From Markstrat to Agentic AI
What a Classic Marketing Simulation Still Teaches Us About Modern Growth
Some lessons age badly. Others just change clothes.
Years ago, during my studies, I worked with Markstrat Online, a business simulation developed to help students translate theory into decisions. On paper, it was about markets, segments, budgets, product launches, communication, and sales force allocation. In reality, it was about something much bigger: how companies make good decisions when information is incomplete, competition is moving, and every choice affects five other choices at once.
That is exactly why the topic still matters today.
Because even if the tools have changed, the real management challenge has not. Whether you are steering a fictional Sonite and Vodite portfolio in a simulation or building modern cloud, AI, and platform strategies in the real world, you still need the same core capabilities: clear goals, good market understanding, disciplined execution, and the ability to adapt faster than the environment changes.
And now, with Agentic AI entering the stage, this old topic suddenly feels very current again.
Not because AI replaces marketing strategy.
But because it changes how quickly, how intelligently, and how continuously strategy can be translated into action.
Why Markstrat still matters in a cloud and AI world
Markstrat Online was designed as a decision laboratory. Six companies compete in two markets, Sonite and Vodite. One of them is already developed and segmented. The other is more immature, less transparent, and more uncertain. That setup alone makes it a pretty good metaphor for real life.
In the real world, companies are almost never operating in a fully stable market. One market is established but crowded. Another is emerging but unclear. One product is your safe cash generator. Another one is your bet on the future. You never manage in a vacuum. You manage in overlapping realities.
That was already visible in Markstrat. Product development, communication, distribution, market research, profitability, and competitive positioning were all deeply connected. You could not make one “simple” decision without touching many others. In other words: the model rewarded integrated thinking, not isolated optimization.
That is also why I find it so relevant today.
Modern digital business works exactly like that. If you build an AI product, a cloud platform, or a new digital service, you do not just need good technology. You need:
a clear target system,
a deep understanding of segments and needs,
a communication strategy,
pricing and product choices,
a route to market,
and a feedback loop that tells you whether the whole thing is actually working.
The difference today is that we have much more data and much faster tools. And that is where AI enters the conversation.
Good strategy always starts with goals
One of the most important ideas in the original work was the role of goal setting. That may sound dry, but it is the opposite. If your goals are vague, everything that follows becomes noise. If your goals are clear, decisions suddenly become much easier.
That was already true in Markstrat. Our simulated company needed to decide whether it wanted market share, profitability, long-term market entry, or a mix of those. Those goals then shaped everything else. Product design, advertising, resource allocation, and sales priorities all followed from that initial choice.
That is still true today, and maybe even more important.
Many companies say they want growth. Others say they want innovation. Others say they want AI transformation. But those are not yet good goals. They are broad directions. Real goals need to make trade-offs visible.
Do you want growth at all costs, or efficient growth?
Do you want market share, or premium margin?
Do you want fast AI adoption, or trustworthy and governed AI adoption?
This is where many current transformation programs still struggle. Too often, companies jump into execution before they have built a real goal system. Then everything starts to drift. Teams optimize locally. Budgets scatter. Initiatives overlap. And suddenly there is a lot of activity, but not much direction.
That is why I still like the old logic from Markstrat. It reminds us that target definition is not bureaucracy. It is architecture.
And now imagine adding Agentic AI into that equation.
An agent without clear goals is just expensive motion.
But an agent system built on top of a well-defined goal architecture can become incredibly powerful. It can monitor signals, suggest actions, optimize campaigns, support product decisions, and detect deviations from the intended path. In that sense, Agentic AI does not remove the need for strategy. It makes good strategy even more important.
Communication is not decoration, it is part of the product
A large part of the original paper focused on communication. Rightly so.
In Markstrat, communication was not an isolated ad budget exercise. It was deeply linked to target groups, market perception, product design, and eventual buying behavior. Attention, interest, desire, action — the classic logic still applied. But the real lesson was deeper: communication only works if it is aligned with who you want to reach, what they care about, and what the product actually stands for.
That is still the case now.
In today’s markets, communication has become faster, more fragmented, and more measurable. But it has also become more unforgiving. The market sees through generic messaging much faster. Audiences are overloaded. The distance between message and product reality has become smaller.
This is where modern teams need to think differently.
Communication is not just the wrapper around the product. It is part of the actual experience. Your landing page, your onboarding flow, your AI assistant tone, your pricing explanations, your GitHub documentation, your support replies — all of that communicates.
And once again, this is where Agentic AI starts to become very interesting.
Because AI agents can already help analyze communication performance, generate content variants, monitor reactions, summarize customer sentiment, and propose adjustments almost in real time. That changes the communication process from a campaign rhythm into a living system.
But there is a trap here.
If you let AI scale communication without strong goals and clear brand logic, you simply produce more noise, faster.
So the winning model is not “let the AI write everything.”
The winning model is “let humans define meaning, direction, and relevance — and let AI increase speed, variation, and responsiveness.”
That is a huge difference.
It is also exactly the kind of balance that the old Markstrat logic already pointed toward: communication only works if it is guided by strategic clarity.
Segmentation still wins — and AI makes it more dynamic
One of the smartest parts of the simulation was the market segmentation. Sonite customers were broken down into different groups with different needs. Vodite segments were less mature and more difficult to read. That meant companies had to decide whom to target, how to design products, and where to focus communication.
That is still a timeless challenge.
You cannot win by being vaguely relevant to everyone.
You usually win by being sharply relevant to someone.
In the analog world, segmentation was often slower and more static. Surveys, reports, research studies, maybe a few focus groups. Today, segmentation is much more fluid. Behavioral data, usage signals, conversion data, customer support interactions, and AI-driven pattern recognition allow companies to update their view of segments much more dynamically.
That is a big shift.
And it opens the door for Agentic AI in a very practical way.
Imagine a system that continuously:
observes shifts in demand patterns,
detects that one segment responds differently to a certain feature set,
adjusts communication recommendations,
and alerts product owners when a formerly strong segment begins to lose interest.
That is no longer classic quarterly market research. That is living segmentation.
But again, the principle itself is not new. The simulation already taught the same idea: understand the segments, align the product, align the message, and then monitor whether reality confirms your assumptions.
The tools changed. The discipline did not.
Distribution and sales force thinking now live inside digital channels
A surprisingly strong part of the old work dealt with sales force sizing and allocation. In Markstrat, the simulated companies had to decide how many salespeople to hire, which products they should support, and in which channels they should operate. Those decisions were expensive, strategic, and not easily reversible.
That may sound like an old-school sales topic. But the logic is still highly relevant.
Today, many companies no longer scale first through traditional field sales. They scale through platform channels, digital marketing engines, partner ecosystems, marketplaces, inside sales, product-led growth motions, and increasingly AI-supported service interactions.
But the core question remains the same:
Where do I place my scarce commercial energy to create the biggest effect?
In 2010-style simulation language, that meant sales force allocation.
In today’s language, it means route-to-market design.
The same trade-offs still exist:
Do I invest more in direct enterprise sales?
Do I build partner leverage?
Do I increase digital acquisition?
Do I support one flagship product harder than another?
Do I focus on emerging demand or defend the installed base?
This is also where AI agents can become highly practical. Not as some magical “AI sales manager,” but as a layer that helps prioritize, simulate, and optimize.
For example, agents can:
surface which accounts or segments deserve more attention,
simulate the likely impact of different allocation choices,
detect underperforming channel mixes,
and generate route-to-market recommendations based on actual performance data.
Again, same fundamental management problem.
Just better instrumentation.
What the simulation taught me about portfolio thinking
One of the most interesting parts in the original work was the shift from Sonite to Vodite. The company initially used one product as a strong market-share driver, while another product was gradually deprioritized. Then new products entered the emerging market and became the new strategic focus.
That is classic portfolio management.
It is also something I see all the time in enterprise technology.
A company has one mature system that still makes money. Maybe even very good money. At the same time, it needs to invest in a younger, less proven area that may define the future. That is never easy, because legacy pays the bills and innovation eats budget.
Markstrat made this tension visible in a simple way.
The old market was established but crowded.
The new market was uncertain but full of upside.
That is exactly the same tension many organizations face now:
traditional application portfolio vs. AI-native products,
legacy ERP vs. cloud platform,
stable cash cow vs. question mark.
The smart move is rarely “kill the old thing immediately” or “bet everything on the new thing blindly.” The real challenge is sequencing. Knowing when to harvest, when to invest, when to double down, and when to exit.
That is another reason why I think this old piece of work still matters. It trained the same muscle we need today: strategic resource allocation under uncertainty.
And in the cloud and AI era, that muscle becomes even more valuable because the speed of change is higher, not lower.
What Agentic AI changes — and what it does not
So let’s connect this clearly to today.
Agentic AI is not just about answering prompts. It is about systems that can perceive, reason, plan, and act across steps toward an objective. In a business context, that means AI is starting to move from tool to active participant in workflows.
In product management, marketing, and go-to-market execution, this can be a huge acceleration layer.
Agents can help gather competitive signals.
They can summarize customer research.
They can optimize pricing experiments.
They can support campaign orchestration.
They can flag underperforming channels.
They can prepare decision options for managers faster than any manual team can.
That is powerful.
But what Agentic AI does not change is the need for managerial judgment.
It does not remove ambiguity.
It does not eliminate trade-offs.
It does not define the company’s ambition for you.
And it definitely does not replace the need for a coherent goal system.
In fact, the better the AI gets at acting, the more dangerous weak strategic framing becomes.
A badly aligned human team creates chaos slowly.
A badly aligned agent system creates chaos much faster.
That is why I think old-school management frameworks still matter. Simulations like Markstrat force you to understand the mechanics behind decisions. They make you think in systems. They teach you that goals, segments, communication, product, and distribution are linked.
That systems thinking is exactly what we need now, because Agentic AI only becomes truly useful when it is embedded in a well-architected strategic environment.
The old lesson that still holds
If I look back at this old draft today, the specific categories may feel dated. Sonites and Vodites. Sales force allocation. Survey reports. Advertising budgets in simulation rounds.
But underneath all of that is a lesson that aged extremely well.
Great companies do not win because they do one thing in isolation. They win because they connect many decisions into one coherent system.
They know what they want to achieve.
They understand who they serve.
They align product and communication.
They invest where the future is.
They learn faster than competitors.
And now, increasingly, they use AI not just to analyze, but to act.
That is what makes this old work still relevant for me.
Because even if the technology stack moved from simulation software to Azure, Microsoft Fabric, GitHub, Copilot, and Agentic AI, the leadership challenge is still beautifully familiar:
turn information into direction,
turn direction into decisions,
and turn decisions into results.
Conclusion
Markstrat Online may have started as an academic simulation, but the logic behind it remains very real. The need for structured goal setting, segment-focused execution, communication discipline, portfolio thinking, and performance control has not gone away. If anything, it has become more important now that markets move faster and AI can accelerate both good and bad decisions.
That is why I still find this kind of work valuable today.
It reminds us that strategy is not an abstract PowerPoint exercise. It is an operating system for decisions.
And in the age of Agentic AI, that operating system needs to be clearer than ever.
Stay clever. Stay strategic. Stay adaptive.
The Cloud Advisor,
Uwe Zabel
🚀 Curious how classic strategy models, Agentic AI, and modern cloud execution come together? Follow my journey on The Cloud Advisor’s book of stories—where cloud, AI, and business strategy converge.
Or ping me directly—because building the future works better as a team.