AI can deliver a polished answer with more confidence than any human before it. Never mind that it's the wrong answer. It's easy to mistake AI confidence for accuracy. So check it. Push back. Find where it's making assumptions, overlooking constraints or pulling in bad stats.
You do not need to challenge everything. But do not be shy about questioning a result.
Eight ways to push back
Flip a card for a fast reference check.
Assumptions
Tap to flipWhat assumptions are you making here?
Generic
Tap to flipWhich parts of this could apply to almost any company?
Prioritizing
Tap to flipWhy is this the priority?
Pillars
Tap to flipWhich idea could most easily belong to a competitor?
Workload
Tap to flipIs this realistic for the team and budget I gave you?
Drift
Tap to flipDoes this conflict with anything in the project context?
Measurement
Tap to flipAre we measuring what matters, or simply what is easy to track?
Before you ship
Tap to flipWhat would have to be true for this strategy to fail?
Part I: While you're drafting
The AI seems to be making assumptions
AI fills gaps by making things up. These questions catch it in the act.
- What assumptions are you making here?
- What about this is a fact?
- Which parts are supported by evidence, and which are inference?
- What information would most change this recommendation?
- Where is the evidence weakest?
Generic recommendations
Generic advice is AI defaulting to the industry average instead of your actual situation.
- Which parts of this could apply to almost any company?
- What would make this more specific to our audience and goals?
- Which recommendation is least differentiated?
- Are you recommending this because it's standard best practice, or because it will meet our objectives?
Push for priorities
AI can generate endless options. Strategy requires choices.
- Why is this the priority?
- What evidence makes this more important than the alternatives?
- What would you deprioritize to make room for it?
- Which recommendation has the weakest connection to our stated goals?
Part II: While you're deciding
Reviewing pillar and content ideas
More ideas are not the same as better ideas. These questions separate the two.
- Which pillar or idea is least distinctive?
- Where is there too much overlap?
- Which idea could most easily belong to a competitor?
- Which ideas require the most effort relative to their likely value?
- Where could one asset support several pieces of content?
- If we cut 25% of this, what should go first?
Feasibility
AI does not get invoices for implementing its recommendations. You do.
- Is this realistic for the team and budget I gave you?
- What is the hidden production workload?
- What would you change if our resources were cut by 25%?
- What would a more sustainable version of this plan look like?
- What would you protect if we had to do less?
AI can drift
A long conversation can quietly wander from what you already approved.
- Does this conflict with anything in the project context?
- Check this against our brand charter, audience priorities, and goals.
- Have you introduced any new directions we didn't approve?
- Which earlier decision led you to this recommendation?
Part III: Before you ship
Measurement review
Easy to track and important to know are not always the same metric.
- Which metrics actually tell us whether we achieved our goals?
- Which are outcome metrics, and which are diagnostic?
- Are we measuring what matters, or simply what is easy to track?
- Do we have the baselines needed to judge success?
Final set of questions
One last pass, before a weak assumption gets built into the plan.
- Which recommendation is least defensible?
- What is the highest-risk assumption?
- What would have to be true for this strategy to fail?
- Which part is most likely to look good on paper but fail in execution?
- What still requires human judgment or validation?
Thanks for stopping by.
Want to connect?
Bring me your next business challenge, a strategy that needs focus or a project your team can’t take on. I bring 15+ years of content and creative expertise, with AI expanding what I can deliver. You get the range of a multidisciplinary team without the overhead.
Let's talk →