# Start & Ship — full content (llms-full.txt)

> This file concatenates the full substantive text of this site — offers,
> FAQ, and every blog post — for AI systems and agents that ingest one
> document rather than crawling multiple pages. For a short curated index
> instead, see /llms.txt. Like llms.txt, this is not a Google ranking
> signal; it's for non-Google AI/agent ecosystems.

A production-ready Next.js codebase for launching content and review sites — legible to AI systems, driven by AI coding agents.

## Offers

Founder: $99 launch price ($149 regular) — the full kit, private repo, lifetime updates.
Founder Pro: $149 launch ($249 regular) — everything in Founder plus 3 vetted niche dossiers, capped at 10 buyers.
Agency: $299 launch ($399 regular) — white-label rights for unlimited client sites.
All sales are final; if delivery itself fails, it's always made right. Launch pricing ends 2026-08-19 and will not be extended.

## FAQ

Q: What exactly do I get?
A: Access to a private GitHub repo containing the full kit described above, plus lifetime updates via git tags. One-time purchase, no subscription.

Q: What literally happens after I buy?
A: Checkout runs through Polar (the merchant of record). The moment payment clears, Polar automatically grants your GitHub account access to the private repo, with no waiting on a manual invite. Lifetime updates arrive as git tags on that repo. For the actual day-by-day path from there to a live site — agentic or manual — see /how-it-works.

Q: What's the refund policy?
A: All sales are final. Once Polar grants your GitHub account repo access, the code can't be un-downloaded, so a blanket 'no questions asked' window on a digital product isn't honest trust-building, it's just an invitation to clone the code and ask for the money back. The one exception: if delivery itself fails, we fix it or refund you immediately, no argument. That's on us, not a favor.

Q: Is the launch discount real, or will it be 'extended'?
A: It's real and it ends 2026-08-19. The launch terms above are signed and dated. No extensions, no re-runs, no coupon that mysteriously always works. If anything changes after launch, prices go up, not down.

Q: Do I need Claude Code or Cursor?
A: No. It's a clean Next.js + TypeScript codebase you can work on by hand. But the agentic layer is the headline feature, and the docs assume you'll point an agent at it at least once.

Q: I'm not a professional developer — can I use this?
A: If you can run Claude Code (or Cursor/Codex) and follow docs, yes: the kit ships with a setup prompt that walks an agent from clone → configure → deploy, and the docs assume the agent does the heavy lifting. If you've never touched a terminal, there's no guided, click-only tier yet — email us if you'd want one, that's the only signal that changes it.

Q: How is this different from ShipFast?
A: Same shape: one-time repo purchase, opinionated architecture. Different substance: ShipFast encodes SaaS plumbing; this encodes 4 months of operating content and affiliate sites, from schema factories to FTC components to the source ledger.

Q: Why not just ask Claude to build this myself this weekend?
A: You could ask it to build *a* site. The question is whether you'd know to ask for a tracked-redirect affiliate registry, FTC disclosure on every recommendation (not just a footer), the correct JSON-LD type per page (a comparison isn't a Review), and a citation-staleness gate that fails the build instead of letting sources rot — an agent starting from a blank repo only builds what it's told to, and it'll happily ship something that looks finished without any of that. We ran exactly this test: a fresh agent, given only this repo's docs and no peek at the internals, shipped two working pages on the first try without opening a single schema file. That's not a claim you could build a Next.js site with an agent — of course you could. It's that getting the specific decisions right, in the right shape, on the first pass, is what you're paying for.

Q: What are the license terms?
A: Founder and Pro: unlimited sites you own. Agency: white-label rights for unlimited client sites. No resale or redistribution of the kit itself.

## Blog

## How to Pick an Affiliate Niche: The Five Checks Before You Build

Published: 2026-07-28
URL: https://startandship.com/blog/how-to-pick-an-affiliate-niche

The architecture is not the hard part. A content site is a solved engineering problem. The niche is the part that decides whether the site earns, and it's the one decision no template can make for you. What follows is the checklist actually used before building the 4 production sites behind Start & Ship. Five checks in a deliberate order, and the keyword research everyone starts with is third.

FIRST, THE MISTAKE WORTH LEARNING FROM

One site in this portfolio was chosen purely on search data. The keyword cluster had roughly ten times the monthly volume of anything else attempted, with a difficulty score under 30. On paper it was the best opportunity of the four.

Today it has the highest Domain Rating of any site here, 32, and the most backlinks, around 1,200. By every input the advice industry tells you to chase, it is the strongest property in the portfolio. It also produced about 3,370 impressions and five organic clicks.

The reason is not technical. There is nothing of the operator in it: no personal interest in the niche, no hands-on hours, nothing a reader couldn't get from any of the ten other pages answering the same query. Keyword research told the truth about where demand was. It could not tell whether this was a person capable of earning that demand.

Then the part that complicates the lesson: that same five-click site has been one of the strongest earners in the portfolio in some months, because the niche sells expensive, seasonal gear and one sale is worth more than dozens elsewhere. Worst traffic, real money. Traffic is not revenue, and the tools measure the first one.

CHECK 1 — CAN YOU ACTUALLY TEST THE PRODUCTS?

This is first because it's the check that fails silently. Every review template in the kit requires a testedOn date and a sourceUrl on every claim, which means a niche you can't get your hands on is a niche where you'd be writing marketing copy with citations bolted to it.

Ask whether you can buy or access a representative product within your budget, whether there's a rental or loaner path, and how long it takes to form a real opinion. That last one is your content velocity ceiling: a niche where a genuine verdict takes a month of use is a niche where you publish twelve times a year.

CHECK 2 — DOES THE AFFILIATE MATH WORK, AND WILL THEY TAKE YOU?

Two separate questions, and the second one sinks more sites than the first.

The math: commission rates vary from about 1% to 10% by Amazon category, while direct brand programs commonly run 10-30%. Average order value matters more than the percentage. Five percent of a $30 product and five percent of a $500 product are not the same business. A recurring purchase category means returning readers rather than one-time search traffic.

The access trap: another site here has been cited by ChatGPT 505 times across 27 pages and shown by Google 65,600 times. Its revenue is zero. Travel-rewards credit cards are among the highest-paying affiliate categories on the internet, with published program terms in the $100-200 per approved application range, and those programs decline brand-new sites without traffic. You need traffic to get into the program and the program to make the traffic worth anything. Read the program terms during niche research, not after thirty pages are live.

CHECK 3 — IS THERE REAL SEARCH DEMAND, AND CAN YOU PROVE IT?

The question is narrow: do people search this, how many, and is the competition beatable by a domain with no authority yet.

The working process is a cluster hunt, not a keyword hunt. For the eSIM site it was a group of "best eSIM in [country]" queries pulling thousands of searches a month at a difficulty score under 30. A cluster is what becomes twenty mutually reinforcing pages. A single keyword is one page and one point of failure.

Start free. Google Keyword Planner gives rough volumes for nothing through a Google Ads account. Mangools has a free tier with a daily cap on difficulty lookups, enough to check a handful of candidate niches properly.

Pay only while researching. When comparing many clusters at once, a paid month is worth it. Mangools KWFinder ran about $61/month on its entry tier as of mid-2026 and is the tool behind the niche selection on these sites. Semrush is the heavier alternative, with deeper competitive analysis and a price to match. Software pricing moves constantly, so check the current number before committing. (Both links on the styled version of this page are affiliate links; the free-tier advice is the same either way.)

This is a research cost, not a platform subscription. Rent the tool for the month you're validating, then cancel. That is categorically different from the $25-60/month a page builder charges forever to keep your own site online.

One more check most people skip: look at who currently ranks. Empty results can mean opportunity, and they can equally mean nobody is looking.

CHECK 4 — CAN YOU SUSTAIN IT?

Do you have enough genuine interest to keep testing and writing after the novelty wears off? This is the check the five-click site failed, and it failed it at selection time, not during execution.

Also ask whether the category is stable. A niche where products, prices, and specs turn over monthly means constant re-verification work. The kit ships a citation-staleness gate that fails the build when sources rot: protection in a stable niche, a treadmill in a volatile one.

CHECK 5 — SCORE IT, HONESTLY

Rate each dimension 1 to 5, in writing, before building anything: you can test the products firsthand; the commission economics are real and the programs will accept a new site; search demand exists and you have checked it; you can sustain content production past the novelty; you would still care about this in six months.

There's no passing score. It's a forcing function to write down what you know versus what you're assuming. The site that produced five clicks would have scored a 5 on search demand and a 1 on the last two rows, and writing that down would have been enough.

WHAT THE TOOLS CAN AND CANNOT TELL YOU

A keyword tool measures demand. It cannot measure whether you're the person who can earn that demand, whether the merchants in the category will accept a new site, or whether one sale is worth $3 or $300. Three of the five checks above are outside its reach, and in this portfolio those three determined the outcomes.

Run the demand check. Just don't confuse it with the decision. For the ongoing numbers behind all of this, including the months that underperformed, the operator's ledger is public at https://realnumbers.online.

FAQ

Q: What keyword difficulty should I look for?
A: Under 30 is the working rule behind every site in this portfolio, and it's a reasonable starting filter for a brand-new domain with no authority. But treat it as a filter, not a verdict: the highest-volume, sub-30-difficulty niche in this portfolio produced five organic clicks. Difficulty tells you whether ranking is plausible. It says nothing about whether you can write something worth ranking.

Q: Do I need a paid keyword tool to pick a niche?
A: No. Google Keyword Planner is free and gives you rough volumes, and Mangools' free tier gives you a limited number of difficulty lookups per day, which is enough to check a handful of candidate niches. A paid tier is worth it when you're comparing many clusters quickly and want difficulty scores without a daily cap. Rent it for the month you're actually researching, then cancel.

Q: Should I pick a niche I'm passionate about or one with good numbers?
A: The numbers decide whether the demand exists. Genuine familiarity decides whether you can produce anything a reader prefers to the ten other pages answering the same question. This portfolio has one site chosen purely on search data and no personal interest, and it gets almost no organic traffic despite the strongest backlink profile of the four. You need the demand check to pass and something real to say. One without the other is what produces a site nobody reads.

Q: How do I know if a niche's affiliate programs will actually accept me?
A: Read the program terms before you build, not after. The highest-paying categories, travel rewards and credit cards among them, publish rates in the $100-200 per approved application range and then decline brand-new sites with no traffic. That's a chicken-and-egg trap worth discovering during research rather than after you've written thirty pages. Check whether there's an Amazon Associates category as a floor, and whether direct brand programs list traffic minimums.

Q: How many keywords should a niche have before it's worth building?
A: Look for a cluster, not a keyword. One high-volume term is a single page and a single point of failure. What you want is a group of related questions people actually search — the best X, X vs Y, X for a specific use case — because that's what becomes twenty pages that reinforce each other, and it's what tells you the interest is a category rather than a moment.

---

## Making a Content Site AI Actually Cites

Published: 2026-07-22
URL: https://startandship.com/blog/making-a-content-site-ai-actually-cites

A citation isn't a ranking. Ranking well in Google and being the source an AI system quotes back to a user are two different outcomes, judged by two different kinds of readers, and most content sites are built for only the first one. This site is a working example of the second: the JSON-LD in its pages, the FAQPage schema under its FAQ blocks, and the /llms-full.txt file concatenating the entire site into one document are all real and live right now, not a mockup of what AI-readiness could look like.

WHAT "CITABLE" ACTUALLY MEANS

Three things, none of them exotic: the claim is dated, so a system can judge whether it's still current; the claim is sourced, so it can be checked instead of trusted; and the page is structured so machine parsing doesn't require guessing what a blob of prose means. A human reader tolerates ambiguity. A retrieval system either finds a clean answer or moves to the next result — there's no partial credit for a page that's directionally right but structurally unreadable.

THE FOUR PIECES, AS THEY ACTUALLY EXIST ON THIS SITE

llms.txt — a curated index at /llms.txt: what the site is, the real offers with real prices, and links to the fuller documents. It's the front door for a system that wants a summary before deciding whether to read further.

llms-full.txt — the whole site's substantive text in one file, for systems that ingest a single document rather than crawl a sitemap. Every offer, the full FAQ, and every blog post's plain-text mirror, concatenated. It exists because crawling is expensive and unreliable for a lot of these systems; one file removes the excuse not to read the whole thing.

JSON-LD, per page type — Product schema on the pricing offers with real prices and availability, Article schema on every blog post with a real publish date, FAQPage schema wherever there's a real FAQ block. Not one schema pasted everywhere; the schema type matches what the page actually is, because a mismatched type is worse than none.

Dated, sourced prose — the part no template can do for you. Every number on this site traces back to site.config.ts or the public ledger at realnumbers.online, so nothing here can quietly drift out of sync with the number a system might cite. Structure gets a page read. Real, checkable content is what gets it trusted.

WHY MOST SITES SKIP THIS

Not because it's hard, exactly — it's tedious in a way that's easy to defer indefinitely. A blank Next.js repo has no JSON-LD factory, no llms.txt convention wired to a config file, no pattern for keeping a plain-text mirror honest against the styled version of a post. Every site ends up rebuilding this from scratch, once, badly, or not at all. It's wired into every page template in Start & Ship from the start, for the same reason the rest of the kit exists: not because any one piece is hard, but because rebuilding the same foundation for every new site is what actually eats the time.

FAQ

Q: What is llms.txt?
A: A plain-text index at the root of a site — /llms.txt — written for AI systems and agents instead of human browsers or Google's crawler. It's a proposed convention, not a guarantee any given system reads it, but it costs nothing to publish and it's exactly the kind of legible summary these systems are built to prefer over scraping rendered HTML.

Q: Does this replace normal SEO?
A: No. Traditional SEO — page speed, backlinks, on-page structure — still matters, and this site isn't claiming otherwise anywhere. The AI-surface layer is additive: it targets a different consumer (an LLM doing retrieval or answering a question) that traditional SEO wasn't built for.

Q: Do I have to maintain the JSON-LD and llms.txt files by hand?
A: The structure is templated so you fill in content once, in site.config.ts or a post's frontmatter, and it propagates to the JSON-LD, llms.txt, and llms-full.txt automatically. The one manual step is a plain-text mirror of each blog post's body, kept in the post file itself, because Next.js Route Handlers can't render React to a string — that's a deliberate constraint, not an oversight, and it's documented right next to the code.

Q: Will this actually get my site cited?
A: No one can promise that, and this site doesn't. What it can promise is that the content is structured the way these systems are documented to prefer: dated, sourced, machine-parseable. That's a real, checkable difference from a site with none of it — not a citation guarantee.

---

## The Fastest Way to Start Affiliate Marketing in 2026

Published: 2026-07-16
URL: https://startandship.com/blog/fastest-way-to-start-affiliate-marketing

The fastest real path: pick one specific niche, build on architecture that already works instead of a blank repo, publish a small number of genuinely useful pages with real citations, and start distributing before the site feels finished. Most of the time people lose isn't spent on any of those four things. It's spent rebuilding the same boilerplate every other content site also needs, from scratch, alone.

This isn't a theory post. Below is the actual sequence, plus what it really looked like the last time we ran it.

THE HONEST TIMELINE

Four production content and affiliate sites, 4 of them live right now, 4 months into actually running this — not a weekend project, and not a case study built for this post. Real domains, real citations, real affiliate programs. We report the revenue behind them, month by month, on the public ledger at realnumbers.online — that ledger covers more ground than just these four sites, so we'd rather point you to the real source than pull a single number out of it for this post.

STEP 1 — PICK A NICHE (THE ACTUAL HARD PART)

Nothing below matters if the niche is wrong. You want somewhere you can credibly write from, with real affiliate programs attached to real products, that isn't already saturated by sites with a decade of head start. Keyword and competition research tools do most of the legwork here — Mangools (https://mangools.com/kwfinder#a69d41a746aee0870acb53af1) is a reasonable, affordable one to start with. If you'd rather skip the research entirely, that's literally what the Founder Pro tier includes: three vetted niche dossiers instead of weeks of it.

Disclosure: some links in this article, including the one above, are affiliate links. We may earn a commission if you buy through them, at no extra cost to you.

STEP 2 — STAND UP THE TECHNICAL FOUNDATION IN A WEEKEND, NOT A QUARTER

This is the step that eats months when nobody tells you otherwise: auth you don't need, a database for content that's just text, a homegrown schema for affiliate links and citations that takes weeks to get right and is invisible when it's done. We built Start & Ship because we kept rebuilding this same foundation for every new site. It's static, one config file, the six page templates that actually earn, and the AI-search and affiliate plumbing already wired in — so this step becomes a weekend, not the reason a site never ships.

STEP 3 — PUBLISH A FEW GENUINELY USEFUL PAGES

Not the full catalog on day one — enough that the site doesn't look empty to the first real visitor. Every page needs a real, checkable source behind every claim; "trust me" content is exactly what both readers and AI-search systems have gotten better at discounting.

STEP 4 — DISTRIBUTE BEFORE IT FEELS DONE

Pick one channel where your actual audience already is. The first post is a real answer with one concrete receipt attached, not an ad. Show up and contribute before you ever mention your own site. The site with three pages and real distribution beats the site with thirty pages and none.

FAQ

Q: Is affiliate marketing dead in 2026?
A: No, but the traffic economics changed — AI answers questions directly now, so fewer visits convert into clicks. It's a real headwind, not a myth, and it's exactly why the technical and citation work in step 3 matters more than it used to, not less.

Q: Do I need a large following to start?
A: No. Content and SEO-driven affiliate sites can work without an existing audience — that's the whole model behind the four sites cited above. It's real work either way; a following just changes which kind.

Q: How much does starting actually cost?
A: A domain runs roughly $10-15/year, and a static site hosts for effectively $0/month. Everything past that — a starter kit, keyword tools, content help — is optional, scaled to how much time you're trying to buy back.

Q: How long does this really take?
A: The technical setup can be a weekend. A site with real traffic and revenue is months, not days — see the honest timeline above. Anyone promising faster than that on the revenue side is selling something.

## Links

Landing page: https://startandship.com
Public revenue ledger: https://realnumbers.online