← All issues

The AI told you to lead with the story. Who told it no?

· The Zolt newsletter

Maya's campaign had no one to argue with it

We run into this with customers all the time, so let's call this one Maya. She runs a small home-goods shop, mostly a handful of products she's proud of and a loyal-ish email list. Next month's campaign needed an angle, and she asked an AI tool a simple question: lead with a discount, or lead with a customer story?

The answer came back in confident, tidy paragraphs. Lead with the story. It gave reasons. They sounded right. Maya liked the story angle anyway, so she felt relieved, and she built the whole campaign around it: the emails, the posts, the hero image on the homepage.

A shop owner at a desk in the back room of her home-goods store, looking at her laptop with a notebook beside her.

Two weeks later, nothing had moved. Opens were fine. Clicks were thin. Sales looked like any other fortnight.

When she finally sat down and looked at how her customers actually shop, the gap was obvious. Her buyers compare prices before they read anything else. The story was nice. It just wasn't the thing standing between them and the checkout button. And nobody, including the tool, had said so.

This is the trap in one AI answer being one opinion. The tool picked one reading of her question and committed to it, in the tone of a verdict. Maya wasn't careless. She just had one voice in the room, and it agreed with her.

A single answer sounds finished. It's really the first sentence of a discussion.


You need friction, not a better answer

A content decision is rarely right or wrong in the abstract. It's right for a particular goal, a particular audience and a particular moment. The weak points only show up when someone looks at it from a different seat.

A business with a proper marketing team gets that for free. The strategist asks what this is for. The creative asks if anyone will remember it. The critic asks what's wrong with it. The person who knows the customer asks whether your buyer would care. Someone asks what everyone is assuming without having checked.

Most owners, solo marketers and small agencies don't have those five people. You have you, and you wrote the question, which makes you the worst-placed person to spot your own assumptions. They feel like facts.

Typing "now argue against it" helps a little, but the objection that comes back is often a polite, token one that the original answer easily absorbs. And on a busy week, when a rushed decision does the most damage, you're least likely to bother.

The fix is to ask five separate questions about the same decision:

  1. Strategy: What is this meant to achieve, and does this approach get there?
  2. Creative: What's the most memorable way to say this?
  3. Critique: What's wrong with this plan? Where will it fall flat?
  4. Audience fit: Would my actual buyer stop, believe it and act?
  5. Bias check: What am I assuming that I haven't checked?

The fifth one is the one people skip, and it's usually the one that pays. "My buyers want a story" isn't a strategy. It's a guess you haven't tested yet. It's exactly what would have saved Maya two weeks.

Don't ask for a better answer. Ask for the other seats at the table.


Try this week

Pick one content decision you've already made this week: an angle, an offer, a headline. Set a timer for ten minutes and write the strongest case against it in three sentences, as a customer who isn't convinced. Then, in one more line, write down the one assumption the plan depends on and how you could check it (a look at your last few orders, a question to two customers, a scan of your own replies).

If the plan survives, publish with more confidence. If it doesn't, you found out before your audience did.

Or skip the routine: Boardroom Mode in Zolt answers one question from five angles at once (strategy, creative, critique, audience fit and a bias check), so the disagreement is on the page before you commit.


From the Zolt blog


No one needs a bigger team to get a second opinion. They just need to remember to ask for it before the campaign goes out, not after. If you'd like help making that a habit, you can Get Started whenever it suits you.

— Your friends at Zolt

P.S. Hit reply and tell us the last content decision you wish someone had pushed back on. We read every one.

Want the full version? One AI answer is one opinion, and nobody argues back on the Zolt blog.