Website & SEO

Your Competitors Have an AI Chief of SEO. You Have 20 Prompts.

A viral thread this week laid out a 20-prompt Claude system for local SEO. It's genuinely good. But a prompt library isn't a system — here's the difference.

MurphMarch 26, 20267 min read

A post went viral this week. An SEO consultant laid out 20 Claude prompts that audit your Google Business Profile, tear apart competitor reviews, build city-specific service pages, and map every keyword gap your site has.

It's genuinely useful. I read the whole thing.

But there's a line in it that stuck with me:

"90% of people reading this will save it and never run a single prompt."

He's right. And it's not because those people are lazy.

It's because a list of prompts is not a system. It's a checklist. And checklists require you to show up every time.


What a System Actually Does

A prompt tells Claude what to do once.

A system tells Claude who you are, runs the work automatically, logs the results, and surfaces what changed — without you having to remember to check.

The difference looks like this:

Prompt: "Audit my competitor's GBP categories and tell me what I'm missing."

System: Every Monday, Claude opens your competitor's GBP, compares their categories against yours, flags anything new, and drops the diff into your dashboard.

You went from "I should do that audit" to "the audit already happened."

That's what we build at VibeTokens. Not prompt libraries — operating infrastructure.


The 20-Prompt Framework Is Actually Good

Let me be honest about this: the article's framework maps the right territory.

It covers:

  • GBP optimization — categories, attributes, photos, posts, descriptions, services
  • Website — keyword gaps, city pages, GSC analysis, review sentiment
  • Backlinks and authority — competitor link profiles, citation consistency, local intent mapping
  • Content and tracking — content gap analysis, entity optimization, monthly reporting

That's the right work. If you ran all 20 prompts consistently, you'd outrank most of your competitors within 90 days. The framework is sound.

The problem is "ran all 20 prompts consistently." That phrase is doing a lot of heavy lifting.


The Part the Article Misses

There's a layer the article doesn't cover at all, and it matters more in 2026 than it did in 2024.

GEO — Generative Engine Optimization.

When someone asks ChatGPT "best HVAC company in [city]" or Perplexity "who's the best plumber in [neighborhood]" — what shows up?

It's not the same businesses that rank in Google Maps. It's the businesses with structured, LLM-readable content. Defined entities. Consistent NAP data across authoritative sources. FAQ content that directly answers questions the way an AI search engine expects.

The 20-prompt system optimizes your GBP for Google Maps. That still matters — it matters a lot.

But the businesses who win the next three years are doing both: map pack and AI citation optimization.


What "AI Chief of SEO" Actually Means

Here's the mental model we use internally:

Your business context loads once. Claude knows your name, your services, your cities, your competitors, your review velocity, your current rankings. It knows this the way a senior employee knows it — not because you reminded it, but because it's already in the system.

From there:

Weekly: GBP competitor scan. Review velocity vs. competitors. Any new categories added. Any new GBP posts from top competitors.

Monthly: Full keyword gap report. Page 2 keywords ready to optimize. Review response performance. GEO audit — are you showing up in AI search for your target queries?

Quarterly: New city page copy. Updated GBP description tests. Entity audit — does Google's knowledge graph know who you are?

None of this requires you to remember to run a prompt. It happens, it lands in a report, you review it in 10 minutes, you approve what to push.

That's the difference between a prompt library and a system.


The Business Case

Home service businesses — HVAC, plumbing, cleaning, landscaping, electrical — are the clearest example of what's at stake here.

These businesses live and die by the map pack. One position in Google Maps = a meaningful difference in call volume per month. For a plumber doing $150–$400 per call, that's not an abstraction.

Most of them have:

  • A GBP they set up 4 years ago and haven't touched
  • 60–200 reviews they've never systematically analyzed
  • Zero secondary GBP categories (meaning they're invisible for half the searches they should rank for)
  • A website with one generic "Services" page instead of city-specific landing pages
  • No structured content for AI search

The gap is real. The fixes are specific. The only question is whether someone runs them manually (20 prompts, 90% never do) or builds them into a system that runs without asking.


If You Want to Try This Yourself

Start with the business context load. This is the one thing you can do right now that changes every subsequent prompt you run.

Before you open a new Claude chat for anything SEO-related, paste in this block:

My business: [Name], [Address], [Phone], [Website], [GBP URL]
Primary service: [What you do]
Service areas: [City 1], [City 2], [City 3], [City 4], [City 5]
Top 5 target keywords: [keyword 1], [keyword 2], [keyword 3], [keyword 4], [keyword 5]
Current GBP status: [X] reviews, [X] stars, [X] new/month, ranking for [X], invisible for [Y]
Top 3 competitors: [Name] — [GBP URL], [Name] — [GBP URL], [Name] — [GBP URL]

Work from this context for everything. Never ask me for this information again.

That alone will improve every audit you run by 10x.

If you want the full system — automated, running weekly, with a dashboard you check once — that's what we build.

→ See how it works


The viral article is right about the playbook. The question is whether you're going to run it manually once, or build it into a machine that runs it for you every week.

One of those compounds. One of them sits in your bookmarks.

Want this for your business?

Tell us what you're building. We'll map out exactly what to build and what it costs.

Start Your Project →

Frequently Asked

What's the difference between a Claude SEO prompt library and a real SEO system?

A prompt library requires you to remember to run each prompt, interpret the output, and take action manually — every time. A system is designed once: it tells Claude who you are, what you're tracking, and runs checks automatically, surfacing what changed without you initiating each cycle. A checklist makes you the operator; a system makes you the reviewer.

Can Claude actually manage local SEO for a business autonomously?

Not fully autonomously — Claude can't log into your Google Business Profile or submit changes directly. But it can analyze your GBP data, draft optimized responses to reviews, write city-specific service pages, identify keyword gaps, and generate a weekly action list. The work becomes structured and repeatable rather than ad hoc.

How do you turn 20 SEO prompts into a system that actually runs?

Consolidate the prompts into a single persistent workflow document. Give Claude standing context about your business, your competitors, and your SEO goals. Schedule regular sessions where Claude reviews input data and produces a prioritized output report. The difference is moving from 'I run these when I remember' to 'this runs on a schedule and surfaces what needs action.'

Jason Murphy

Written by

Murph

Jason Matthew Murphy. Twenty years building digital systems for businesses. Former CardinalCommerce (acquired by Visa). Now running VibeTokens — AI-built websites and content for small businesses.

The window is open.

It won't be forever.

Start Your Project →