ChatGPT will tell you exactly how it ranks your category. You just have to ask.

Issue #4 · April 18, 2026

When ChatGPT recommends the best automatic coffee machine, it’s not pulling from a vibe.

I asked it yesterday. Two prompts, same session.

Prompt 1: Best automatic coffee machines.

It returned five machines across five slots. Jura Z10 at $4,499 as best overall. De’Longhi Dinamica Plus at $1,499 as best value. Philips 5500 LatteGo at $898 for milk drinks. Breville Barista Touch Impress at $1,499 as the hybrid. Jura GIGA 10 at $5,499 as ultra-premium.

Four price tiers. Five jobs-to-be-done. Every slot filled.

Prompt 2: What criteria did you use to rank these? Give me the weighting as percentages.

It gave me the rubric. Cited three sources: Consumer Reports, Specialty Coffee Association, Wirecutter.

• Coffee Quality (extraction + grinder) — 35%
• Ease of Use & Automation — 20%
• Milk System Performance — 15%
• Maintenance & Reliability — 15%
• Customization & Features — 15%

This is the scorecard. If your product page, your PR coverage, and the reviews ranking for your category don’t signal these five things in this order, you’re not in the answer.

Two things most operators miss.

The rubric is category-specific. A generic electric toothbrush rubric leads with gum safety and replacement head cost. Coffee leads with extraction pressure and burr grinder consistency. ChatGPT is not applying a universal product framework. It’s synthesizing the authorities that own your category.

The rubric names its sources. Consumer Reports. Specialty Coffee Association. Wirecutter. If none of those three cover you well on those five criteria, the rubric is rigged against you by default. Not by ChatGPT. By the three publications it trusts for coffee.

Open a fresh ChatGPT session. Run these two prompts for your category:

1. Best [your category]
2. What criteria did you use to rank these? Give me the weighting as percentages.

Then ask three questions.

Does your product detail page lead with the criterion that got the highest weight, or with a feature ChatGPT ranked fifth?

Do the publications ChatGPT cited cover your product? If not, that’s your PR target list. Not tech press. The specific outlets the rubric named.

Is there a slot in the top five that nobody owns? A price tier, a use case, a job-to-be-done? That’s where a challenger wins.

Under 10 minutes. Do it before Monday.

Sources and citations →