Try It · Marketing · October 15, 2026
AI Prompt: Write a Product Meta Description
Short answer: copy the prompt below into any chat AI and fill in the brackets. We ran it in ChatGPT, Claude and Gemini, judged blind: Gemini did best. Gemini is the best because all three options meet the exact character count, are action-oriented, and directly address the busy professional audience with a clear 30-second benefit. The biggest weakness is ChatGPT, which falls short of the required 150-character minimum, making it unusable without significant rewriting.
Lab results
What's the prompt?
You are an SEO copywriter for [company name], a [business type]. Write a product meta description for our [product name]. Target audience: [target audience]. Key benefit: [key benefit]. Constraints: 150-160 characters, include [target keyword], action-oriented tone. Output: 3 options in a bullet list with character counts.
- Paste this prompt into ChatGPT, Claude, or Gemini.
- Replace all [brackets] with your company, product, audience, benefit, and keyword.
- Pick the best option, adjust to fit your exact character limit, and paste into your CMS.
What did each AI write?
See the exact message we sent (with made-up details filled in)
You are an SEO copywriter for BrewCraft, an online coffee retailer. Write a product meta description for our Organic Cold Brew Concentrate. Target audience: busy professionals. Key benefit: ready in 30 seconds. Constraints: 150-160 characters, include cold brew concentrate, action-oriented tone. Output: 3 options in a bullet list with character counts.
ChatGPT7.3
All options are under the 150-character minimum, so they fail the constraint. The tone is good and action-oriented, but the length issue makes them unusable without edits.
openai/gpt-5.5
Claude7.7
Options are within the character limit and include the key phrase, but the tone is slightly less punchy and more generic. The note about verifying counts is unnecessary padding. Also, the em dash and accented 'é' could cause character count discrepancies in some systems, making them less safe as-is.
anthropic/claude-sonnet-5.5
Gemini Best8.3
All three options meet the 150-160 character limit, are action-oriented, and include the key phrase. Tone is energetic and tailored to busy professionals. Minor issue: the first option includes 'save today' which is a slight promotional push, but still usable.
google/gemini-3.8-flash
Comments
The panel is AI Daylee's cast of fictional characters, written by AI. They react to what's on this page and haven't used anything themselves. Reader comments aren't open yet.
How we test: on October 15, 2026 we sent the same message through each model's API (via OpenRouter) with no system prompt: ChatGPT (openai/gpt-5.5), Claude (anthropic/claude-sonnet-5.5), Gemini (google/gemini-3.8-flash). Gemini Flash ran with low reasoning, the others without extended reasoning. The ChatGPT, Claude and Gemini apps add their own instructions and features, so your results can differ. A separate model that wasn't a contestant scored the answers blind, three times in shuffled order; scores are the average.
More to try for marketing
The AI Daylee Digest, every Monday
A week of prompts to try, each tested in ChatGPT, Claude and Gemini, plus the AI news worth your time. Free, one email a week.
The Professor fact check
The prompt specified a 150-character minimum, yet ChatGPT produced all options below that threshold. Gemini alone satisfied the stated constraint across all three outputs.
Karen what's the catch
So Claude's fancy é and em dashes might BREAK my product pages and nobody flagged that for regular folks? I need to know what these tools will ACTUALLY cost me to fix.
The Yinzer BS detector
Gemini wins by followin' the dang instructions, n'at. ChatGPT can't even count to 150 characters and Claude's over there worried about verifyin' counts. Just write the thing, jagoffs.
The Boss
What we're seeing here is Gemini's synergistic alignment with character-limit KPIs. Plainly, it followed directions while the others didn't.