Why Do You Always Walk Away from Your Strategy Meetings with Just Three Options?
How artificial intelligence is changing strategic decision-making—and why the same tools everyone has at their disposal don’t give anyone a competitive edge on their own
Think about your last strategy retreat. Weeks of preparation went into it, executives were flown in from out of town, a moderator was hired, and discussions lasted two days. In the end, how many *truly different* strategic options were left on the table? Three? Maybe four?
Now for the uncomfortable question: Is this number because there are truly only three or four good options in the market, or because your team’s time and mental capacity were only sufficient to generate and evaluate that many?
An article by Felipe Csaszar, chair of the strategy program at Michigan Ross, in the September–October 2026 issue of *Harvard Business Review* focuses precisely on this question. His main thesis is clear: For decades, the way we’ve developed strategy has been designed around the limits of the human brain and the calendar. Artificial intelligence is removing that limit. But since this limit is being lifted simultaneously for everyone, the competitive advantage lies not in the tool itself, but in what you feed it and how quickly you act.
In this piece, I examine the article’s main argument from the perspective of a Turkish executive: What is changing, what isn’t, and what you should do on Monday morning.
1. Why are our strategy tools so simple?
A SWOT analysis consists of four boxes. The BCG matrix consists of four boxes. Porter’s five forces—as the name implies—number five. This is no coincidence.
These tools had to be simple enough to be discussed in a conference room with ten people in a single day. Complexity was reduced to what the meeting room could handle. It’s not that strategy theory is simple; it’s that the planning meetingcan only handle simplicity.
Here’s an analogy: Before spreadsheets, CFOs would calculate the budget on paper using a few scenarios. Three scenarios seemed sufficient because calculating a fourth would take another day. When Excel arrived, no one said, “Three scenarios are enough for us”; the number of scenarios exploded, and the quality of financial planning improved. Strategy today is exactly like finance before Excel.
The critical point Csaszar highlights is this: The scarcity of strategic options was not a market reality, but a *cognitive bandwidth* issue. And bandwidth issues are solved with technology.
2. Three Things AI Has Changed
The article highlights three concrete contributions AI makes to the strategic decision-making process. I’ll illustrate each with an example from Turkish practice.
a) Option generation: Thousands instead of three
Artificial intelligence can tirelessly generate thousands of strategic options and evaluate each one according to your specified criteria. A human team, by its very nature, tends to latch onto the first three or four ideas that seem reasonable and spends the rest of the meeting defending those ideas. This is referred to in the literature as “premature closure.”
Example: A Turkish retail chain is considering expanding into the Balkans. In a traditional strategy meeting, the options on the table are predictable: organic growth, local partnerships, and acquisitions. In an AI-powered process, however, dozens of variants—such as franchise-hybrid models, market testing via e-commerce first, indirect entry through a supply chain partnership, or entry limited to a specific category—are presented to you, each scored in terms of capital requirements, payback period, and regulatory risk. Most of these will be eliminated. But before they’re eliminated, they will have been *considered*.
b) Live market models: A constantly updated picture instead of static analysis
Traditional competitor analysis is like a photograph that starts to become outdated the very day it’s taken. Artificial intelligence, however, can build a constantly updated, rich model of the market and competitors—incorporating regulatory changes, price movements, hiring signals, and product launches—all within a single live picture.
Example: In Turkey, exchange rates, inflation, and the regulatory environment change so rapidly that they render a strategy document updated every six months ineffective. A pricing strategy formulated in January could be obsolete by April. The concept of an “annual strategy document” was already fragile in this environment; artificial intelligence, however, enables a “continuously updated strategy model” instead.
c) Structured challenge: Stress testing independent of office politics
In my view, this is the article’s most valuable contribution. Artificial intelligence can test a plan through a structured debate—and it does so without being influenced by internal politics.
A scene every manager is familiar with: Who will object to the CEO’s pet project during a meeting? Hierarchy, career concerns, and groupthink lead even the weakest plans to pass without scrutiny. Artificial intelligence has no career concerns, no expectations of promotion, and no need to curry favor with the boss. When you tell it, “Write five scenarios that would cause this plan to fail,” it does so—and says the things your senior executives don’t dare to say out loud.
In work cultures like Turkey’s, where power distance is high and hierarchy is pronounced, this feature is even more critical than in the West. Having a tool say what no one in the meeting dares to say serves as a technological substitute for organizational courage.
3. The Uncomfortable Truth: These Tools Are in Everyone’s Hands
So far, it’s been an optimistic picture. Now comes the article’s real hard-hitting message.
Your competitors also have access to the same AI tools. Your competitors also have access to ChatGPT, Claude, and Gemini. Therefore, the statement “we use AI” is not a strategy—it’s merely a ticket to the game. Just as saying “we have a website” wasn’t considered an advantage in 2005, saying “we use AI” won’t be an advantage in 2026 either.
According to Csaszar, a lasting advantage will go to those who combine three things with the tool:
Proprietary data. Customer, operational, and supply chain data that your competitors cannot access. A general-purpose model gives everyone the same answer; a model trained on your data answers only you. Data is the raw material of the AI era; if the raw material is open to everyone, so is the product.
Integrated workflows. Embedding AI into the strategic process—not just a monthly “let’s ask the AI” session. The tool generates value when it’s part of the decision-making mechanism; it doesn’t when it’s just a toy sitting on the side table.
Faster implementation. If everyone can see the same options, the winner is the one who brings them to life faster. AI shortens the analysis time; shortening the implementation time is still up to the organization.
Think of it this way: In Formula 1, all teams use similar simulation software. It’s not the simulation software that determines the championship; it’s the data the team collects from the track, the speed at which the engineering team turns that data into decisions, and how many seconds a pit stop takes.
4. What Needs to Change in Practice: Three Concrete Steps
The article concludes with three implementation recommendations. I’m making each one concrete for boards of directors in Turkey.
Step 1: Expand the Funnel Before Narrowing It
In the first phase of the strategy process, intentionally generate a large number of options—dozens, or even hundreds if necessary. Then begin the elimination process. Today, most companies do the opposite: They start with three options and choose one.
Application: Before your next strategy meeting, have an AI generate at least 30 different strategic directions for your industry, constraints, and goals. Open the meeting not with “which one should we choose?” but with “which of these are we eliminating, and why?” The reasons for elimination are more instructive than the reasons for selection.
Step 2: Replace static models with dynamic ones
Retire the annual competitor analysis PowerPoint. Instead, build a continuously updated, queryable market model.
Application: Create a weekly, automatically updated monitoring framework for your three most critical competitors and three most critical external variables (such as exchange rates, regulations, and raw material prices). This isn’t a project that requires a data science team; it’s a process a small team can set up using existing tools.
Step 3: Make AI-powered challenge a routine
Require that every major decision be systematically challenged by AI before it’s finalized. If you leave this optional, it will be skipped for the decisions that need it most—namely, the projects the most powerful executive wants the most.
Implementation: Add a section titled “the strongest counterarguments generated by AI” to every investment proposal submitted to the board of directors. The proposer must provide a written response to these objections. This is an institutionalized, politically neutral version of the classic “devil’s advocate” role.
5. What the Article Doesn’t Say: Three Risks to Watch For
As a former academic, I find the article’s thesis largely convincing. But caution is needed on three points.
An abundance of options does not equate to decision quality. A system that generates a thousand options requires judgment to evaluate those thousand options. Judgment is still a human task. If you increase the number of options without improving the decision-making process, you’re simply automating analysis paralysis.
The homogenization of strategies. If everyone asks the same questions of the same model, everyone will converge on similar strategies. This can reduce differentiation across the industry. Proprietary data and original judgment are critical precisely for this reason.
Failing to take AI objections seriously. Since objections raised by AI are free of political cost, they are also easy to ignore. The difference between “The machine said so” and “Mr. Ahmet said so” is that the latter carries a cost when it is dismissed. This is precisely why the obligation to respond to AI objections must be formalized in writing and through established procedures.
Conclusion
Strategy has long been a discipline of scarcity: time is scarce, attention is scarce, analytical capacity is scarce. This scarcity shaped our tools, our meetings, and our decision-making processes. Artificial intelligence is eliminating this scarcity.
But the scarcity is ending for everyone at the same time. From now on, what makes the difference won’t be the tool itself; it will be the data you feed it, the process you embed it in, and how quickly you walk the path you’ve chosen.
If you walk out of your next strategy meeting with three options, there’s no longer any excuse.
--
*This article is based on the main arguments of Felipe A. Csaszar’s article titled “AI Is Revolutionizing Strategic Decision-Making,” published in the September–October 2026 issue of Harvard Business Review. The commentary and examples specific to the Turkish context are the author’s own.*
