You Built It. Why Isn't Anyone Coming?

More stuff. Same doors.

Vibe Coding drove the cost of shipping down to almost nothing. The less-mentioned flip side: if you can ship a product in ten minutes, so can everyone else. What gets built is multiplying in plain sight. The doors to user attention are still the same few. Competition didn't vanish — it just moved from "can you build it" to "will anyone see it."

A lot of indie hackers follow this path: grind for a month, ship, post it on WeChat Moments (or Twitter), get thirty visitors, then zero a week later. Conclusion: "the product is bad." Most of the time that verdict is early. The product may be fine. The problem is it never entered any discovery channel at all.

Shipping is only the start. Being found is what counts as launch. A product nobody visits and a product that was never built look the same to the market.

Try it · Three doors, flip them one by one

How users find a new product almost never leaves these three paths. Each card is one path. Open it and read its temperament: who drives it, how long until it works, how much you can control.

Three roads traffic takes Opened 0 / 3

Click a card to open it. Finish all three before you move on

Search engines

Users show up with a question. Open it and read its temperament.

Who drives itThe user. They type a question. Your page has to be the answer

Time to workSlow. Indexing takes weeks, ranking takes months — but once it lands, people keep coming

What you controlContent and structure are entirely in your hands. Of the three, this is the one you control most

AI engines

Users stopped searching. They just ask. Whether you're in the answer is the AI's call.

Who drives itThe AI. It reads the web for the user, then decides whom to cite. Users see an answer. What you're fighting for is one citation inside it

Time to workDepends on whether you made it into its index and retrieval sources. Once you're in, you can get cited

What you controlSome of it. You can improve how you write (all of Lesson 3). Whether it cites you is still its decision

Social spread

Someone else talks about you. It arrives fast. It leaves fast.

Who drives itThe platform and the people who share. Whether a post takes off — you can influence it, you can't decide it

Time to workFast. You'll see the curve the same day. It usually falls back within a week

What you controlThe least. It's a spark, not a foundation

All three open — the division of labor is clear. Social spread lights the fuse. Search engines are the long game. AI engines are the new door still growing up. The rest of this Part only covers the last two, because they share one thing: slow to work, but you can control them, and doing it right once lasts. How social spread becomes word-of-mouth was already taught in the Lei Jun course. We won't reteach it here.

The true nature of word-of-mouth: exceeding expectations · Lei Jun's Entrepreneurship CourseThe method for the social-spread door lives in that lesson: managing expectations, and care that users can actually feel.

Try it · The indexing funnel: which layer are you leaking?

"Nobody's coming" is a blurry complaint. Split it into five layers: you built it, it can be crawled, it got indexed, it got shown, it got clicked. Each layer leaks a batch. The four switches below map to four common diseases. Try turning them off, then on, and watch the numbers on each layer change.

Indexing funnel simulator Four switches

Switches start all off — that's the real state of a lot of new sites at launch. Flip them on one by one and watch which layer gets fixed

Body text visible on the serverThe body is already in the page source. Crawlers can read it without running JS. The favorite mistake of AI-generated frontends: render everything in JS

Submitted sitemap and webmaster toolsTell the engines "I'm here, and here are my pages." Indexing stops being luck and becomes something you can measure

Every page has a human title and descriptionThose two lines are what users see on the results page. Write them like a person or nobody clicks

The content answers a question people searchHowever many pages you have — if nobody searches the question they answer, you'll never get shown

Bar values are illustrative indexes — they show relative loss at each layer, not real stats.

After playing it you should see a pattern: the four switches each fix a different layer. None of them can stand in for another. Visible body text fixes "can be crawled." sitemap fixes "got indexed." title fixes "got clicked." Topic choice fixes "got shown." Before you complain about traffic, find which layer you're leaking. Later lessons prescribe by layer.

Try it · Three minutes: is your product even there?

No tools to install. Three moves give your site a rough diagnosis. If you have a product now, do it as you read. If not, try it on some small tool site you use. Tap an item when you've done it.

3-minute rough diagnosis Done 0 / 3

Tap each item when you're done, and see what that step is diagnosing

✓ Type site:your-domain in the search box Do it once on Google and once on Bing. The result count is your index count This step diagnoses "are you indexed?" Zero on both sides means the engines don't know you exist — go to Lesson 2 and submit. Google has you, Bing doesn't? Lesson 4 explains why that's a big deal.

✓ Ask an AI a problem your product can solve Use your target user's own words. Ask ChatGPT or Kimi. See if you're in the answer — and if your competitors are This step diagnoses "do AI engines see you?" If the answer is all competitors, someone is already harvesting this door and you're not at the table. If nobody is in the answer, the question is still open — that's an opportunity.

✓ Search your own brand name Add "what is" or "review" after the product name. Is page one you — or someone else introducing you? This step diagnoses "are you holding your brand terms?" If you can't find yourself by your own name, users who heard of you still can't reach you. That's the unluckiest break in the discovery chain.

All three done — you already have a rough diagnosis. Write down the three results. Come back after this Part and check again. The change is your report card.

It's 2026. Why still learn SEO?

A fair objection: users went off to ask AI. Is search-engine optimization a dead craft? Split it open and the opposite is true. First, search isn't dead: a lot of "find this site" demand still goes through the search box, especially with transactional intent (prices, tools, comparing services). Second, and more important: before an AI engine answers, it has to "search" too. Its retrieval sources are search-engine indexes and the content on your site. The SEO basics — crawlable, readable — are exactly the ticket to being cited by AI.

What you do	What it does for search engines	What it does for AI engines
Body text in the HTML	Can be indexed	Can be read — most AI crawlers don't run JS
A human title	People click in the results	The first clue an AI uses to judge what question this page answers
Submit a sitemap	Indexing gets faster and more complete	You're only in an AI's retrieval sources if you're in the index

That's why this Part teaches SEO and GEO together: not two courses, two exits from the same set of basics. Lesson 2 is the shared foundation. Lesson 3 is the AI-specific part.

Pick one · A new site, three months, no traffic

An AI-built tool site has been live for three months. Search traffic is basically zero. What should you check first? Single choice

Wrong answers come with an explanation. Keep going until you hit the right one

ANot enough features yet — add more first

BCheck the index count with site:, then see if the page source has body text

CBuy a burst of ads and push the volume up first

DMake the homepage flashier — add animation and video

Map of this Part

The next five lessons walk the funnel from top to bottom. Each one closes one layer:

Lesson 2 fixes "can be crawled, got indexed": the SEO minimum viable checklist. All ten items are one-time actions.
Lesson 3 fixes "can be cited by AI": GEO's four levers — answer blocks, FAQ, llms.txt, dates.
Lesson 4 is a real operation: the full audit and rebuild of this site in August 2026.
Lesson 5 is the long fight: topic ranking, trading real experience for trust, and black-hat landmines to avoid.
Lesson 6 closes it: a 12-item acceptance list plus four measurement definitions. Progress lives on your machine.

Key Takeaways

✓

Shipping is not the same as being found: once the cost of execution collapsed, more stuff got built. The doors to being found didn't multiply. A product nobody visits and a product that was never built look the same to the market.

✓

Three doors, three temperaments: social spread lights the fuse but you control it least; search engines are slow but you call the shots; AI engines are the new door still growing up. This Part only covers the last two.

✓

"No traffic" has to be split into five layers: you built it, it can be crawled, got indexed, got shown, got clicked. Each layer has a different disease and a different medicine. Locate the leak first, then treat it.

✓

The 3-minute rough diagnosis is always available: site: for the index count, ask an AI once, search your own brand. Write down today's results. Come back after the Part and compare.

✓

SEO and GEO are two exits from the same set of basics: AI engines retrieve before they answer. Crawlable and readable homework works on both sides.

Source: Original to Xiaoshan Academy's "Getting Found" series. Methods and cases come from this site's real SEO / GEO rebuild in August 2026 (the full recap is Lesson 4).