Stop letting AI make your marketing decisions
AI can support marketing implementation, but experienced marketers must guide strategy, evaluate work and stay accountable for results.
Published on 2026-10-07
Share this article
Table of Contents
Before you hand a marketing decision to AI, ask yourself whether you could evaluate that decision without it.
I have spent more than 28 years in digital marketing. My career started in offline marketing, followed by sales, marketing management, a director role and agency work. Since founding Brick Marketing in 2005, we have worked with and helped more than 600 B2B and B2C companies, from small local businesses and midsize organizations to enterprises.
That experience shapes how I use AI. I see real opportunities to help agency and internal teams improve results, provided they understand the work. My concern is what happens when someone accepts a recommendation without knowing enough to question it.
An answer can sound reasonable and still overlook something as basic as how long customers take to buy. Someone has to recognize what is missing before that advice becomes a marketing plan.
Lessons to share with those responsible for marketing
Know the work before using the AI tool
Owning a dental chair and a dentist’s instruments does not qualify you to place a crown. Access to AI does not qualify someone to run marketing, either.
Marketers should know how to do the work without it. You need that foundation to catch mistakes, recognize weak recommendations, and decide whether the output is useful. Otherwise, how are you judging it?
Consider a report that recommends increasing website traffic. Before accepting its recommendation, I want to understand who already visits the website and what happens after they arrive. More visitors might help, but if the offer is unclear or the website attracts people who will never become customers, increasing traffic could leave the business with the same problem.
The marketer needs to make that distinction. Knowing how to ask AI a question is helpful, but it does not replace understanding customer behavior, reviewing performance, or writing a message that gives someone a reason to respond.
For someone learning marketing, I would make the work itself part of the training. Have them review a campaign and explain what they would change before asking AI for suggestions. Then compare the reasoning.
Keep your strategy with experienced marketers
I want AI to help with implementation. I do not want it to decide positioning, priorities, or investments. Those decisions require knowing the business, its customers and its sales process.
Budget constraints and sales cycles matter here, too. A recommendation might be sensible for a company with an established marketing department and completely unrealistic for a business where one person handles everything. Even a good idea has to fit the resources available to carry it out.
The same applies to positioning. Before changing how a business describes itself, someone needs to understand why its existing customers chose it in the first place. A new description might sound stronger while removing the detail that made the business a good fit.
Suppose a company sells a service that takes six months to purchase. Evaluating a campaign after several weeks requires some patience and a closer look at the conversations it has started. An immediate sale may not be a reasonable expectation. That does not mean the company should keep spending indefinitely. It means the evaluation should reflect how that company wins business.
An experienced marketer should decide what information belongs in that discussion and what evidence would justify a change. AI can help organize the information or suggest questions worth exploring. I would still keep the decision with someone who understands the consequences.
Executives without marketing experience should have a marketer evaluate AI advice before acting on it. I recommend a discussion with the people responsible for delivering it, especially when it changes their priorities or commits their budget.
Choose the assignments carefully
I have found AI useful for ad variations and tagline ideas. I do not consider it a substitute for substantive copywriting. Those assignments require different levels of judgment.
With an ad, I can work from a defined offer, a specific audience and a clear action I want someone to take. I can review the suggestions against those requirements. An option that sounds good but promises something the business cannot deliver gets rejected.
An article needs our experience, original thinking and voice. I still need to stand behind every word. The question is whether the article says something useful we would actually say to a client. A polished paragraph that could appear on any agency’s website does not give a reader much reason to trust ours.
Before assigning work to AI, decide what a finished piece needs to accomplish and who can review it properly. An editor can improve readability, but someone familiar with the subject needs to evaluate the substance.
I would also separate generating ideas from approving them. There is nothing wrong with considering a suggestion and deciding it does not fit. You do not owe the tool a place for every answer it produces. Match the assignment to your ability to evaluate the output, and leave room to do the work yourself when that produces a better result.
Ask AI to give proof for its search marketing advice
Search marketing recommendations deserve scrutiny because acting on them can create work across an entire website. Before making changes, ask what business problem the recommendation solves and why it applies to your situation.
If AI suggests publishing more pages, get in the habit of questioning it. Which audience would those pages serve? What information is missing from the website? Will adding content help someone make a decision, or is the recommendation simply asking the team to produce more?
That advice may not address an existing conversion problem. A business could already have relevant visitors who cannot find a clear explanation of its services or an obvious way to schedule a conversation. If that is the case, I would investigate what those visitors need before committing to a larger publishing schedule.
Ask AI to cite where its advice came from, then read the sources. Check whether they support the actual recommendation and whether the circumstances resemble yours. A link alone is not enough. Neither is a confident explanation that leaves you unable to explain why the change should help.
Next, test the recommendation before making changes across the whole website. Decide what improvement you expect, how you will measure it, and when you will review the results. Ask whether the work is likely to contribute to growth six months from now.
Keep a written record of what you changed and why. If you revise the content, navigation, and contact process at once, it becomes harder to tell which change helped. A smaller test gives the team something specific to learn before adding even more work to the schedule.
Stay accountable for the results
An experienced marketer should approve the work and own the outcome. That responsibility does not shift to AI because it helped produce an article, an advertisement, or a recommendation.
Check accuracy, relevance, and brand voice before publishing. Make sure the business can support any claims being made. Junior marketers should explain their decisions, including why they accepted one suggestion and rejected another. That conversation helps a manager see whether the person understands the work or is simply passing along an answer.
Measure qualified leads, sales opportunities, and revenue against the goal. If the purpose of a campaign is to generate conversations with potential customers, publishing more content is not enough to call it successful.
Include review and correction time when calculating efficiency gains. Producing a draft faster means little if an experienced team member has to spend longer fixing it. Watch for declining quality as production increases, and give the people reviewing the work enough time to do it properly.
I have learned that if output increases while results plateau, reconsider the approach before asking AI to produce more. The next step may be to revise the offer, improve existing content, or speak with the sales team about what prospects are asking.
I believe AI can help us improve results and scale implementation. The useful question is where it supports work your team understands and can evaluate. Invest in your team’s ability to question recommendations and recognize weak work. The responsibility for knowing whether that marketing works stays with us.