12 Claude Skills Every

The twelve Claude skills worth installing for a small business are not twelve prompts you retype. They are twelve commands that read your saved business context and write their results back, so a plan runs on your numbers, an email sounds like you, and a re-run reports what changed. You install them once in about five minutes, answer a setup interview once in about thirty, and from then on the work compounds instead of resetting every time you open a new chat.

The reason this matters is that small businesses are already experimenting with AI and getting very little from it. Most of that use is a smarter search box: a clever answer that lands in a chat window and is gone the moment the tab closes. A skill is different in one specific way. It does not start from a blank slate. It reads a small folder of notes about your business first, then does its job, then saves the result with a date. That single property is what turns a chatbot into something that operates on your business. For the system view of how this fits together, see Claude as a business operating system.

• Small businesses are already experimenting with AI, but much of it is shallow and one-off.
• Twelve skills cover the recurring revenue work, grouped by job: diagnose the gap, fill the pipeline, convert, prove it, and run the day.
• Each one reads your saved business context, so the output fits your business instead of a generic template.
• Install takes about five minutes; the one-time setup interview that builds your context runs about thirty.
• All free, plain markdown, one command each, no coding required.

What Is A Claude Skill, And How Is It Not Just A Prompt?

A skill is a short instruction file that teaches Claude one job and triggers from a single command, like /free-plan or /daily. That part sounds like a saved prompt, and if it stopped there it would be one. The difference is what the skill reads before it acts.

A saved prompt carries only the words you typed. A skill reads your business context first: a small folder of plain text notes that describe what you sell, who you serve, how you write, and the numbers you want to move. So when the plan skill runs, it already knows your offer and your target customer. When the email skill runs, it already knows your voice. And when any skill finishes, it writes its result back to that folder with a date, so the next run has something to compare against.

That is the whole reason to install rather than paste. A prompt gives you a fresh answer every time and forgets you between tasks. A skill gives you an answer grounded in your business and remembers what it found. The twelve below are grouped by the job they do, because nobody installs a tool, they install a way to get a specific piece of work done.

Which Skills Diagnose The Gap?

Before you fix anything, you need to know where the money is getting stuck. These four read a page or an export and grade it, then save the score so next month's run can show movement instead of starting over.

1. /free-plan: paste a URL, get a graded cross-channel revenue snapshot with the quick wins ranked. It runs with no setup, so it is the right first touch.
2. /seo-audit: a graded read across sixteen categories that saves a health-ledger line, so the score becomes a trend you can watch.
3. /ads-audit: a budget-weighted read of your Google and Meta exports that grades each platform and hands back an action plan, so wasted spend is named, not guessed at.
4. /seo-optimize: feed it a Search Console export and it returns a ranked P0 to P3 roadmap of the clicks you are leaving on the table.

The reason these four belong together is that they all produce a graded number, and a number you save is a number you can move. The free plan is the wide-angle first look across every channel. The SEO and ads plans go deep on the two places small businesses most often bleed money: a site search engines cannot read, and paid budget spent on clicks that never convert. The optimize skill is the follow-through, turning a Search Console export into a ranked list of pages that are one fix away from more traffic. Run them once for the picture, then re-run monthly so the grade tells you whether last month's work landed.

Which Skills Fill The Pipeline?

A diagnosis is useless if nothing new is coming in. These two build and clean the top of the funnel against your actual customer profile, not a generic template of who might buy.

5. /lead-research: builds a sourced prospect list from your customer profile and scores each one Hot, Warm, or Cold, so you work the list in priority order.
6. /lead-qualify: takes a raw list you already have and gates it against your customer profile, with a one-line reason for every verdict, so a "no" is a decision and not a guess.

Which Skills Convert The Pipeline Into Revenue?

Leads do nothing until someone follows up well. These two write the outreach in your voice, read from your context, so the email does not read like it was generated.

7. /email-sequence: designs a full follow-up, welcome, or sales flow, every email written in your voice from your brand notes.
8. /email-personalize: writes one real first line per lead for a cold list, built from actual research on that prospect, not a swapped-in first name.

Which Skills Prove It Worked?

The work that closes the next deal is the proof you banked from the last one. These two turn your own results into reusable assets, traced to your numbers so nothing is invented.

9. /win-loss: reads your closed deals and writes a winning-customer profile plus a red-flag kill list, so your targeting sharpens with every deal you close or lose.
10. /case-study: turns a real result from your goal ledger into a written case study, traced to the source line and flagged for consent before anything ships.

These two close a loop most small businesses leave open. Every deal you win or lose is data, and without a habit of capturing it, the lesson walks out the door with the prospect. The win-loss skill reads the pattern across your closed deals and feeds it straight back into your customer profile, so the lead research that comes after targets the customer who actually buys, not the one you assumed would. The case-study skill then takes a real, dated result and writes the proof asset, traced to the exact ledger line so the claim is defensible and flagged for consent so nothing ships without a yes.

Which Skills Build And Run The Whole Thing?

The first two skills are the ones that make the other ten worth installing. One builds the memory; the other runs the day off it.

11. /setup: the one-time interview, about thirty minutes, that writes your business context: your offer and pricing, your customer profile, your voice, your goal ledger, and a first-win plan. Every other skill reads what this one writes.
12. /daily: a short morning brief built from your numbers, plus inbox triage and a session log, so the system files what happened today into the right place without you sorting it.

Two more are worth knowing about once the basics are running. /business-review rolls your entire goal ledger into one periodic scorecard, which is where the month-over-month compounding becomes visible on a single page. And /churn-watch scores your existing customers for risk and writes a save play per account, because keeping a customer is cheaper than the lead research it took to win one. The point of the twelve is to start; the set goes further as the need shows up.

Can I Feel The Difference Before I Install Anything?

Yes. You do not have to install a thing to see why the saved-context version is worth it. Open Claude, paste the prompt below, and swap in your homepage URL. This is the manual, one-shot version of /free-plan: it reads the page and grades it, but it has no memory, so next month it starts from zero. That gap is exactly the point.

That prompt gives you a snapshot you have to re-explain every time. The installed skill does the same read, saves the result with today's date against your business context, and next month opens with what changed instead of a blank read. The prompt is the taste. The installed skill is the system that remembers, and remembering is the entire reason it compounds.

What Can These Skills Do, And What Can They Not Do?

Honest limits matter more than a feature list. These skills are good at reading, drafting, grading, and tracking against your saved context. They are not a replacement for your judgment, and they are not a database with guarantees. Here is the line, plainly.

The right mental model is a fast, tireless analyst who has read everything about your business and never forgets a number. You still make the calls. The skills remove the manual work around them and show you the change, so the calls are better informed than they were the month before.

Prompts Or Installed Skills With Memory?

This is the one comparison that decides whether AI pays off for a small business. The same model sits behind both columns. What changes is whether it reads your business first and writes its work back, or starts cold every time.

Read the bottom row first. A prompt resets; a skill compounds. Everything else in the table is a consequence of that one difference, and it is the difference between using AI and operating on it.

How Do You Install The Twelve?

Conversion Skills is the public method reference for these patterns. Conversion OS stays private, but the repeatable structure is visible: context first, then skills that read and write against it.

1. Start from the Conversion Skills product page and choose the first useful pattern.
2. Write a small context folder before you automate anything.
3. Run /setup. Claude interviews you and writes your business context in one session of about thirty minutes.
4. Run /free-plan on your own site to see a real result land and save. After that, every skill uses the context you just built.

You do not have to install everything on day one. Most owners start with two, the setup interview and the free plan, then add the rest as a real need shows up. If you want the next step automated, /autopilot sets up a scheduled agent that files meetings, chat, and email into your context for you, so the memory keeps filling without you typing notes. But that is a later move. Week one is one saved result, not a finished system.

Which Skills Should A Small Business Install First?

The order is the same for almost everyone, because the value depends on the memory existing before the other skills can use it.

Day one: run /free-plan with no setup, just to feel a graded result. Then run /setup to build the context. Without that folder, nothing compounds, so this is the step you cannot skip.

Week one: add /seo-audit on your site and /lead-research against your profile. One tells you where you are getting stuck; the other starts filling the top of the funnel. Both now save against the context you just built.

When the need shows up: reach for /email-sequence when a launch needs follow-up, /case-study when you land a result worth showing, and /daily once you want a morning rhythm. Install on demand, not all at once.

What Are The Common Mistakes?

Treating them like prompts. If you skip /setup and just run commands cold, you get a chatbot with extra steps. The context folder is the whole point; build it first or none of the compounding happens.

Skipping the first baseline. A plan with no saved result is a snapshot. The value is in the second run that shows the change, so save the first one even when the score feels rough.

Waiting for perfect data. These skills work from manual exports and public pages. Connectors and /autopilot are accelerators, not a prerequisite, so start with what you have on your disk today.

Installing all twelve at once. A wall of commands on day one is paralysis. Two skills you actually run beat twelve you installed and forgot.

Methodology

The skill descriptions reflect the public Conversion Skills method: each skill has a job, reads a defined context, writes a useful artifact, and leaves a trail the owner can inspect. Conversion System publishes no client outcome numbers, so none appear in this article; the proof here is the method shown, not a result claim. The aim is to give a small-business owner a concrete starter set grouped by job, so the first install is a decision and not a guess.

Related: Claude as a business operating system · The solopreneur Claude stack · Claude skills for an SEO plan · Run a free AI system scan with Claude