AI scenario planning for marketing: build a practical what-if playbook

Choose the decision and the time horizon

Scenario planning becomes vague when the team tries to simulate everything at once. Start with one decision: entering a new segment, changing positioning or launching a campaign. Set a useful planning horizon and identify what the team can actually change within it.

For an illustrative small software launch, the decision might be whether to emphasize support or low price over the next campaign cycle. The team can change copy, onboarding and channel allocation. It cannot control a competitor’s announcement or the wider economy. That distinction helps keep the playbook actionable.

Pick uncertainties that could change your action

List what you know, what you assume and what could materially change the decision. Examples include a competitor discount, customer concern about implementation effort or limited delivery capacity. Avoid building different scenarios that all lead to your preferred conclusion.

Select a baseline with no major new shock and two alternatives with clearly named changes. A price-pressure scenario could introduce a competing discount. A trust-pressure scenario could introduce a challenge to a public claim. These are hypothetical situations for planning, not assertions that either event will happen.

• Baseline: current offer and known constraints.
• Price pressure: a competitor makes a lower-cost offer.
• Trust pressure: buyers question the evidence behind your claim.

Explore the scenarios in Mirror

Create a source packet with the offer, audience research, public competitor information and your constraints. Upload it in Mirror and describe the baseline. Review the knowledge graph for missing stakeholders and incorrect relationships before interpreting the simulation report.

Run the alternative situations separately and record exactly what you changed. Keep the source packet and evaluation questions consistent. Mirror’s generated results may vary even when the brief changes little, so use repeated exploration to inspect fragility rather than claiming a precise effect from a single run.

A reusable what-if planning brief

Turn the report into signals and responses

A useful playbook connects each scenario to evidence the team can monitor. For price pressure, that might be repeated pricing objections in qualified sales calls. For trust pressure, it could be recurring questions about a specific claim. A generated narrative alone is not a signal that the scenario is happening.

Write a response option, an owner and a review date. Favor reversible actions while uncertainty is high. Improving an explanation or publishing a sample deliverable may be easier to reverse than a permanent price cut. Define what evidence would cause the team to stop or change the response.

Review the playbook against reality

When new evidence arrives, compare it with the assumptions behind the scenario. Do not mark a scenario as accurate merely because one detail looks similar. Ask whether the proposed reaction sequence occurred and whether the suggested response was useful.

Keep a record of unexpected outcomes. They can reveal missing stakeholders or a misleading source packet. The goal is not to prove that the AI foresaw the future. It is to make uncertainty explicit and prepare a team to respond thoughtfully when circumstances change.

Common questions