Here is a simple experiment any business owner can run in ten minutes.
Open Claude, ChatGPT, Gemini and Grok. Type exactly the same question into each one:
“I run a restaurant in Ho Chi Minh City with a monthly marketing budget of 30 million VND. Where should I spend it?”
Read the four answers side by side. Then ask yourself one question: which one is right?
What you will notice
You will most likely see overlap. Every answer will mention familiar ideas: Google Maps, social media, reviews. But the priorities will rarely match. One may lean towards paid ads, another towards content, another towards loyalty programmes. The suggested budget splits will be different.
Why the answers are different
This is not a bug. Each AI is built by a different company, trained on different data, and tuned with different goals.
- Claude (Anthropic) tends to give careful, structured answers and often adds caveats.
- ChatGPT (OpenAI) is broad and practical, and often gives the most list-style plan.
- Gemini (Google) is connected to Google’s ecosystem and leans on what Google’s tools can do.
- Grok (xAI) is connected to X and tends to be more direct and opinionated.
None of them know your restaurant. They are each making a reasonable guess about an average restaurant.
The uncomfortable conclusion
If four expert-sounding sources give four different plans, the tool is not your answer. It is your input.
Someone still has to decide. That person needs to know things none of the four AIs knew:
- Your average bill size and your margin per table.
- Whether your customers find you by walking past, by searching, or by friends’ recommendations.
- Which nights are empty and which are fully booked.
- What your competitor two streets away just started doing.
With that information, choosing between the four answers becomes easy. Without it, you are choosing the one that sounds most confident.
Confidence is not correctness
AI answers all sound equally sure of themselves. That’s the danger. A business owner reading one answer feels informed. A business owner reading four feels confused, which is actually the more accurate feeling.
The confusion is useful. It shows you exactly where the real work is: in your own data and in judgment.
Where DMG fits
At DMG we use AI exactly this way. We let it generate options quickly, then test them against the client’s real numbers, customers and market. AI can give you four plans. Your data tells you which one to run.
Ran the experiment and got four different answers? Bring them to us, and we’ll help you pick the one that fits your business. Talk to DMG.
