We use AI every day. Generative, predictive, analytical. And we see our clients using it too: some with strategic intelligence, others with naivete that costs them dearly.
We're going to be straight: we sell marketing strategy and execution. Not "prompt magic." So if we recommend something about AI, it's because it's honest, not because it serves us.
What we use internally and what it taught us
Claude, ChatGPT, Gemini, Perplexity: yes. Midjourney and DALL-E: yes. Google Analytics with AI attribution models: yes.
We use AI for:
- Accelerating content production: briefs, drafts, first versions
- Analyzing data: patterns in campaigns, optimization recommendations
- Generating ideas: creative angles, messaging variations, testable hypotheses
- Automating repetitive work: segmentation, tagging, data reformatting
What it taught us: AI is a productivity multiplier for people who already know what they're doing. For those who don't, it's an error multiplier.
What we see working with clients
Good practice 1: Use AI to amplify, not replace, strategy
Successful clients define strategy first, in human. Then they ask AI: "Here's our direction. Give me 10 content variations that respect it." Not: "Tell me what strategy to have."
AI excels at transforming clear direction into hundreds of executions. It struggles at inventing direction.
Good practice 2: Put a human critic between AI and the client
Best clients don't take AI output and launch it directly. Someone reads, critiques, adjusts. Removes hallucinations. Strengthens what works. Rejects what's generic.
This is where real value gets created.
Good practice 3: Integrate AI into your process, not outside it
Instead of "an AI agency will make content for us," it's: "We have our internal process. AI accelerates step 2. We add value to steps 1, 3, and 4."
AI becomes a tool in your machine, not the machine itself.
Good practice 4: Measure real AI ROI, not just speed
Many clients celebrate: "Using ChatGPT, we produce 3x more content." Great. But does it convert 3x more? Attract 3x more qualified customers? Or are we just making 3x more mediocre content?
Good clients measure: does this change our actual metrics? Not just our production capacity.
Good practice 5: Stay intentional about brand identity
AI can write. But it doesn't know your brand like you do. Successful clients keep a human in charge of: "Does this sound like us? Does it respect our tone, values, promise to customers?"
Otherwise you get competent, interchangeable content. Exactly like your five competitors using ChatGPT too.
What we see failing with clients (and too often in AI agencies)
Trap 1: Believing AI replaces strategy
"Use Claude to write your marketing strategy." Wrong. AI gives you competent, generic strategy. Structured. Readable. But without risk-taking, testable hypotheses, or distinctive positioning.
Working strategy comes from you. AI can help structure it. Not invent it.
Trap 2: Confusing volume with results
"We produce 100 articles monthly with AI." And how many customers does that generate? Zero? Then it's noise with keywords.
Trapped clients think more output equals better ranking. Wrong. Producing thoughtfully against clear strategy ranks well. Volume without intent is evolved spam.
Trap 3: Believing AI eliminates the need for marketing expertise
AI is a tool. Not a person. It doesn't ask hard questions. It doesn't form testable hypotheses. It doesn't say "no, this won't work."
A client thinking "now we have AI, we don't need marketers" will learn expensive lessons.
Trap 4: Letting AI decide direction based on probabilities
AI often proposes what statistically worked before. Rarely what will stand out in your market or differentiate you.
"Every blog post in our industry should start with a shocking statistic." Works. Also works for your 15 competitors using the same AI. Result: noise, not signal.
Trap 5: Delegating quality validation to non-experts
We see this often: "A junior will check that AI didn't hallucinate." But the junior lacks strategic context. Can't judge if it serves the brand. Can only check if it's readable.
Validation should come from someone understanding your market, audience, positioning.
So, how to recommend AI correctly to your clients
Step 1: Define first, generate after
Start with a strategy day WITHOUT AI. Define: who do we talk to? What problem do we solve? How are we different? Which content/messages address these directly?
Then: use AI to amplify that.
Step 2: Integrate AI into your process, not as a separate process
Instead of "do something with AI and launch it," it's: "AI does step X, then we do Y, then AI does Z."
AI becomes an internal tool, not an external resource.
Step 3: Establish critical review
Someone reads EVERYTHING AI generates before it reaches the client. Not because it's broken, but because it's strategically aligned. Adapted. Differentiated. Intentional.
This is where real work begins.
Step 4: Measure what matters
Not: "Let's make 3x more with AI." Rather: "Are our conversion, engagement, and market share actually moving?" What really changes?
If AI produces more but not better, it's false economy.
Step 5: Keep humans responsible for brand identity
AI helps. But someone needs to say "Does this sound like us?" And answer honestly.
Non-delegable.
What this means for partnership with an agency
If an agency says "we use AI to work faster and better," that's standard in 2026. Everyone does it.
Real question: does the agency stay strategically involved? Does it critique outputs? Adapt them to your brand? Measure real impact?
Or does it launch AI output directly to you saying "here, done fast"?
Difference: one creates value. Other creates volume.
What we honestly recommend
Yes, use AI. Accelerate production. Test faster. Generate more variations.
But:
- Keep strategy human. It must come from you, not AI
- Put an expert between AI and your clients. Critique is where value lives
- Measure ROI, not productivity. "More" isn't better. "Better" is better
- Preserve identity. AI helps execute it, not define it
- Test continuously. AI accelerates testing, but YOU decide what to test
The honest truth
We could sell you "let us do your marketing with AI and it's cheaper and faster." Technically true. Also why 90% of clients doing this end up with mediocrity.
So we do the opposite. We tell you the truth: AI is a tool. Good tool. But tools don't replace strategy, expertise, and judgment.
If you work with an agency, ask them: Will you stay strategically involved? Will you put an expert between AI and me? Will you measure my actual ROI?
If the answer is no, it'll be cheaper elsewhere. Probably not better though.
