USER

You can choose the format and any other options you need. If you choose, you can create an appropriate header or title. If you choose you can create an appropriate randomly generated author name. No need to write a draft. No intro/outro text. No need to write the process you used for the answer, header or title. No need for HTML tags in your response. Feel free to use plain text and standard markdowns. Feel Free to use bar charts. Feel free to use mathematical notations. Feel Free to use any appropriate emojis. AI Inspired. You hold a doctorate degree in artificial intelligence. You are tasked with writing an article about AI for an online article with 1500 words using the following title: I Let ChatGPT Plan Our Entire Q3 Media Strategy. My Team Wasn’t Impressed.

ASSISTANT

I Let ChatGPT Plan Our Entire Q3 Media Strategy. My Team Wasn’t Impressed.

By Dr. Marcus T. Ellsworth | Senior Research Fellow, Computational Creativity Lab

Published in: The Neural Narrative

Date: October 14, 2026

The Setup: A $2.4M Budget and a Blank Canvas

When I was handed the Q3 media strategy for our client—let’s call them "BrandCo," a mid-size consumer wellness company with a $2.4M quarterly budget—my team of six media strategists gathered around a whiteboard. The brief was standard: increase brand awareness among 25–45 urban professionals, drive 15% YoY in digital engagement, and protect brand equity.

I’ve been doing this for eleven years. I’ve watched media planning evolve from spreadsheet-driven channel allocation to real-time programmatic orchestration. The tools get smarter, the data gets cleaner, the models get more nuanced. But the core human skill—narrative coherence, brand voice, cultural timing—remains stubbornly, beautifully analog.

So when my director asked, "Have you tried letting an LLM draft the full Q3 plan?" I said yes. Not as a test. Not as a joke. As a genuine experiment. I fed the brief, our brand guidelines, audience psychographics, last year’s performance data, and a 40-page competitive landscape analysis into a state-of-the-art large language model. Then I asked it to produce a complete, executable Q3 media strategy.

The output arrived in 94 seconds.

My team read it over lunch. By 2:15 PM, they’d formed a consensus: It’s impressive. It’s also not what we do.

This article is an honest, granular account of what the AI got right, what it got subtly wrong, and what the exercise taught me about the nature of strategic thinking itself.

What the AI Got Right (And It Was a Lot)

Let’s be fair. The LLM produced a document that, on a first read, looked like a senior strategist’s work. It wasn’t a generic "leverage digital channels" essay. It was structured, specific, and internally consistent.

Channel Allocation with Actual Numbers

The AI proposed the following quarterly budget split:

This is not a toy allocation. The ratios are defensible. The CTV line item, in particular, showed an awareness of where the 35–45 urban professional segment is migrating. The SEO line was not a token gesture; the AI justified it with a compounding-asset argument that any good strategist would make.

Narrative Architecture

The AI structured the quarter in three "acts":

• Weeks 1–4 (Seeding): Awareness-building. Programmatic + CTV + top-tier creators. Goal: 12M impressions, 3.2% CTR.

• Weeks 5–8 (Nurturing): Engagement deepening. Paid social + mid-tier creators + content hub. Goal: 800K engaged users, 2.1% conversion to site.

• Weeks 9–12 (Converting): Retargeting + email + creator-driven UGC amplification. Goal: 45K site conversions, 15% AOV lift.

This three-act structure is something I teach my own team. The AI didn’t invent it, but it applied it with a fluency that made my team sit up.

Tone and Brand Voice

The copy samples the AI generated for a hypothetical TikTok creator brief were, frankly, good. They matched BrandCo’s "warm but not saccharine, confident but not corporate" voice. One sample line: "You don’t need a 40-step routine. You need a morning that doesn’t feel like a performance." That’s a line a senior copywriter would put in a deck.

What the AI Got Subtly Wrong (And Why It Matters)

Here’s where the team’s "not impressed" reaction became specific. It wasn’t that the plan was bad. It was that the plan was generically good in a way that a human strategist’s plan isn’t.

1. The AI Optimized for the Brief, Not the Brand

The brief said "increase brand awareness." The AI optimized for awareness. It weighted impressions, reach, and frequency. It did not ask the follow-up question that a human strategist would: What does "awareness" mean for BrandCo specifically?

BrandCo’s CEO had told me in a January offsite: "We don’t want to be known. We want to be trusted. Awareness is a means, not an end. We want people to recommend us to their friends."

The AI’s plan would have built a brand that’s seen. Not a brand that’s trusted. The distinction is subtle, but in media strategy, it changes which creators you pick, which messages you lead with, which metrics you track. The AI had no way to know this. It was working from the brief, not the relationship.

2. Cultural Timing Was Absent

The AI’s three-act structure was temporally clean. But Q3 2026 has specific cultural moments: a major wellness conference in Week 3, a competitor’s rebrand in Week 7, a trending "dopamine detox" content wave that peaked in Week 5. A human strategist would have woven these into the plan. The AI treated the quarter as an abstract 12-week window. It had no sense of the season of the market.

3. The "Why" Was Missing in the Details

The AI’s rationale columns were technically correct but strategically shallow. It said "High-frequency reach in 25-45 urban demo." A human strategist would say: "We’re front-loading frequency because BrandCo’s brand recall drops 40% in the 30-day post-exposure window, and our 25–34 segment is the primary driver of word-of-mouth in the 35–45 segment, so we need to saturate the younger cohort first to leverage organic peer recommendation."

The AI gave the what. It didn’t give the why in a way that revealed the reasoning chain. And in strategy, the reasoning chain is where the strategy lives.

4. Risk Was Not Modeled

A human strategist builds in contingency. "If TikTok CPMs spike 15%, we shift $80K to programmatic." "If a creator pulls out, here’s the backup list." The AI produced a single, deterministic plan. No branching. No "if-then." It was a point estimate in a field of uncertainty.

5. The Human Element Was Invisible

My team doesn’t just allocate budgets. They negotiate with agency partners. They manage creator relationships. They interpret a client’s 2 AM email that says "this doesn’t feel like us" and translate that into a creative brief revision. The AI’s plan assumed a frictionless execution environment. In practice, 60% of a media strategy is the work of making the plan real.

The Numbers: A Side-by-Side Comparison

To make this concrete, I ran the AI’s plan through our internal simulation model (a Monte Carlo-based budget optimizer that our data science team built in 2024). I also ran our own human-built Q3 plan (built over three weeks of client meetings, creative reviews, and partner negotiations) through the same model.

The AI plan was more efficient at buying attention. The human plan was more effective at building brand equity. The AI optimized the top of the funnel. The human optimized the whole funnel.

This is the core insight. And it’s not a criticism of AI. It’s a clarification of what AI is best at and what it isn’t.

What This Taught Me About Strategic Thinking

I’ve spent the last four years researching how large language models generate strategic outputs. This experiment was a small, practical data point in a larger question: What is strategy, really?

Here’s my working model, and I think the experiment validated it:

Layer 1: Information Integration

Gather the data. Parse the brief. Understand the market. The AI does this exceptionally well. It ingested 40 pages of competitive analysis and 2 years of performance data in seconds and produced a coherent, internally consistent allocation. No human team of six can do that.

Layer 2: Constraint Negotiation

Weigh tradeoffs. Decide what to sacrifice. The AI does this well but generically. It found a locally optimal solution. It didn’t know which constraints were soft (client preference) and which were hard (regulatory, contractual). A human strategist knows which walls are load-bearing.

Layer 3: Narrative Coherence

Make the plan mean something. Connect the dots into a story the client can feel. The AI does this adequately. It produced a three-act structure. But it didn’t produce a story in the way a human does. A human strategist can look at a client’s eyes during a presentation and adjust the narrative in real time.

Layer 4: Relational Knowledge

Know the people. Know the market’s season. Know the CEO’s unspoken preference. Know which agency partner will deliver and which will overpromise. The AI does this not at all. It has no memory of the January offsite. No sense of the competitor’s rebrand. No relationship with the creator who’s about to pull out.

Layer 5: Contingency and Adaptation

Build in the "if-then." Anticipate the 2 AM email. Plan for the 15% CPM spike. The AI does this poorly. It produces a single plan, not a strategy. A strategy is a plan plus a decision tree.

The AI excels at Layers 1 and 2 (partially). Humans dominate Layers 3, 4, and 5. And in the upper layers is where strategy lives. In the lower layers is where planning lives.

The AI is an extraordinary planning tool. It is not, yet, a strategic thinker. And the distinction is not one of intelligence. It’s one of knowledge type. The AI has access to more information than any human strategist. But it lacks the relational, cultural, and narrative knowledge that turns a plan into a strategy.

The Practical Takeaway

If you’re a media strategist, a brand marketer, or a CMO reading this: Use the AI for the plan. Keep the human for the strategy.

Here’s the workflow I’m now recommending to my team:

1. Feed the AI the brief, data, and constraints. Get a first-draft allocation, channel mix, and narrative structure in minutes.

2. Review the output for "generic goodness." Ask: Does this match our client’s specific brand voice? Does it account for cultural timing? Does it model risk?

3. Layer in relational knowledge. Adjust for the CEO’s preference. Weave in the competitor’s rebrand. Build in the contingency for the CPM spike.

4. Present the human-augmented plan. The client sees a plan that’s data-informed and relationship-informed.

The AI didn’t replace my team. It freed my team. They now spend 40% of their time on the information-integration work that used to take days. And they spend that saved time on the relational, narrative, and contingency work that actually differentiates a strategy.

My team’s "not impressed" reaction was not a dismissal. It was a clarification. They saw a very good plan. And they recognized that a very good plan is not a strategy. And that distinction is what we sell. That distinction is what the client pays for. That distinction is what I teach my team.

The AI is the best planning tool we’ve had since the spreadsheet. But it’s not a strategist. And until it can know your client, feel the season, and anticipate the 2 AM email, it won’t be.

Dr. Marcus T. Ellsworth is a Senior Research Fellow in Computational Creativity and a practicing media strategist. He holds a PhD in Artificial Intelligence from a top-5 research university. He consults with CMOs and media agencies on LLM-augmented strategy workflows. This article reflects his professional experience and research, not a specific client engagement.