Getting Found · 5 / 6

Solo Priorities: Write Questions People Actually Search

The first three lessons covered the technical checklist — all one-time jobs. The real daily fight is content: what to write, and what to write first. A one-person company has no SEO team and no keyword-tool budget. Your only weapon is ranking your bets: spend effort on questions people search, trade real experience for trust, and stay off the crooked path.

Topic checkupQuestion-first thinkingLong-tail drillBlack-hat landmines
Write the question first, then add features

Indie hackers' content habit is a "feature lens": changelogs, feature roundups, version comparisons — written diligently, searched by nobody. Users don't search "what's new in XX tool v2.3." They search their own trouble: "special vs regular VAT invoice" or "cat vaccine expired, what now." First rule of topic selection: write the sentence users will actually type into a search box or an AI chat.

A good topic has two dimensions: intent strength (how close is the searcher to using your product) and competition (how many people already answered this well). Strong intent + weak competition is a gold mine. One-person companies dig those. Leave the fat-head keywords to the big players.

Try it · Four candidate topics, check them one by one

Same example: an invoicing tool for freelancers. The site owner listed four candidate topics. Open each one and check it — watch how the intent and competition meters move.

Topic checkup desk Checked 0 / 4
The meters are illustrative; the method is real: for each topic, ask "who would search this, and how close are they to paying"
"How much tax do freelancers pay when invoicing a company"
Intent
Competition
People searching this are stuck in the invoicing flow — one step from using your tool. Few pages answer it seriously; most are tax-bureau boilerplate. Gold-mine topic. Write it first.
"What is an invoice"
Intent
Competition
Search volume looks big, but most people typing this are students doing homework — a world away from paying. Encyclopedias, official sites, and portals already own the answers. Fat-head, high competition, low intent. A one-person company touching this is just pacing the giants.
"Why tax-bureau invoice requests get rejected"
Intent
Competition
The searcher just got rejected and needs a fix now — emotion and intent are both at the peak. Few people write up an unflattering "got rejected" story. If you do, you're scarce supply.
"Invoice Owl v2.3 changelog"
Intent
Competition
Competition is zero because search volume is also zero. Nobody but you searches a tiny tool's version number. Write the changelog (old users and AI engines will read it), but don't expect it to bring new traffic, and don't make it your content mainstay.
After all four checkups, the ranking writes itself: write 1 and 3 first (strong intent, weak competition); skip 2 (you're just pacing the giants); keep 4 with zero expectations. Run every topic list through these two meters. That's a one-person "SEO team."
Click · Question title vs feature title

Same piece of content — the title decides whether it has a door in. Click the one you think is easier to get found.

Two titles, same article A/B duel
The content is identical. Only the title changes.
Option A
petdiary.example.com/blog/reminder
New smart reminder engine: multi-dimensional, configurable, understands you

We rebuilt the reminder system — custom rules by vaccine type and pet age…

Option B
petdiary.example.com/blog/reminder
My cat's vaccine expired a few days ago — can I still catch up? What the vet said, and what I did

Short answer: within two weeks, one catch-up shot usually works; past a month, restart the series…

Try it · Translate one feature into five questions

To switch from feature-speak to question-speak, there's a mechanical drill: take one feature and write five questions a user might type into a search box. Each sentence is a content topic. Below we demo with the invoicing tool's "contract recognition" feature. Click the five empty slots to reveal them.

Feature → question translator Translated 0 / 5
Feature: upload a contract PDF, auto-read amount, tax ID, and invoice type
Click a slot to see which five user questions this feature translates into
1Click to translate sentence 1
2Click to translate sentence 2
3Click to translate sentence 3
4Click to translate sentence 4
5Click to translate sentence 5
See the pattern: none of the five sentences mention your product. They're all the user's own trouble. They come for the trouble, stay because you answered well, and then notice your tool happens to solve it. That's how content and product should lock together. Your turn: take one feature of your product and write five sentences in this format. If you can't, that feature may not be needed at all.
Trade real experience for trust

Topic selection answers "what to write." What's left is "why you." Search engines and AI engines have been tightening the same valve for years: they prefer first-hand experience. Google wrote Experience into its content rater guidelines. When AI engines pick citations, they clearly favor paragraphs with concrete numbers and lived detail — you saw that in last lesson's citation auction.

That's huge news for a one-person company: you have no editorial team, but you have something nobody can copy. Real bills, real error logs, real negotiation notes, real holes you fell into. This site's OPC part used our own trademark rejection as teaching material and a real domain-negotiation quote as a case — those pages are among the most cited. Write what you actually did, with the real numbers from the process. That's a fingerprint AI mass-spam pages can never fake.

Inventory what you may already have. Each item is a piece nobody else can write:

For redaction, follow this site's rule: keep amounts, dates, and process (that's where trust comes from); hide counterpart names, document numbers, and personal information.

One-person company OPC · Rights, told from first-hand casesTrademark rejection, domain bargaining, software-copyright filing — all first-hand records, and a ready demo of "trade real experience for trust."
Write new, or go back and update the old

Once the tracker is running, you'll face a weekly micro-decision: does this week's slot go to a new topic, or to rewriting an old article? The rule is simpler than it sounds:

SituationActionWhy
Old page ranks, but the data is staleUpdate firstAdding fresh data to a page that already ranks is the highest-ROI move of all: the foundation is there, you just add a layer
Old page has no rank and no clicksDiagnose, then decideGo back to lesson 1's funnel and find which layer leaks. If nobody searches the topic, drop it without regret. If the title is the drag, two lines of copy will do
The tracker has a strong-intent, weak-competition candidateWrite newGold-mine topics expire. If someone else answers first, that competition cell changes

Rough rhythm: for every two or three new pieces, go back and refresh one old one. When you update, also update dateModified — lesson 3 covered this. AI engines really do look at that date.

Where AI sits in content production

By this point in the course, you'll probably think: just let AI write the content too. The answer depends on the job. AI is a great assistant and a terrible printing press: use it to speed up your real experience and you double output; use it to mass-produce pages from thin air and you slide into the spam pages on the anti-list below.

StageLet AI do it (assistant seat)Don't let AI do it (press seat)
TopicsTranslate one feature into twenty candidate questions; you pickLet it decide what to write — it will only pick fat-head keywords
DraftTurn your voice notes, logs, and bills into a structured first draftInvent a "deep, quality piece" with no source material
FactsList the numbers and sources you still need to checkLet it fill in the numbers — it will invent them with a straight face
BatchDraft one-sentence summaries for hundreds of pages; you review, then shipGenerate hundreds of standalone pages nobody checked

The line is really one sentence: does the page contain something only you can provide? If yes, AI can help as much as it wants and it's still your content. If no, even if you typed it by hand, it's spam. This site's "in one sentence" blurbs are "batch-draft, then human-review" — hundreds of pages, one person, two days.

Try it · Six tactics: which ones get you killed

Last, the anti-list. The SEO world always has people selling shortcuts. Some work in the short term. The cost is a whole-site penalty once the engine catches on — for a one-person company with one site, that's a wipeout. Judge each of the six: safe, or a trap.

Black-hat risk desk Judged 0 / 6
Pick Safe or Trap for each. Wrong answers still get an explanation.
Spend 500 yuan on a batch of "high-authority backlinks"
Batch-add FAQPage structured data to help-doc Q&A pages
Have AI mass-generate five hundred landing pages by city name
Put a block of white-on-white keywords in the footer
Refresh a two-year-old article with new data, republish, update dateModified
Sign an SEO vendor who "guarantees homepage ranking in three months"
After all six, the pattern is one line: anything that tries to fool the engine is a trap; anything that helps the engine read real content is safe. A one-person company can't afford a whole-site penalty. Don't test that line.

When a hustler shows up, the pitch is always three lines: "internal channel," "guaranteed ranking," "do nothing and competitors will pass you." Hear any one of them and the conversation is over. The money you save will renew every TLD variant for three years.

Ship it: a minimum viable content tracker

Topic method only counts once it lives in a table. No special tool — four spreadsheet columns are enough:

Topic (user's own words)IntentCompetitionStatus
How much tax do freelancers pay when invoicing a companyHighLowPublished 08-02
Why tax-bureau invoice requests get rejectedHighMedWriting
How do I tell if a contract amount is tax-inclusiveMedLowScheduled
What does a red-letter invoice reversal (红冲) meanMedMedCandidate

Two rules: new topics enter as "Candidate"; each week take only one or two from the top down. After a month live, add a column of real data (search impressions, whether AI cited you). Let the next ranking stand on last round's facts. Keep it for a quarter and this table will know your users better than any keyword tool.

Key takeaways

Write the question first, then add features: users search their own trouble, not your changelog. Run every topic through two meters — intent strength and competition — and dig only the strong-intent, weak-competition gold mines.

One feature, five questions: if you can't write five sentences in the user's own words, suspect a fake need. The questions themselves are your content calendar.

Real experience is a fingerprint nobody can copy: bills, errors, negotiation notes, holes you fell into. Engines and AI are both weighting first-hand experience. That's a one-person company's only moat against content mills — and the deepest one.

Black-hat in one sentence: fooling the engine is a trap; helping it read real content is safe. Vendors who promise rankings go straight on the blocklist.

Source: Original to Xiaoshan Academy's Getting Found series. Topic-checkup meters are teaching illustrations. "Experience" in content rating: Google Search Quality Rater Guidelines (E-E-A-T). Scaled-content crackdown: Google's March 2024 core update public notes.