Small Business Marketing Automation 05: Can AI Turn Research into a Useful Marketing Plan?

 

Market, customer, and competitor research can produce a large amount of information.

It can reveal what frustrates customers, how competitors price and position their offers, and what changes are emerging in the market.

But organizing research does not create a marketing plan. A business still needs to decide what to prioritize, which customers to target, what value to offer, which channels to use, and how success will be measured.

AI can help automate much of the work involved in connecting research findings to business goals, customer problems, channels, messages, campaigns, and performance metrics.

However, AI should not set the business goal or make the final decisions. The business owner or responsible manager must decide and approve the goals, positioning, budget, and scope of execution.

Research Findings and Marketing Plans Are Not the Same

Research findings provide evidence for understanding the current situation.

This may include identifying recurring complaints in customer reviews, comparing competitors’ prices and offers, or detecting changes in the interests of a particular customer segment.

A marketing plan determines what the business will do with that evidence.

The U.S. Small Business Administration recommends considering the target market, competitive advantage, sales methods, goals, action plan, pricing, budget, and performance measurement when developing a marketing and sales plan.

Research findings should therefore lead to decisions about:

• The customer problem to solve

• The business goal to pursue

• The customer segment to prioritize

• The value and offer to present

• The channels to use and the role of each channel

• The campaigns and experiments to run

• The budget and staff required

• The performance metrics to track

Research answers, “What have we learned?” A marketing plan must answer, “What will we do about it?”

AI Can Connect Research Evidence to an Execution Plan

AI is useful for organizing scattered research into a consistent structure and linking each proposed decision to supporting evidence.

A basic sequence might look like this:

Research evidence → Customer problem → Business goal → Value proposition → Channels and messages → Campaign → Performance metrics

AI should not simply summarize the research. It should show how each piece of evidence supports a proposed action.

For example, if the same complaint appears repeatedly in customer reviews, AI can classify it as a possible customer problem. If the business decides to address that problem, AI can suggest messages, channels, campaigns, and measurement criteria that support the approved goal.

Proposals without supporting evidence should be excluded or marked as requiring further validation.

AI should connect approved business goals with relevant evidence and possible actions.

The Business Must Define the Conditions First

External research alone cannot tell AI about a business’s internal goals and constraints.

The business must provide information such as which products are most profitable, whether employees can manage a new campaign, and whether the company can handle an increase in orders or inquiries.

Before generating a plan, the business should define at least:

• The business goal for the plan

• The customer segment to prioritize

• The positioning that must be maintained

• The maximum available budget

• The staff time and resources available

• The actual capacity for sales, bookings, or production

• Claims and language that cannot be used

• Decisions that always require human approval

Without these conditions, AI may create a plan that appears logical but is not realistic.

It might recommend expanding advertising when the business lacks staff to manage additional bookings. It could also prioritize a product because of market interest even though that product produces little profit.

A useful plan requires both research evidence and the business’s internal goals and constraints.

How Does Marketing Plan Automation Work?

Marketing plan automation is not simply a matter of entering a question and receiving one answer.

It requires a workflow that retrieves the necessary business conditions and produces a plan and action list according to predefined rules once the research has been reviewed and approved.

• Trigger: New research findings are reviewed and approved

• Inputs: Research evidence, business goals, budget, staffing, operational capacity, and prohibited actions

• Rules: Separate verified evidence from assumptions and flag proposals with insufficient support

• Outputs: Priorities, campaign briefs, action items, KPIs, and revision conditions

• Authority: AI prepares the draft; the responsible manager approves the goals, budget, channels, and major proposals

• Records: Store the evidence used, assumptions, approvals, and change history

• Stop conditions: Insufficient evidence, budget limits exceeded, inadequate capacity, or conflict with business goals

• Exception path: Send items that cannot be resolved automatically to the responsible person

This structure helps the business apply consistent standards whenever research and planning are repeated.

It also makes it easier to understand why a plan was created and identify what must change when business conditions change.

AI Should Build Scenarios, Not Predict One Certain Outcome

Even strong research cannot predict customer behavior or competitor responses with certainty.

An AI-generated plan should therefore be treated as a hypothesis to test, not a guaranteed forecast.

The plan should distinguish verified facts from assumptions that still need validation. Instead of presenting one outcome as certain, AI can prepare scenarios such as:

• Performance improves more than expected

• The campaign works only for a particular customer segment

• Customers engage with the campaign, but purchases do not increase

• Customers respond negatively to the message or its frequency

Each scenario should be linked to specific metrics and next actions.

If performance exceeds the target, the business may expand the campaign. If the campaign works only for one group, it may narrow the audience. If there is no meaningful response, the message, channel, or original assumption may need to be reconsidered.

Research on current large language models also shows limitations in complex, long-term strategic planning. However, the SPIN-Bench study does not directly evaluate the performance of AI-generated marketing plans. It is included only as supporting evidence that current LLMs still face limitations in complex, long-horizon strategic planning.

This allows the AI-generated plan to function as a working hypothesis that can be tested and revised.

AI Must Also Consider Priorities and Dependencies

Having several good ideas does not mean a business can execute all of them at once.

Advertising expansion, an email campaign, a website redesign, and new content production may all appear useful. But a small business with limited time and money must choose among them.

AI can compare possible actions using criteria such as:

• Expected business impact

• Required cost and time

• Whether the current team can execute the work

• Whether results can be measured

• Whether another task must be completed first

• Whether the action conflicts with another campaign

• Whether it can be stopped if problems occur

• The opportunity cost of alternatives that cannot be pursued

An idea may have high potential but still be impossible to start if customer data is not ready or no one is available to manage it.

AI can identify these dependencies and suggest an order of execution. The business must then decide what to prioritize based on its goals and actual capacity.

A Practical Example

Suppose a small e-commerce business with repeat-purchase products finds the following pattern in customer reviews and sales data:

Customers do not receive enough guidance after their first purchase, which may be preventing repeat purchases.

If the business approves “increasing repeat purchases among first-time customers” as its goal, AI could prepare the following plan:

• Customer problem: Customers do not fully understand how to use the product after purchase

• Proposed direction: Emphasize product-use support rather than additional product promotion

• Channel role: Use post-purchase email to provide usage instructions and answers to common questions

• Test scope: Begin with a limited group of customers whose consent status has been verified

• Performance metrics: Email engagement, repeat-purchase rate, and changes in related customer inquiries

• Expansion condition: The test group produces a meaningful improvement in repeat purchases

• Revision condition: Customers read the emails, but repeat purchases do not change

• Stop condition: Unsubscribes or complaints exceed a predefined limit

AI can convert this plan into a campaign brief and a list of execution tasks.

Automation should therefore do more than generate ideas. It should create a traceable structure that connects evidence, execution, and measurement.

Approved Plans Can Be Converted into Actual Work

Once a plan is approved, AI can convert it into manageable work items.

The outputs may include:

• A campaign brief defining the goal and target audience

• Channel-specific messages and offers

• A list of required content and materials

• Task owners and deadlines

• A proposed budget allocation

• Task dependencies

• Performance metrics

• Conditions for expansion, revision, or termination

When connected to a project management system, approved tasks can be assigned to team members and their status recorded. The system can also notify the responsible person if required materials are missing or a deadline is missed.

However, publishing advertisements, increasing budgets, making new promises to customers, or changing brand positioning should require separate approval.

Automating plan preparation and automatically executing marketing actions are not the same thing.

Plans Must Change in Response to Results

A marketing plan is not a document that should remain unchanged after it is created.

Before execution, the business should record a baseline for metrics such as conversion rate, repeat-purchase rate, inquiry volume, and cost. After execution, actual results should be compared with that baseline.

AI can identify gaps between expected and actual performance and suggest possible causes and additional checks.

It should not present one explanation as certain simply because performance is lower than expected. The message may not address the customer’s real problem. The channel or timing may be unsuitable. Price or product satisfaction may be the more important cause.

Small message changes or adjustments to the test audience may be allowed within predefined limits. Major budget changes, new target audiences, and changes in positioning should require renewed approval.

Not Every Business Needs Complex Automation

The more frequently a business conducts research and creates marketing plans, the more valuable automation may become.

Businesses that manage several products and customer segments, run campaigns regularly, or require different plans for multiple locations may be able to reduce a significant amount of repetitive work.

A business that runs few campaigns or has limited research data does not need to begin with a complex system. It can start with a simple planning template that combines research findings, business conditions, and an AI-generated draft.

The economic value should be assessed using more than time savings. The calculation should also include data preparation, tool integration, review, maintenance, and error-recovery costs.

The purpose of automation is not to produce more plans. It is to execute evidence-based plans more quickly and consistently.

MTC Rating

Criterion	Rating	Reason
Impact	4/5	Connecting research to action can reduce unnecessary campaigns and wasted resources while improving consistency in marketing execution.
Urgency	4/5	Businesses using or preparing to use AI for planning should define their goals, budgets, and approval conditions before execution begins.
Business Fit	4/5	The approach applies to many small businesses that conduct research and develop marketing plans repeatedly.
Cost to Respond	2/5	A business can begin with existing research and a simple planning template, although data preparation and human review still require time.
Evidence Confidence	Medium	Evidence supports the need for core planning elements and human oversight of AI, but the actual performance of AI-generated plans varies by business and execution environment.

MTC Recommendation: ACT NOW

The immediate priority is not to allow AI-generated plans to be executed automatically.

The priority is to build a structure that converts research into a plan containing goals, priorities, budgets, execution tasks, and measurement criteria.

Begin with one recent set of research findings. Separate verified facts from assumptions, then connect one customer problem to a business goal approved by the responsible person.

Provide AI with the budget, staffing, operational capacity, and prohibited actions, and ask it to prepare a plan. Instead of automating the entire marketing operation at once, begin with one customer problem and one campaign.

Operational Conditions: The responsible person must provide and approve the goal, positioning, budget limit, and operational capacity. Every assumption must remain traceable to supporting evidence. Important budget, channel, and brand decisions must not be approved automatically.

Final Takeaway

AI can help automate the work of converting research into a marketing plan with goals, execution tasks, and performance metrics.

However, AI should create and compare options, while the business decides the goals, budget, positioning, and final priorities. Automation becomes useful in actual business operations when verified evidence is separated from assumptions and the plan can be revised in response to results.

---

References

U.S. Small Business Administration — Marketing and Sales
https://www.sba.gov/counseling/manage-your-business/#marketing-and-sales

UK Department for Business and Trade — How to Create a Digital Marketing Strategy
https://www.business.gov.uk/export-from-uk/learn/categories/prepare-sell-new-country/digital-marketing/how-to-create-a-digital-marketing-strategy/

NIST — Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile
https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.600-1.pdf

NIST — Artificial Intelligence Risk Management Framework
https://nvlpubs.nist.gov/nistpubs/ai/nist.ai.100-1.pdf

SPIN-Bench — How Well Do LLMs Plan Strategically and Reason Socially?
https://arxiv.org/html/2503.12349v3