What I learned building Kyndryl marketing from zero

The situation on day one

Kyndryl spun out of IBM in November 2021 as the world’s largest IT infrastructure services company, with $16 billion in annual revenue and approximately 90,000 employees. What the spin-off did not include was any meaningful digital marketing presence.

Before the separation, the IBM brand carried everything. IBM.com ranked for the keywords. IBM’s domain authority provided the SEO foundation. IBM’s content library created the reference ecosystem that drove traffic to IBM Cloud and IBM services properties. When Kyndryl became an independent company, it inherited the revenue base and the workforce. It did not inherit the brand equity, the traffic, or any of the digital infrastructure that had been built over decades inside IBM.

On day one we had a new domain, zero accumulated SEO authority, an empty content library, no marketing technology platform, no paid media infrastructure, and a customer base that had been IBM customers and now needed to understand what Kyndryl was and what Kyndryl offered.

The pressure was immediate. A business that size needs pipeline to operate. The board needed to see that Kyndryl could stand on its own as an independent entity in the market. And the marketing team was still being assembled. We had no runway.

This is what I mean when I say we built from zero. Not from low. From zero.

Sequencing: the decision that determined everything else

In a zero-to-one situation, the most important strategic decision you make is about sequencing. Everything cannot happen at once. Resources are constrained, team capacity is limited, and attempting to do everything in parallel produces a situation where nothing is done well enough to compound.

The temptation is understandable. There is urgent pressure from the business for pipeline, from leadership for market visibility, and from the board for evidence that the marketing function can perform independently. All of that pressure pushes toward trying to do everything immediately.

We did not do that. We made explicit sequencing decisions and held to them.

First ninety days: core infrastructure. We built the minimum viable marketing technology stack to capture and route leads. We stood up basic paid search capability in the highest-priority markets. We launched the SEO foundation: the technical structure of the site, the initial taxonomy, the content architecture that would tell search engines what Kyndryl was and what problems it solved. We did not try to be comprehensive. We tried to be functional.

The principle was: build the floor before you build the ceiling. We needed the foundation to be solid before adding capability on top of it. Every hour spent on content programs or social media infrastructure in the first ninety days was an hour not spent on the foundation that everything else would depend on.

Months four through nine: analytics and content. Once the core infrastructure was stable, we invested in two things: the analytics layer and the content architecture. These were sequenced together because they are dependent on each other. The analytics layer tells you what content is working and why. The content architecture gives you the material to analyze.

The analytics investment in this phase was not sophisticated by later standards. But it was accurate and trusted. We built the reporting frameworks that the marketing leadership team used to make decisions, and we made sure those frameworks were credible to the finance organization from the beginning. The trust work starts early or it does not happen.

Months ten through eighteen: expansion and optimization. With the foundation built and the analytics layer providing a reliable picture of what was working, we expanded. Additional channels. Additional markets. Additional content programs. Additional team capacity. This is the phase where the compounding effects of the foundation work became visible. The SEO authority we had been building for nine months started to produce exponential traffic growth. The attribution framework we had built allowed us to reallocate paid media toward the channels that were actually producing pipeline.

By month eighteen, Kyndryl.com was driving 11.9 million annual visits. Not because we were spending more than comparable companies. Because we had built the right foundation in the right sequence and then expanded intelligently.

Building the team: the decisions that held up and the ones that didn’t

Team construction in a zero-to-one build is different from team construction in an established function. In an established function, you are filling defined roles. In a zero-to-one build, you are making bets about what capabilities matter most and in what order.

The bets we made that held up.

We hired for ownership over expertise. The people who performed best in the early Kyndryl build were not necessarily the ones with the deepest functional expertise. They were the ones who were comfortable operating in ambiguity, making decisions without complete information, and owning outcomes rather than activities. At an established company you can hire specialists. In a build situation, you need people who can hold more than one thing at once.

We used agencies for execution, not strategy. The agency model works when the agency is executing a strategy owned and directed by the in-house team. It does not work when the agency is setting direction. The agencies we used were selected for their execution quality in specific channels: paid search, SEO, content production. They operated inside frameworks and with briefs that came from the in-house team. The ones that tried to own strategy became a problem quickly.

We made the analytics hire relatively early. In retrospect, it should have been even earlier. The analytics function turned out to be the highest-leverage investment we made in the team. The person in that role shaped every subsequent decision about where to invest and where to pull back. Hiring that role late means operating without a reliable decision-making framework for a longer period than necessary.

The bets that did not hold up.

We hired some roles too early at too senior a level. In the first twelve months, what you need is people who can do the work. Senior leadership can be hired after the function is established enough that there is something to lead. Some of the early senior hires produced management overhead without proportionate output.

We tried to build the social media program too early. In a zero-to-one build, organic social is low leverage in the first year. It takes time to build an audience, the attribution is hard, and the opportunity cost is high when the paid and organic search programs are not yet mature. We invested earlier and more heavily in social than the results justified.

The inflection points

The build did not progress linearly. It progressed in recognizable steps, and understanding what drives each step changes how you run the program.

The 90-day mark. The program was functional. Traffic was growing from a base of near zero. Leads were being generated and routed correctly. The infrastructure was stable. The numbers were not impressive in absolute terms, but the trajectory was positive and the foundation was solid. This was the moment when the organizational narrative shifted from “we are building this” to “we are operating this.”

The six-month mark. The compounding effects of the SEO investment started to become visible. Organic traffic was growing faster than most of the business had expected, because most of the business had not seen organic traffic built from scratch before. The search engine takes several months to develop confidence in a new domain, and then the growth curve steepens. At six months, the steepening became visible.

The twelve-month mark. The program had produced enough data to make intelligent channel allocation decisions. We knew which paid media channels were producing pipeline at an acceptable cost. We knew which content programs were driving organic traffic and which were not. We began a significant reallocation based on that data, pulling budget from underperforming areas and concentrating it in programs where the results were verifiable.

The eighteen-month mark. 11.9 million annual visits. The performance media program had scaled to $6 million in annual spend. The team had grown to 30 people globally. What had been a build was now an operating function.

What I’d do differently

Invest in the data layer earlier. We built the analytics and data infrastructure in parallel with the content and media programs. It was always slightly behind. In the first nine months, we were consistently making decisions on data that was less reliable than it should have been, because the data infrastructure had not caught up to the operational complexity of what we were running.

If I made this decision again, I would establish the data layer as the first investment, not a concurrent one. The cost of building it first is a slower start on the programs. The benefit is making every subsequent decision on better information. The payoff on that investment is not visible in month three but it is very visible in month twelve.

Hire the analytics lead on day one. The analytics function was the most leveraged hire in the organization. It should not have been the fourth or fifth hire. It should have been the first.

Be more aggressive about deprecating programs that are not working. In a build situation, there is a natural reluctance to declare something a failure, because the organization has invested in it and the team wants the chance to make it work. The opportunity cost of a program that is consuming resources without producing results is real, and the programs that were not working in month six were almost never working in month twelve. The earlier you make the reallocation decision, the more of the program you can save.

Build the governance framework for the martech stack earlier. By the time we were deploying sophisticated AI tools in 2023 and 2024, the martech stack had grown to a point where governance was a significant overhead. Decisions about which tools could access which data, what the approval process was for new tool deployments, and how platform changes were evaluated and approved took organizational bandwidth that could have been avoided with earlier investment in governance frameworks.

The Kyndryl build was the most intense professional experience I have had. It was also the most clarifying. You learn what matters in a zero-to-one build in ways that a well-resourced expansion cannot teach you. The sequencing discipline, the team construction principles, the data-first investment philosophy: these are the things I would carry forward and do again, largely the same way. The rest I would do differently.