You can build the thing. That stopped being the bottleneck, and if you skipped
the plain-English version, the shape of the offer is what decides whether it pays. This
lesson holds one named model all the way down to the numbers: offer shape → unit
economics → the two-week validation loop. What it actually costs you to deliver one sale
now that a model sits in the loop, what to charge so a sale clears its own cost with margin to
spare, and the fourteen-day test that returns a signal you can act on instead of a feeling you
can argue with.
Updated August 1, 2026Refreshed monthlySources: Anthropic · Stripe · Gumroad · Paul Graham
Core concept 01
The build is cheap. The delivery has a meter now.
On February 2, 2025, Andrej Karpathy, formerly Tesla's AI director and a founding OpenAI
engineer, named the whole move: "there's a new kind of coding I call 'vibe coding,' where you
fully give in to the vibes, embrace exponentials, and forget that the code even exists." Describe
what you want, let the model write it. The phrase stuck hard enough that Collins Dictionary
named "vibe coding" its 2025 Word of the Year. Simon Willison, a longtime software engineer,
drew the honest line under it: if the AI wrote it and you reviewed it, tested it, and can explain
it, that is software development; if you can't and you ship it anyway, that's vibe coding, and his
rule is to never commit code to a real project he couldn't explain to someone else. Inside that
fence the tools are genuinely good. Claude Artifacts turns a chat description into a real
working page you can publish to a public link since July 13, 2026, and Gamma does the same
for decks and sites.
Here is what changes at this depth. When your product just hands over a static file, the build
cost is a one-time afternoon and every later sale is nearly free to deliver. But the moment a
model call sits inside each delivery, a real AI feature, a generated report, an agent that does
the work per buyer, delivery stops being free and starts running a meter. Every sale now
burns API tokens, and those tokens are a cost of goods you have to price against. The rest of
this lesson is that meter: what one delivery costs, what to charge so the sale clears it, and how
to read whether real people will actually pay before you build the fancier version.
Two lines come out of every sale before you keep a cent. The first is the processor
fee, and it is the one everyone fixates on and the one that matters least. Stripe charges
2.9% + 30¢ per successful charge with no monthly fee, Payment Links included; Gumroad charges
10% + $0.50 on a direct sale with no subscription. On a $29 offer, Stripe keeps about $1.14 and
Gumroad about $3.40. Real money, fixed and knowable, and never the thing that decides whether the
offer works.
The second line is the one this lane exists for: the cost to deliver one more unit with a
model in the loop. Anthropic prices Claude by the token, input and output separately. As of
2026, Claude Haiku 4.5 runs about $1 per million input tokens and $5 per million output tokens,
and Claude Sonnet 5 about $3 in and $15 out (check their pricing page, it moves). A light
delivery, say twelve thousand tokens on Haiku to generate one buyer's output, costs you roughly
three cents. A heavy delivery, an agent that loops a dozen premium-model calls to produce a full
report, can cost four to seven dollars every single time. Same $29 button, two completely
different businesses underneath it. Your cost to deliver is a real number you can measure by
running one delivery and reading the token count, and it is the difference between a margin
and a mirage.
Go deeper: measure it before you guess it
Do not estimate the token cost from vibes. Run one real delivery end to end and read the
actual input and output token counts the API reports back, then multiply by the per-token
price of the exact model you called. That is your measured cost of goods per sale, and
it is usually different from what you assumed, because the input includes every instruction
and every piece of context you stuffed in, not just the buyer's request.
Two levers pull that number down without touching the price. A cheaper model for the same
job: much of what people run on a premium model delivers just as well on Haiku at a fifth
of the output price. A shorter prompt: trimming the context you send on every call
cuts the input tokens on every sale forever. Both are engineering decisions that move the
unit economics, which is exactly the kind of decision an engineer gets paid to make.
Unit-economics waterfall · the fee is fixed; the token meter is the variable you designCore concept 03
Price to a contribution margin, not to the sticker.
Contribution margin is the number that survives one sale: price − processor fee −
cost to deliver one. It is what one more buyer actually contributes toward everything
else you have to pay for, and it is the only figure worth pricing against. A price you picked
because it "felt right" is a guess. A price you set so the contribution margin clears a floor you
chose on purpose is a decision, and it survives contact with the token meter.
Work it in that order. Measure your cost to deliver one. Add the processor fee. Decide the
margin you need per sale to make the whole thing worth running. Then set the price so the math
lands there, and check it against what the outcome is worth to the buyer and the next-best option
they already have. If a $29 delivery costs you $6 in tokens and $1.14 in fees, you are keeping
$21.86, and the honest question is whether $21.86 a sale, at the volume twenty warm people can
plausibly produce, is a business or a hobby. Sometimes the answer is to charge more. Sometimes it
is to make the delivery cheaper. It is almost never to hope for more volume.
Go deeper: the contribution margin is what covers your fixed cost
Every offer has two kinds of cost. Variable cost scales with each sale, the fee and
the tokens, and it comes straight out of the price. Fixed cost is what you pay whether
you sell zero or a hundred: hosting, the tools in your stack, the subscriptions the build
leans on. Contribution margin is the part of each sale left over to cover that fixed cost, and
after the fixed cost is covered, every further sale's margin is profit.
This is why a negative or razor-thin contribution margin can never be fixed by volume. If
each sale contributes two dollars and your fixed cost is two hundred a month, you need a
hundred sales just to break even before you have earned a cent. Reprice or re-shape the offer
until each sale carries real weight, then the volume math becomes a reason to celebrate a good
day instead of a trap.
Anthropic's advice to its own developers, the same line from Lesson 01, is the discipline
here: start with the simplest thing that works. Paul Graham told founders the same thing in 2013,
"do things that don't scale," solve the problem by hand as long as you can. A one-page site with a
real payment link and a delivery you run manually is the by-hand version, and it is the correct
choice, not the lesser one. You are not being paid to build the automated pipeline yet. You are
being paid to find out whether the automated pipeline would ever earn back its build cost, and the
cheapest way to learn that is to charge real money for the manual version first.
So run the loop. Build the smallest sellable version, charge real money for two honest weeks
in front of twenty real people, and measure. Real money is load-bearing: a free signup, a
"would you pay for this," a waitlist, none of them clear the bar, because nothing separates polite
interest from a buyer like a card. The signal you are reading is not revenue, twenty people will
never make you rich. The signal is conversion under real stakes, and it points at exactly
one of a few different fixes. Nobody clicked is a reach problem. Plenty clicked and nobody paid is
an offer problem. A couple paid is a live yes and a reason to build the next rung. None of those
fixes is "build something fancier," which is the thing you were tempted to do first.
Build → charge → measure → read · the loop returns a signal, not a feeling
Is it real in two weeks? · read the signal, then keep going or change exactly one thing
The two-week signal
What it means
Keep going, or change
Almost nobody clicked the button.
Reach
The offer might be fine; the wrong twenty people saw it, or the headline never earned the click. Keep the offer, change the audience or the promise above the button. Kill only after three genuinely different audiences all return zero.
Plenty clicked, nobody paid.
Offer
The interest is real and the offer is off: price, shape, or trust. Change one variable, not all three, and run it again. This is a reason to reprice or re-shape, not to abandon the idea.
One or two actually paid.
A live yes
Keep going. Two buyers out of twenty warm people is a real conversion, not luck. Don't celebrate a launch; build the next rung and reuse this rung's asset to make it cheaper.
They paid, then went quiet or asked for a refund.
Delivery
They bought the promise and the thing under it did not land. Fix delivery before you sell a second batch, or the margin you measured is a mirage that refunds itself away.
The margin is under your floor even at full price.
Economics
Kill this shape, or re-engineer it. A sale that can't clear its own fee and tokens is negative before volume even enters the room. No number of buyers fixes a unit that loses money. Reprice, cut the delivery cost, or change the offer.
Core concept 05
Price isn't set in a vacuum: watch CAC and payback.
The two-week loop uses twenty warm people you already have, and warm reach is free. The moment
you want more buyers than your own list can produce, you start paying to reach them, and that cost
has a name. Customer acquisition cost, CAC, is what you spend to get one new buyer:
total spend on reaching people divided by the buyers it produced. Spend $80 on an ad set and get
four sales, your CAC is $20. That number does not exist during the warm-list test, which is
exactly why the test is honest, and it becomes the whole game the second you try to scale.
CAC only means something next to your contribution margin. Payback is how many sales, or
how long, it takes for a buyer's margin to earn back what you spent to acquire them. A one-time
$29 offer with $21.86 of margin and a $20 CAC pays back on the first sale with $1.86 to spare,
thin but alive. The same CAC against a $2 contribution margin never pays back at all on a one-time
offer, which is why recurring revenue and CAC belong in the same sentence: a subscription earns
margin every month, so it can absorb a CAC a single sale never could. You do not need to run ads
to ship this month. You do need to price so that when you eventually pay for reach, the margin can
cover it, because a price set with no room for CAC is a ceiling you build before you start.
Break-even volume · fixed cost is a line; margin is the climb that has to cross it
The number you leave with
Your real unit economics, with the token meter in it
Fill these six lines for the offer you are actually considering, with real 2026 numbers you can
swap for your own. It takes about five minutes and it is the whole difference between a priced
offer and a hopeful one. Measure the cost to deliver by running one real delivery, not by
guessing.
01Your pricethe number on the button, e.g. $29
02Minus the processor feeStripe 2.9% + 30¢ ≈ $1.14 on $29 · Gumroad 10% + $0.50 ≈ $3.40
03Minus your cost to deliver oneAPI tokens per sale: ~12K tokens on Claude Haiku 4.5 ($1/1M in, $5/1M out) ≈ $0.03; a dozen premium calls can be $4–$7; $0 if no model call — check the pricing page
04That leaves your contribution marginline 01 − line 02 − line 03 — e.g. $29 − $1.14 − $0.03 ≈ $27.83 light, or ≈ $21.86 heavy
05Your fixed monthly costhosting + the subscriptions the build leans on, a real number, e.g. $200
06Sales to break even = fixed ÷ marginline 05 ÷ line 04, rounded up — every sale after that is profit. If break-even feels impossible from 20 warm people, the price or the delivery cost is what changes, not the effort
Fresh from the lab
What changed this month
Gumroad shipped a proposal-and-approve agent loop on top of its checkout API, on July 2,
2026. The new Agent tab reads a request in plain language, drafts the action, a refund, a
payout query, a support reply, and waits for your approval before it executes against your
account. It runs next to the Gumroad CLI the company shipped earlier this year, so a full
store, products, sales, payouts, licenses, webhooks, is now scriptable by a human or an agent,
the same read-propose-approve pattern most builder tools now default to for anything that
touches money.
Canva Code 2.0 dropped the cost of "AI builds the storefront" to zero for the first prompt,
on July 15, 2026. Every account, including free, gets 20 prompts a day; paid tiers get 60,
across Canva's 265 million monthly users. It's now a real third rail next to Claude Artifacts
and Gamma for the build step in this lesson's workflows, worth factoring in whenever a client's
budget covers the offer but not the page.
Claude Cowork's mobile and web rollout, starting July 7, 2026, changes your review
math. A build task now keeps running unattended after you close your laptop, and
Anthropic extended doubled usage limits through August 5. More of your product can ship between
the moments you actually look at it, so the review pass before anything touches payment data or
credentials needs to scale with the length of the unattended run, not with the size of the diff.
These are the words on every pricing
review and every "why isn't this profitable" post-mortem. Learn all six and the numbers under your
own offer stop being a mystery and start being a decision.
Unit economics
The profit and loss of a single sale, in isolation: what one buyer brings in and what that
one sale costs you to deliver. If the unit doesn't work, no amount of scale rescues it, it
just loses money faster.
Contribution margin
Price minus the variable cost of one sale, the processor fee and the delivery tokens. It is
what each sale contributes toward your fixed cost, and after that's covered, toward profit.
The number you actually price against.
COGS / cost to deliver
Cost of goods sold. What one more delivery costs you: near zero for a static
download, real dollars once a model call sits in the loop. Measure it by running one real
delivery and reading the token count.
CAC
Customer acquisition cost. Total spend to reach buyers, divided by the buyers it
produced. Zero on your warm list, which is why the two-week test is honest, and the whole game
the moment you pay for reach.
Payback
How many sales, or how long, before a buyer's margin earns back what you spent to acquire
them. A one-time offer must pay back on the first sale; a recurring one can absorb a CAC a
single sale never could.
Break-even
The number of sales where cumulative contribution margin finally overtakes your fixed
monthly cost. Before it, you're paying to run; after it, every sale is profit. It's a line
your margin has to climb past.
Make it yours
Ship one priced offer, with its numbers worked
Same model, your business. Pick your
lane and follow the steps. Every workflow below ends in one live, priced page in front of twenty
real people, with its contribution margin worked and a two-week test defined before you launch.
Budget an afternoon to build and two weeks to read the signal. A free Claude or Gamma account and a
Stripe or Gumroad account are enough to start.
For female founders
I have three offers I keep meaning to put a real page up for, and "this
weekend" has been the plan for two months straight.
The pilot cohort, priced to a real margin
Tool: Claude Artifacts or Gamma for the page · Stripe Payment Links for checkout · the mathbox above for the numbers
Pick the one offer you've already sold informally, and set a pilot price you can
defend against the outcome it delivers, not against your cost.
Work the unit economics before you build. Price, minus Stripe's 2.9% + 30¢, minus
your real cost to deliver one seat (your hours at a rate you'd accept, plus any per-buyer
model calls). That's your contribution margin per seat.
Have AI build the page in Claude Artifacts or Gamma, then add a Stripe Payment Link
at the pilot price with seats capped at what you can deliver well.
Define the two-week test: twenty warm people, real card required, and the kill line
written down first, e.g. "if nobody pays in fourteen days, the offer changes, not the effort."
Read the signal, not the revenue. A couple of paid seats is a live yes and a reason
to build the recurring version. Clicks with no buyers means reprice or re-shape one thing.
The payoff: the offer that lived in your head becomes a priced page with a
margin you can name, and a two-week answer instead of a two-month maybe.
For artists
My best pieces live in a folder on my phone, and there's nowhere anyone
could actually buy one.
The print shop, with the margin per print worked
Tool: Gamma for the page · Gumroad for print and commission checkout
Pick a tight drop: three to five prints or one open commission slot, each priced
against what the piece is worth to a buyer, not what the paper cost.
Work the margin per print. Price, minus Gumroad's 10% + $0.50, minus your real
per-print delivery cost: printing, packaging, shipping, and your handling time. What's left is
the contribution margin, and it needs to clear your floor before you list it.
Have AI build the shop page in Gamma from your images and details, then add Gumroad
checkout per size or a Stripe Payment Link for a commission deposit.
Define the two-week test: your close-friends list and studio mailing list, real
checkout, and a written kill line for zero sales.
Read the signal. A print or two sold is a live yes; open the next drop. Views with
no buyers is a price or a framing problem, change one.
The payoff: the finished work finally sells at a margin you actually kept,
instead of a number you picked and hoped covered the shipping.
For actors
I coach other actors through self-tapes between my own auditions, and I have
never had one link to send when someone asks how to book me.
The coaching session, priced by the hour it costs you
Tool: Claude Artifacts or Gamma for the page · Calendly for booking, Stripe Payment Links for payment
Pick one session type, a 45-minute self-tape review, priced once against the result
it gets, not the DM rate you've been mumbling.
Work the margin per session. Fee, minus Stripe's 2.9% + 30¢, minus the honest cost
of your time in the room at a rate you'd accept. A live service is nearly all hours, so the
margin is what tells you if the price respects your time.
Have AI build the page in Claude Artifacts with your reel and bio, then embed a
Calendly link and a Stripe Payment Link so a booking is paid before it's confirmed.
Define the two-week test: your scene-study group and last class, real payment at
booking, and a kill line for zero bookings.
Read the signal. One or two paid bookings is a live yes; add a small group session
as the next rung, which delivers more margin per hour.
The payoff: the side hustle stops running through ten DM threads at a fuzzy
price and starts running through one link at a margin you chose.
For musicians
People ask "do you do weddings?" and "can I buy the record?" constantly, and
I still don't have one link that answers either.
The one-link shop, with two margins side by side
Tool: Gamma for the page · Gumroad for the digital bundle, Stripe Payment Links for booking deposits
Pick two priced things: a digital bundle, the record plus two unreleased tracks, and
a private-event booking rate. One is nearly pure margin, one is your hours.
Work both margins. The download: price minus Gumroad's 10% + $0.50 minus near-zero
delivery, so margin is roughly 90%. The booking: rate minus Stripe's fee minus your real
time. Knowing which is which tells you where the real income is.
Have AI build one page in Gamma that holds the shop and the booking pitch cleanly,
with Gumroad for the bundle and a Stripe Payment Link for booking deposits.
Define the two-week test: mailing list plus the fans who comment on everything, real
checkout on both, and a kill line for each.
Read the signal. Bundle sales prove the passive lane; a booking deposit proves the
premium one. Chase the margin that showed up, not the one you assumed would.
The payoff: the midnight "do you do weddings?" gets answered by a link at a
known margin, and you learn which half of the business actually pays.
For fitness pros
I keep coaching people one-on-one in my DMs for free, because I've never
built the page that would let them just buy the program instead.
The 4-week program, priced two ways
Tool: Gamma or Claude Artifacts for the page · Stripe Payment Links or Gumroad for checkout
Pick the program you already coach from memory, and decide its shape: a self-paced
download (near-zero delivery) or a live cohort (your hours every week).
Work the margin for the shape you chose. Download on Gumroad: price minus 10% + $0.50
minus almost nothing, so it scales while you train other people. Live cohort on Stripe: price
minus the fee minus your coaching hours, a very different margin at the same sticker.
Have AI build the page in Gamma or Claude Artifacts laid out week by week, with the
matching payment link.
Define the two-week test: the clients you've coached this exact program on for free,
real checkout, and a written kill line.
Read the signal. Buyers on the download means passive income is real; buyers on the
cohort means premium is real. Refunds mean fix the delivery before the next batch.
The payoff: the program you gave away in the DMs starts paying you back at a
margin you can see, in the shape that actually fits your week.
For coaches
I run a great workshop live in my head every few months, and it's never once
had a real page or a real way to pay for it.
The paid workshop, priced to fill and clear its cost
Tool: Gamma or Claude Artifacts for the page · Stripe Payment Links for the ticket
Pick one workshop, one date, one price, set against the outcome attendees leave with,
not the hours it takes you to teach.
Work the margin per seat. Ticket minus Stripe's 2.9% + 30¢ minus your near-zero
per-seat delivery cost (a live workshop's cost is mostly fixed, your prep time, spread across
every seat, which is why filling it matters).
Have AI build the page in Gamma with the outline, date, and a clear price, and cap
seats at the group size you can run well.
Define the two-week test: the clients who've said "I wish you taught this," real
payment for the ticket, and a break-even seat count written down first.
Read the signal. Hit break-even seats and it's a live yes; run it again and record it
as a download. Empty seats means the price, the topic, or the audience, change one.
The payoff: the someday workshop becomes a dated page with a break-even
number, so you know exactly when it started paying.
For therapists
I answer the same three questions on every intake call, and I've never built
the page that would answer them without me on the phone.
The self-serve resource, priced and privacy-first
Tool: Gamma or Claude Artifacts for the page · Stripe Payment Links for checkout · no client details in the builder, ever
Draw the line first, then pick the thing: a downloadable coping-skills guide or a
niche workshop. Zero client information goes into any AI tool, only your own general practice
content.
Work the margin. A guide is near-pure margin: price minus Stripe's fee minus almost
nothing, so a second sale is essentially free. A workshop is your prep time spread across
seats. Both are simple to price once you know which one you built.
Have AI build the page in Gamma answering the repeat intake questions, with a Stripe
Payment Link kept fully separate from your clinical booking system.
Define the two-week test: your waitlist and referring colleagues, real checkout, and
a written kill line.
Read the signal. Buyers on the guide is a live yes and a passive lane; the intake
hour shrinks either way because the page did the explaining.
The payoff: the repeat questions get answered by a page that pays a real
margin, and the privacy line stays exactly where it should.
For authors
My next book has a real preorder date and zero pages online that actually say
so.
The reader-magnet page, with the preorder math honest
Tool: Gamma or Claude Artifacts for the page · your retailer's or publisher's preorder link · Gumroad for a paid companion
Pick the funnel: a free chapter that leads to the preorder, plus one optional paid
companion, a workbook or a bonus story, that carries its own margin.
Work the margin on the paid piece. A digital companion on Gumroad is price minus 10%
+ $0.50 minus near-zero delivery. The free chapter is a reach cost, not a product, so treat it
as marketing, not revenue.
Have AI build the page in Gamma to move a reader from free chapter to preorder in one
scroll, with the real retailer link and an email capture for launch day.
Define the two-week test: your newsletter and last ten reader emails, a real preorder
click (and a real purchase on the companion), and a kill line for silence.
Read the signal. Preorders and companion sales together prove demand before launch
week; low click-through means the promise above the button, not the book, needs the change.
The payoff: preorders start counting before launch-week panic, from a
page that earned its own small margin weeks earlier.
For chefs
Private dinner requests come in over text, and I quote every single one from
scratch, usually at 11pm.
The private-dinner package, priced to clear food and hours
Tool: Gamma for the page · Stripe Payment Links for the deposit
Pick one package, a fixed headcount and price per head, instead of an infinite custom
quote at midnight.
Work the margin per dinner. Price per head times guests, minus Stripe's fee on the
deposit, minus your real cost to deliver: ingredients, travel, and your hours in the kitchen
and the room. That number tells you if the package pays or just keeps you busy.
Have AI build the page in Gamma with the menu style, price, and minimum, and a Stripe
Payment Link for a deposit that locks the date.
Define the two-week test: past guests and the followers who've asked, a real deposit
required to book, and a written kill line.
Read the signal. A booked deposit is a live yes at a margin you worked; requests that
stall at the price means reprice the package or trim the delivery cost.
The payoff: quoting leaves your 11pm texts for a page that quotes itself at a
margin you already checked.
For consultants
Every new client starts with the same free 45-minute call, and I still don't
have a page that would let them just book and pay for it.
The paid diagnostic, priced as a real deliverable
Tool: Claude Artifacts or Gamma for the page · Stripe Payment Links for the diagnostic fee
Turn the free call into a paid diagnostic, priced against the value of what they
leave with, not the hour it takes you.
Work the margin. Fee minus Stripe's 2.9% + 30¢ minus your real cost to deliver: your
prep and call time, plus any per-client model work if you generate part of the deliverable
with AI (measure those tokens, don't guess them).
Have AI build the page in Claude Artifacts so it reads like a real service with two
short case studies, and add a Stripe Payment Link due at booking.
Define the two-week test: your warmest pipeline, real payment at booking, and a kill
line if nobody converts.
Read the signal. A paid diagnostic booked is a live yes and a filter for buyers who
are ready; the paid version screens out tire-kickers the free call attracted.
The payoff: the free call that ate your calendar becomes a priced first
deliverable with a margin, and it pre-qualifies every buyer.
For health advocates
Families ask the same five questions before they ever become a client, and I
answer them one exhausted phone call at a time.
The resource guide, priced and privacy-first
Tool: Gamma or Claude Artifacts for the page · Stripe Payment Links for checkout · no client details in the builder, ever
Draw the line first: no client or patient information goes into a general AI tool.
Build the page from your own general process, then pick the priced thing: a guide or a paid
navigation session.
Work the margin. A guide is near-pure margin once built. A paid session is your hours;
fee minus Stripe's fee minus your real time. Price against what it saves the family, not what
it costs you.
Have AI build the page in Gamma answering the five recurring questions, with a Stripe
Payment Link kept separate from any clinical system.
Define the two-week test: referring providers and your waitlist, real checkout, and a
written kill line.
Read the signal. A sale on the guide is a live yes and a passive lane; the first
exhausted phone call gets shorter regardless, because the page already explained.
The payoff: the repeat explaining becomes a page that pays a margin, with
the privacy line intact and the phone quieter.
For makers
My next collection is finished before I ever get around to building the page
that would actually sell it.
The drop, with the margin per piece worked
Tool: Gamma for the page · Gumroad for the drop
Pick one drop, a fixed number of pieces on a set date, and price each against what a
buyer will pay, not what the materials ran.
Work the margin per piece. Price minus Gumroad's 10% + $0.50 minus your real cost to
deliver: materials, packaging, shipping, and the time to make and mail it. What's left is what
the drop actually earns you per unit.
Have AI build the drop page in Gamma from your photos with a countdown to the
release, and Gumroad checkout per piece that goes live when the drop opens.
Define the two-week test: your waitlist and best repeat customers, real checkout, and
a written kill line.
Read the signal. Pieces sold at that margin is a live yes; schedule the next drop.
Views with no buyers means the price or the piece, change one and re-drop.
The payoff: the finished collection gets a release day and a margin per
piece, instead of a shelf and a guess.
For nonprofit leaders
Our biggest campaign of the year still runs through a link I made in 2019, and
it shows.
The campaign page, with the true cost per dollar raised
Tool: Gamma or Claude Artifacts for the page · Stripe Payment Links or your donor platform's checkout
Pick one campaign, one goal number, one deadline, with a clear ask instead of a
generic donate button.
Work the true cost per dollar raised. Every gift loses the processor fee (Stripe's
2.9% + 30¢, or your donor platform's rate), so a $50 gift nets about $48.55. Knowing the net,
and any per-donor tooling cost, keeps your goal honest and your impact claims accurate.
Have AI build the page in Gamma with the goal, a progress bar, and two real program
details, and add the checkout your board already trusts.
Define the two-week test: your board and top past donors, real gifts, and a written
line for what "working" looks like before you scale it wider.
Read the signal. Gifts from the warm list is a live yes; take it to the full list.
Silence means the story above the button, not the cause, needs the change.
The payoff: the campaign gets a page that matches the mission and a net you
can report honestly, not a leftover link from six years ago.
For photographers
Every mini-session season I rebuild the same booking spreadsheet instead of
the page that would run itself.
The mini-session page, priced to fill and clear its cost
Tool: Gamma or Claude Artifacts for the page · Stripe Payment Links for the session fee
Pick one package, a fixed slot length, price, and number of slots, priced against the
gallery a client leaves with, not the hour behind the camera.
Work the margin per session. Fee minus Stripe's 2.9% + 30¢ minus your real cost to
deliver: shooting time, editing hours, and any print or gallery-host cost. Editing is the
hidden line, count it or the margin is fiction.
Have AI build the page in Gamma with the gallery, price, and slots, and a Stripe
Payment Link that reserves a slot on payment.
Define the two-week test: last season's clients and your mailing list, real payment
to book, and a break-even slot count written down first.
Read the signal. Slots filling at your margin is a live yes; retire the spreadsheet.
Empty slots means the price or the promise, change one before next season.
The payoff: the spreadsheet retires and the page fills slots at a margin that
counts your editing hours, not just your shutter time.
For realtors
I know this neighborhood cold, and I've never once turned that into a page
that brings people to me first.
The neighborhood guide, with the true cost per lead
Tool: Gamma or Claude Artifacts for the guide · your CRM's lead form for the opt-in
Pick one neighborhood you actually farm, and treat the guide as a lead magnet: the
product is the buyer or seller it eventually produces, not a sale on the page.
Work the cost per lead, not a margin. Your time to build the guide plus any tooling
cost, divided by the leads it captures, is your CAC for this channel. Compare it to your commission
so the effort is priced against the outcome.
Have AI build the guide in Gamma so it reads like a neighbor wrote it, with a
lead-capture form tied to your CRM in exchange for the full download.
Define the two-week test: past clients in that area and your sphere, a real opt-in
required, and a written line for how many leads makes it worth repeating.
Read the signal. Real opt-ins from warm people is a live yes; expand the reach. No
opt-ins means the guide's promise, not the neighborhood, needs the change.
The payoff: your expertise becomes a lead magnet with a cost per lead you can
compare to a commission, working while you're at a showing.
For stylists
New clients still find me through a DM back-and-forth instead of a page that
would just let them book.
The book-me page, priced by the chair time it costs
Tool: Gamma or Claude Artifacts for the page · Stripe Payment Links or your booking software
Pick one signature service as the front door, priced against the result, not a full
menu that overwhelms a new client.
Work the margin per appointment. Price minus the processor fee minus your real cost
to deliver: chair time and any product used. A service is mostly your hours, so the margin is
what tells you the price respects them.
Have AI build the page in Gamma with the price, what's included, and three photos,
and add a Stripe Payment Link for a deposit that holds the appointment.
Define the two-week test: current clients (they refer), your bio link, and last
month's DM inquiries, with a real deposit to book and a written kill line.
Read the signal. Deposits from new clients is a live yes; booking moves out of the
DMs for good. Inquiries that stall at the deposit means reprice or reframe the front-door
service.
The payoff: booking leaves the DMs for a page that works at 1am at a margin
you chose, deposit already in hand.
The house rule
That's the last lesson in the school, so this is where Lauren's line has to hold for the whole
thing: what belongs to you and keeps your humanity, you keep. The offer, the price, the way you
talk to your buyers, that's yours, always. What frees you to be more human away from your laptop,
you hand off, and everything this lane taught you, the token meter, the margin, the two-week
loop, is in service of that trade, not the other way around. You learned the engineering so the
numbers would stop being a mystery, not so the numbers could run you. Now go price one real thing,
charge real money for it, and read the signal. The party started. You're not late.