What Do You Know That We Don’t?

Marketing teams are remarkably good at producing content. They have editorial calendars, campaign plans, production workflows, distribution channels and dashboards that measure everything from impressions to conversions.

Now they have AI, which can generate more material, more quickly, at a fraction of the previous cost. But here's a question that rarely appears in the annual content planning process: What does your company know that nobody else does?

Not what does it believe, or claim, or predict. What does it actually know, based on evidence that it has collected, analyzed and interpreted?

The distinction matters because the cost of producing content is collapsing while the value of original knowledge is rising. Anyone can publish another AI-generated article about the future of AI. Far fewer can tell you what enterprises are actually doing with AI, how their investments are performing, and what has changed over the past year.

A content calendar answers the question, "What should we publish next?" A research program answers a more fundamental question: "What can we discover that the market doesn't already know?"

The production trap

The marketing industry has become extraordinarily efficient at generating material. According to the Content Marketing Institute's Technology Content and Marketing Trends: 2026 Insights, 95% of surveyed technology marketers using AI for content creation employ it to generate or optimize written content. Among those users, 90% report productivity improvements.

But only 35% say AI-assisted creation has improved content performance. Eighteen percent say it has actually reduced content quality. These numbers suggest an uncomfortable possibility: marketing organizations are becoming better at producing content without becoming correspondingly better at producing something worth reading.

The cost of producing content is collapsing while the value of original knowledge is rising.

The problem predates generative AI. For years, content marketing has largely been organized around the demands of distribution. Search algorithms needed fresh material. Social platforms rewarded frequency. Marketing automation required a steady supply of assets for each stage of the funnel. The content calendar became both the organizing principle and the production mandate. 

AI has made that system dramatically more efficient. It hasn't necessarily made the underlying thinking more valuable. A company can now publish five articles where it once published one. But if all five recycle publicly available information, what has been added to the market's understanding?

Volume is not the same as knowledge. And knowledge is not the same as authority.

What research creates

I've spent much of my professional life trying to understand things that weren't yet well understood.

That began with ethnographic fieldwork in Lahore, Pakistan, where I studied urban culture through the intersecting lenses of class, caste and Islamic nationalism. The objective was to build a coherent interpretation of that culture from original fieldwork.

Decades later, at ServiceNow, I was applying different research methods to a very different question: What were enterprises actually doing with generative AI, and what measurable value were they getting from it? The project became the Enterprise AI Maturity Index.

ServiceNow's Enterprise AI Maturity Index 2025, developed in partnership with Oxford Economics.

Initially, it was something of a skunkworks experiment. We wanted to understand how organizations were adopting generative AI, where the technology was producing practical benefits, and where the promised returns weren't materializing. We weren't simply asking executives what they thought about AI. We were investigating what their organizations were doing with it. 

As the research continued, successive waves allowed us to compare results over time. We could identify emerging patterns, track changes in adoption and distinguish between the ambitions companies expressed and the outcomes they reported. That longitudinal dimension was crucial. A single survey provides a snapshot. Repeated research can reveal a trajectory.

Over time, the Index evolved from an editorial research initiative into a franchise that helped power ServiceNow's go-to-market strategy. We published findings on our owned channels, promoted them through corporate and executive social media, pitched them to journalists and supplied sales teams with evidence they could use in customer conversations.

Senior leaders cited the research during mainstage presentations at Knowledge, ServiceNow's annual conference in Las Vegas. Findings appeared on signage across the show floor. By then, the Index was doing something that no content calendar could accomplish on its own. It was giving an entire organization a growing body of evidence around which to build a market narrative.

Beyond the report

This is where companies often misunderstand the value of proprietary research.The immediate temptation is to think about deliverables. A research report can become a series of blog posts, infographics, webinars, executive presentations, press pitches and social media campaigns. Agencies routinely promote the idea of turning one study into a year's worth of content.

There's nothing wrong with that. We did it at ServiceNow. But it gets the relationship backward.The real asset isn't the collection of content that research generates. It's the underlying evidence, and the interpretation of that evidence, that makes the content worth consuming.

Don't build a research program to feed the content calendar. Build a content calendar to distribute what the research program discovers.

Consider the difference between publishing an article about AI adoption and maintaining a dataset that tracks actual enterprise AI adoption over several years.The article can summarize conventional wisdom. The dataset can challenge it.

The article competes with thousands of other articles addressing the same topic. The dataset has the potential to become a reference point that journalists, analysts, executives and customers return to.

The article has a publication date. The research program has a history, a methodology and the possibility of becoming more valuable with each successive wave. That's what makes it infrastructure rather than simply another marketing campaign.

The evidence test

In a recent Field Note, "The buyers you never see," I argued that the most consequential audience for B2B thought leadership may be people who never attend a vendor presentation.

Finance, security and technical stakeholders increasingly shape enterprise purchases after the vendor has made the shortlist. Their job is not to admire the presentation. It's to test the proposition.

According to G2's 2026 Buyer Behavior Report, evaluation has become the longest stage of the software buying journey. Forty percent of buyers identify it as their biggest time sink.

What happens when a prospective customer's internal champion must defend a purchase to the CFO? Assertions about innovation and transformation won't get very far. The champion needs evidence.

Research serves two audiences: people making decisions and the machines helping them gather information.

That might include credible performance benchmarks, quantified outcomes, comparisons with peers or an independently defensible account of the problem the product addresses.

Original research can supply that evidence. It also matters before the shortlist is formed. As I argued in "Citation Is the New Page Rank," generative answer engines increasingly mediate how buyers discover and evaluate vendors. Statistics, attributable findings and verifiable claims give those systems something concrete to retrieve and cite.

A generic opinion about the future of work is interchangeable with countless others. A well-documented finding about how workforce practices have changed can become part of the evidence that both humans and machines use to understand a market.

Research can therefore serve two audiences simultaneously: people making decisions and the machines helping them gather information. Neither audience needs another polished restatement of conventional wisdom.

Survey ≠ research

Commissioning a survey doesn’t automatically create authority. The market is full of branded surveys built around loaded questions, unrepresentative samples and findings engineered to validate a sponsor's marketing claims.

In short: A branded statistic isn't necessarily a credible fact. Useful research starts with a consequential question, employs a defensible method and remains open to inconvenient answers.

In the ServiceNow research, the absence of measurable AI returns was potentially as interesting as evidence of success. A study that can only confirm the sponsor's preferred narrative isn't doing much investigative work.

If your company stopped publishing tomorrow, what would the market no longer know?

Nor does every company need a major longitudinal survey. Proprietary insight can come from product telemetry, anonymized customer data, operational experience, structured interviews or systematic field observation. The method matters less than the question and the integrity of the answer.

And original research should not be confused with exclusive ownership of a truth. The ultimate test is whether outsiders find the evidence credible and useful enough to adopt, challenge, reference and build upon. That's how a company's knowledge begins to acquire authority beyond its own channels.

Flip the script

As marketing organizations enter their planning cycles, the instinct will be to ask how AI can make next year's content program more productive.

That's a reasonable operational question. But it isn't the most important strategic one. A better starting point is to identify what the company is uniquely positioned to know.

What patterns can it observe that competitors cannot? What customer problems does it understand particularly well? What evidence could it collect systematically, and what might that reveal about the direction of its market? Only then should the editorial planning begin. 

A research program may yield an annual benchmark, quarterly findings, executive commentary, conference presentations and a continuous stream of material for sales and marketing. But those outputs should follow from the knowledge being developed, rather than determine what gets investigated.

In other words, don't build a research program to feed the content calendar. Build a content calendar to distribute what the research program discovers. A company that produces original, credible knowledge gives customers a reason to pay attention, journalists something to report, analysts something to evaluate and executives something substantive to say.

It may also give the market a new way to understand a problem, a category or an emerging technology. That is the beginning of authority. And it suggests a useful test for any marketing organization:

If your company stopped publishing tomorrow, what would the market no longer know?