weakness faster.

Marketing can’t repair an operating problem

The problems in the McDonald’s and MSG stories don’t begin with advertising. They begin with business decisions about how much information to collect, what to infer, and how much control to give the person being described.

Marketing often arrives after those decisions are made. Its job becomes explaining the loyalty program, promoting the personalized experience, or reassuring customers that surveillance exists for their benefit.

But marketing can’t permanently compensate for an unfair data relationship.

Better creative can’t repair customer mistreatment. Personalization can’t manufacture trust. A new platform can’t make unreliable information accurate. A privacy policy can’t create meaningful understanding if ordinary people can’t reasonably follow what happens to their information.

When the marketing story and operating reality separate, reality eventually wins.

Marketing should identify that separation, not help conceal it.

Politeness doesn’t solve hard problems

Fixing these systems requires people inside organizations to ask questions that may be unwelcome.

• Do we need to collect this information?
• What harm could result if it’s wrong, misused, or stolen?
• Would customers reasonably expect us to create this profile?
• Can they see it, correct it, and delete it?
• Who inside the company can use it, and for what purpose?

Business culture often rewards people for making difficult conditions sound acceptable. Surveillance becomes frictionless security. Behavioral prediction becomes personalization. Massive collection becomes customer intelligence.

The language protects the meeting. It doesn’t protect the customer or the business.

Ignoring a hard question transfers the cost to someone else. The customer loses control. The employee learns to remain quiet. Leadership continues to make decisions with incomplete information. The company faces a growing security and reputational liability.

Researchers call this organizational silence: people withhold information about problems because they believe speaking up is unwise. Research published by the Academy of Management describes how that silence can become a shared organizational behavior.

It may feel safe. The business risk remains.

Radical candor needs both halves

I have written before about how marketing leadership asks for radical candor but often wants compliance.

Kim Scott defines radical candor as caring personally while challenging directly. Both parts matter.

Challenge without care becomes aggression. Care without challenge becomes ruinous empathy: protecting someone from immediate discomfort while allowing a preventable problem to continue.

Radical candor isn’t permission to say anything in any manner. It requires evidence, respect, and a willingness to hear that your interpretation may be wrong.

If I expect partners and stakeholders to accept direct questions from me, I must accept direct questions from them. No one gets immunity, including me.

Earlier in my career, I thought self-awareness meant reading the room and adjusting myself to fit it. Now I see it differently. Self-awareness means understanding the room without automatically surrendering my judgment to it.

I can reconsider my delivery without abandoning the issue. I can admit an error without pretending the original concern had no merit.

That isn’t conflict for its own sake. It’s how incomplete information gets corrected.

Candor is part of better data

Bad data doesn’t begin only inside a database.

It begins when someone notices that an assumption is wrong but decides not to question it. It grows when a McDonald’s customer needs to request 515 pages to understand how lunch became a prediction model. It grows when a fan trades verified identity fields for an undisclosed chance to win. It grows when an MSG visitor can’t see the risk assessment connected to a night out.

It also grows when an employee can’t challenge a conclusion or a marketer can’t tell leadership that the evidence doesn’t support the story.

Every data system reflects the behavior of the people operating it.

If the culture rewards agreement, the information will eventually reflect what leadership wants to hear. If people can challenge assumptions, admit uncertainty, and correct mistakes, the data becomes more useful.

AI will generate more content, predictions, and customer profiles than any human team could create. The answer isn’t more automation built on the same weak assumptions.

Marketing needs better judgment, greater humility, and systems that allow people to participate in how they’re understood.

Better data needs context. It needs consent and correction. Most of all, better data starts with treating people like people.