Lesson 1 of 12 · Claude Code Marketing

Two questions before you automate anything

Before a single build: the intake diagnostic. You will name a client's constraining step, pick the automation ladder rung it needs, and quote from evidence instead of a feature list.

Built from the course by Nick Saraev Source span 0:47:17–1:15:23 ~15 min lesson 6 quiz checkpoints

▶ Watch this chapter of the original video

💡 Dotted words are definitions

Any underlined term has a plain-English definition. Hover on desktop, tap on mobile.

✅ Quizzes keep score

Each section ends with a check. Running score in the top bar; final tally at the end. Your teacher-agent takes follow-up questions any time.

01 · The pipeline

Every business is the same four-stage pipeline

Nick Saraev frames every business, service or ecom, as one pipeline: marketing gets strangers to notice you, a sales transformation turns interest into money, fulfillment delivers what was promised, and an administrative wheel keeps the machine running. 0:48:34

Service-business pipeline from strangers through marketing, sales, fulfillment and administration to profit — 0:48:34 · slideStrangers in, profit out. The course automates only the left side of this picture.

On top of that pipeline he lays the RACE model: Reach, Acquire, Close, Expand. The line that matters for an agency sits between Acquire and Close: everything left of it is marketing, everything right of it is sales. This course, and your offers built from it, live entirely on the left. 0:49:58

RACE framework with Reach, Acquire, Close, Expand and the sales-transformation line — 0:51:28 · slideRACE with the marketing/sales boundary drawn between Acquire and Close.

Where does the marketing/sales boundary sit in the RACE model?

Between Acquire and Close. Reach and Acquire are the marketer's territory (impressions, then intent: form fills, bookings, opt-ins). Close and Expand are the sales transformation. When you quote a client, you are pricing improvements to Reach and Acquire.

02 · The levers

Six sliders on one control panel

The course's five marketing functions map onto six adjustable levers, grouped under the two marketing stages. Under Reach: creative (AI-generated images, video, audio at volume), copy (per-recipient personalization), and outreach (outbound at scale). Under Acquire: speed to lead, booking percentage, and follow-ups. 0:53:00

Marketing control panel with six sliders grouped under REACH and ACQUIRE — 0:53:00 · slide"Claude Code lets you push every slider up at once." The pitch of the whole course in one caption.

Two claims anchor the Acquire side, and both are the speaker's own numbers, logged in the claims ledger: he claims speed-to-lead systems can lift companies "300 to 400%", citing a home-services company that went from about $3M to $9M in monthly sales 0:55:37, and that shortening a 10-question form to 3 or 4 questions raises booking percentage, with Claude researching the prospect to replace the lost qualification data.

A client's ads get plenty of clicks but form fills sit unanswered for hours. Which lever is that?

Speed to lead. The clicks and form fills mean Reach is working; the leak is after intent appears. That is the Acquire side, and it is why the diagnostic in section 6 matters: more ad spend would amplify the wrong stage.

03 · The constraint

Bottleneck thinking: the factory math

Before building anything, the course installs bottleneck thinking. The factory analogy: 100 units of raw material arrive daily, step A can process 100 units per day, step B can process 10. The factory's output is 10 units per day. Double step A to 200 and output is still 10. Double step B to 20 and the whole factory doubles. 0:58:24

Factory diagram with 100 units in, step A at 100 per day, step B at 10 per day — 0:59:43 · slideThe narrowest step sets the throughput. Nothing else does.

Pipeline diagram warning DON'T UPGRADE THE WIDE PIPES, fix the booking flow first — 1:01:15 · slideThe marketing version: creative and leads are wide pipes; booking is the narrow one. Fix that first.

The consequence for your quotes is blunt: an automation system produces monetary return only when it targets the current constraining step. A creative system for a company whose constraint is booking is billable work that moves nothing. 1:02:17

In the factory example, doubling step A's capacity from 100 to 200 units per day changes total output by how much?

Nothing changes. Step B is the constraint; everything upstream of it just piles up. This is the single most useful sentence in a client meeting: "improving a non-bottleneck produces no extra output."

04 · The ladder

Prompt, skill, loop, routine

The second half of the diagnostic is the automation ladder: four rungs distinguished by what triggers them. A prompt runs when you type. A skill runs when you press run. A loop runs while your laptop is on. A cloud routine runs whether your laptop is on or not. 1:11:24

Progression from manual prompts through reusable skills and loops to scheduled routines — 1:05:00 · slideManual cart to self-driving car: the same process at four levels of autonomy.

The ladder slide with handwritten annotations around loops and routines — 1:09:00 · slideThe annotated version: loops are brittle (they need your machine, logins, and plugins intact); routines are the client-ownable surface.

Use cases per rung, straight from the source: prompts for quick experimentation, skills for repeat tasks you still supervise, loops for monitoring and scraping on your own machine, routines for daily production a business can own. The speaker claims routine economics of roughly $1 in tokens per ~$100 of deliverable value, logged as a claim. 1:08:22

What actually separates a loop from a cloud routine on this ladder?

The trigger surface. A loop is your laptop's cron; it dies with your battery, your logins, your plugin drift. A routine runs in a cloud environment on a schedule, API call, or webhook, which is exactly why it is the rung a client can own after you hand it off.

05 · The two rules

"Typed twice" and the 51.2% problem

Two rules route all the work you will do in later lessons. First, the escalation rule: the moment you type the same prompt twice, turn it into a skill. Not everything converts; if a task cannot pass the autonomy barrier, the human stays in the driver's seat, and that is information too. 1:09:53

Split panel contrasting repeated prompts with reusable skills — 1:11:00 · slide"A prompt you have typed twice should become a skill."

Second, the division of labor. Chain three steps that each succeed 80% of the time and the whole chain succeeds 0.8 × 0.8 × 0.8 = 51.2% of the time. The fix is not better luck: minimize the number of chained steps, and keep humans on the two ends, designing the process and judging the output, while AI executes the middle. 1:13:41

Whiteboard math showing 0.8 times 0.8 times 0.8 equals 51.2 percent — 1:14:27 · whiteboardThree 80% steps make a coin flip. Step count is a design decision.

Why does the course insist humans keep the design and evaluation steps rather than delegating the whole chain?

The multiplication. Every chained step compounds failure (0.8³ = 51.2%). Letting AI pick the process, run it, and grade itself multiplies three fallible judgments; a human-designed, human-judged process with AI in the middle keeps one fallible link instead of three.

06 · The win

The two-question intake diagnostic

Now assemble the lesson's deliverable, the thing you run on every Alexana client before quoting: a written diagnostic that answers exactly two questions.

Question 1: which step is the constraint? Walk the client's funnel left to right (creative → leads → speed-to-lead → booking → follow-ups) and find the narrow pipe: where does volume pile up or leak? Evidence beats opinion: response times, booking rates, follow-up counts. Question 2: which rung does it need? Match the fix to the lowest rung that solves it: a one-off analysis is a prompt; a repeated deliverable is a skill; monitoring is a loop; a system the client owns after handoff is a routine, which is also what you price as a product.

The output is one page: the named constraint, the evidence, the rung, and what you would explicitly NOT build yet (the wide pipes). Print the intake diagnostic card and the ladder card; they are this lesson compressed for a client call.

A client asks you to automate ad creative. Their forms answer within a minute, but 60% of booked calls never get a follow-up. What does the diagnostic say?

Sell the constraint, not the request. Follow-ups are leaking booked revenue; creative is a wide pipe. The diagnostic gives you the evidence to redirect the engagement, which is worth more to the client and positions you as the person who found the narrow pipe. (Lesson 10 builds exactly that machine.)

✔ Wrap-up

Your score

0 / 8

Answer the quizzes above to see your result.

One-sentence summary

Automation earns money only when it widens a client's narrowest funnel step, so every engagement starts with two written answers: which step constrains, and which ladder rung fixes it.

📖 Reference

Glossary

Every term marked with a dotted underline above. Skim it, or search the page for a word.

⚙ For your system

Ideas worth stealing into your own setup

Proposals, not done deals: nothing gets added to your system until you choose it.

01 The intake diagnostic as an Alexana biz-dev SOP

Saw: The course routes every build through bottleneck-plus-ladder before any tool is opened.
Why: Turns "what should we build you?" into an evidence document, and makes your quotes trace to a revenue lever.
Form: A one-page intake template in alexana-brand's biz-dev SOPs.
Source: 1:02:17 in the video

02 Speed-to-lead as the flagship client package

Saw: The claimed $3M-to-$9M home-services case rests on exactly the funnel stage your real-estate clients leak at.
Why: You already own the stack (forms, GHL, SMS); Lesson 8 builds the package end to end.
Form: A productized build for next-real-estate / rr-investment.
Source: 0:55:37 in the video