Launch week is a spike

Launch week feels like the work. You pick a Tuesday. You write a thread. You ship a Product Hunt post, a Betalist card, a handful of "we launched" emails. The graph goes up. Then the graph goes down. The comments freeze. The URL you fought for is still live, but nobody has a reason to open it again.

That is not a moral failure. It is how spikes work. A spike is a moment of attention. A listing is a place. If you want software to stay findable, you need both, and you need to stop treating the spike as the whole plan.

This is a long look at why launch-week directories forget you, what search and answer engines actually keep, and how to leave a public record that still makes sense six months later. MakeItLast is one of those records. The advice holds even if you never list here.

What a spike is good for

A spike is good at one job: telling people who already watch launches that you exist today.

Product Hunt, Hacker News, a well-timed tweet, a founder Slack, a niche newsletter. These surfaces are timed. They rank by recency, upvotes, or the mood of a room. They are honest about that. The front page is a front page. It is not an archive you should expect to carry you.

Use a spike for:

• A first hundred visits from people who like new things.
• A sentence you can reuse later ("We launched on Tuesday, 800 people tried the demo").
• Bug reports from strangers who will never become customers, but will tell you the onboarding is confusing.
• A screenshot of social proof you can put on a landing page.

Do not use a spike for:

• Your only inbound channel.
• Your only public description of the product.
• A substitute for a page that names the product, the category, and the URL in plain language.

If the launch is the only public artifact, you are renting attention. When the rent is due again, you pay with another launch.

Why the thread dies and the listing should not

A launch thread dies because the audience moved. That is the design. The next post is someone else's. Your comments are still there, but they are not a query result. Nobody types your category into Product Hunt six months later and expects a stable, complete answer.

Search and answer engines do something different. They look for a page that still matches a question:

• What is this product?
• Who is it for?
• Is it real software, or a parked page?
• Where is the official site?
• What else is like it?

A directory listing that stays public, uses the product's real name, and does not vanish when the launch week ends is a better answer to those questions than a frozen thread.

That is the whole idea behind MakeItLast. The catalog is not a launch week. It is a list of software built to stick around. You submit once. If review passes, the card stays up. Search can keep reading it. So can an answer engine that is willing to fetch llms.txt or a markdown twin.

The three places a product description lives

Most makers accidentally keep three different descriptions, and none of them agree.

The homepage. This is the source of truth. It should say the name, one sentence, and what a person can do in the first minute. If the homepage is a vibe, a reel, or a waitlist with no product, directories and crawlers have nothing to quote.

The launch copy. This is written for a room. It is punchy. It has a joke. It assumes the reader already knows the category. It goes stale the moment you change pricing.

The directory card. This should be boring on purpose. Name. Tagline. A few short paragraphs. Category. Tags. Official URL. That is the version Google and ChatGPT can cite without inventing a sentence you will hate.

When those three drift, you get a product that "sounds different" everywhere. Crawlers do not merge your vibe. They pick one snippet and repeat it.

The practical rule: write the directory card as if a stranger will paste it into an answer. Then keep the homepage close to that card. Launch copy can be louder. It should not contradict the card.

What actually gets indexed

People talk about "getting in front of Google" as if a directory submission were a vote. It is not a vote. It is a URL with text.

A useful listing page has:

• A stable URL that does not change when you edit the tagline.
• The product name in the title and the first paragraph.
• A description long enough to be a real answer, not a slogan.
• Structured data that repeats the same facts.
• An outbound link to the official site, marked clearly.
• A machine-readable twin, if you care about answer engines.

MakeItLast listings are public pages under /tools/{slug}. Each one has a markdown twin at /tools/{slug}.md. The catalog is published as llms.txt and llms-full.txt. The point is not that markdown is magic. The point is that a crawler does not have to execute your React homepage to know you exist.

If your own marketing site is a client-only canvas with no title tag, a directory page can be the first thing that is actually readable.

That is an uncomfortable sentence for designers. It is still true.

Launch directories vs lasting catalogs

A launch directory is a calendar. A lasting catalog is a filing cabinet.

Launch directories optimize for a day:

• Ranking by votes in a 24-hour window.
• A badge that says "Product of the Day" and then becomes trivia.
• Comments that are social, not factual.
• A profile that is hard to update without launching again.

Lasting catalogs optimize for a year:

• A page that still matches "markdown notes app" in November.
• Categories and tags a human can browse without a clock.
• A way to correct the name when you rebrand.
• Review that blocks junk, not taste.

You can use both. Plenty of good tools do. The mistake is stopping at the calendar.

If you only do Product Hunt, you have a spike and a URL that will be outranked by the next spike. If you only do a catalog, you may grow slower in week one. You will still have a sentence in public when a reporter, a model, or a customer searches the category in March.

A 30-day plan after the launch

Day 0 is the spike. Days 1 through 30 are where most products go quiet. Use the quiet.

Day 1. Write down what people actually said. Not the compliments. The confusion. "I thought this was a CRM." "Does this replace Notion?" Those sentences are your tagline candidates.

Day 2. Put a plain-language description on the homepage. Name, one sentence, what happens after signup. If you cannot describe it without a metaphor, you are not ready for a directory card either.

Day 3. Submit a lasting listing. On MakeItLast that means get listed, paste the URL, edit the draft, save. Review runs in the same request. If it passes, you are live.

Day 4. Pick the outbound link. Free dofollow if you put the badge on the homepage. A one-time paid dofollow if you do not want a badge. Nofollow if you just want to be in the catalog. Do not stall here. A nofollow listing that exists beats a dofollow you never finish.

Week 2. Add the missing facts. Pricing model in a sentence. Platforms. Who it is not for. Directory copy that tries to be for everyone reads like no one.

Week 3. Cite the listing from somewhere you already own. Site footer. Docs. A "Press" or "Also listed on" row. You are not building a link scheme. You are making it obvious which public page is the canonical description.

Week 4. Search your own name and your category. If the launch thread outranks the listing, that is fine for now. If nothing but the homepage appears, the listing may not have been crawled yet, or the description is too thin to be a result. Thickening the card is cheaper than another launch.

What "software that lasts" actually means

It does not mean the product will exist forever. Nothing in software promises that.

It means the public record is built as if the product might still matter next year:

• The name is the name customers use, not a codename.
• The URL is the current site, not a expired domain.
• The description is still true after a pricing change, or you update it.
• The page does not depend on a vote total from a Tuesday.

MakeItLast reviews for adult content, spam, malware, phishing, gambling, scams, and parked domains. It does not review for taste, novelty, or whether a stranger on Twitter would upvote you. Agencies and studios are allowed. That is deliberate. A lot of "directories of startups" quietly punish service companies. People still search for them.

If you are building a custom software studio, you still need a public sentence. A listing is that sentence.

The homepage test

Before you submit anywhere, open the site in a private window, on a phone, with images blocked. Read only the text.

You should be able to answer:

1. What is the product called?
2. What does a user do in the first session?
3. Is there a way to start, or only a way to watch?

If you fail that test, a directory will fail it too. AI fill can draft a card from the page, but it cannot invent a product that is not on the page. Garbage in, a polite hallucination out, then a reject or a listing you will have to rewrite.

Makers sometimes ask why fill "did not get" the product. Usually the page never said it. A hero video is not a description. A three-word slogan is not a description. "The OS for X" is not a description unless you also say what the person clicks.

Write the sentence on the site. Then let a directory copy it.

How answer engines use a catalog

Answer engines do not owe you a citation. When they do cite, they prefer pages that look like answers:

• Stable URL.
• Clear entity name.
• Surrounding context (category, tags, similar tools).
• A format they can fetch without a browser farm.

That is why MakeItLast publishes markdown twins and a catalog index. A model that is allowed to read the web can fetch llms.txt, pick a tool, and open /tools/{slug}.md without guessing your CSS.

Will that get you into ChatGPT tomorrow? Nobody honest will promise that. What you can control is whether the facts exist in a form that does not require JavaScript. Launch threads are HTML comments. Catalog twins are documents.

If you care about AI search, do the boring work:

• Keep the official URL in one place.
• Keep the name consistent.
• Keep a public paragraph that is not a slogan.
• Give crawlers a text twin.

Then go back to building the product. The document is not the business. It is the map to the business.

Mistakes that waste a listing

Launching the directory on the same day as Product Hunt and then ignoring it. The listing is the long copy. Write it when you still remember what the product is. Do not write it six months later from memory.

Stuffing every keyword into the tagline. "AI-powered next-gen platform for teams and solopreneurs." Nobody talks like that. Crawlers see a blob. Humans bounce.

Using a different name on every site. The GitHub repo is one string. The domain is another. The Hunt post is a third. Pick the customer-facing name and repeat it.

Leaving a parked page live. "Coming soon" with a Mailchimp form is not a product listing. Automated review will often reject it, and it should. List the live site.

Buying ten spam directories the same afternoon. You will get ten thin pages that all say the same three words. Some will be toxic. A single lasting catalog page is cleaner.

Treating dofollow as the product. A followable link is a nice extra. It is not a growth engine by itself. If the listing is thin, the link is a thin link.

A worked example

Imagine you ship a local-first notes app. Launch day: 1,200 visits, 80 signups, 12 angry tweets about import. Homepage headline: "Notes, finally." That is a vibe. It is not a listing.

Rewrite the homepage to: "Papercut is a local-first notes app for people who write in Markdown and sync with a folder. No account required on the desktop app."

Now a directory card can say the same thing, plus category (notes), tags (markdown, local-first, desktop), and the URL. Six months later someone searches "local first markdown notes." The launch thread is gone from every feed they follow. The listing can still be a result. So can the homepage, if you kept the sentence.

If you later add iOS, you update the listing. You do not need a second launch to change a tag.

That is the difference between a spike and a place.

What to do this week

If you launched in the last year and went quiet:

1. Open your homepage and write one boring paragraph.
2. Submit the URL to a lasting catalog. On MakeItLast, AI can draft the card. You still edit it.
3. Choose badge, paid dofollow, or nofollow. Do not wait for a perfect design.
4. Add the listing URL to your footer or docs.
5. Stop planning the next launch until the public sentence is true.

If you have not launched yet:

1. Do not wait for Product Hunt to write the sentence. Write it first.
2. Put it on the site.
3. List it the same week you launch, not as a consolation prize after the spike dies.

Short answers

Is Product Hunt a waste? No. It is a spike. Use it like a spike.

Will a listing replace SEO? No. It is one public page with your facts. Your own site still has to work.

Do I need a badge? Only if you want the free dofollow on MakeItLast. The listing can exist without it.

What if someone else listed my site? Claim it. Prove you control the URL, take the card, then pick the link type.

What if review rejects me? Read the reason. Parked domains, adult content, spam, and scams are out. Software, agencies, and studios are in. Fix the site or the copy and submit again.

Launch week is a Tuesday. A listing is still there on Wednesday. Build for Wednesday.

A live listing also grows review, pricing, alternatives, and compare URLs when neighbors exist. If Search Console parks those URLs under crawled, currently not indexed, read Crawled, currently not indexed. If you want the pair rules, read software alternatives without doorway pages.