Chapter 1 · Pick Your Game: Pick Wrong and All the Work After Is Wasted

Where you are, and what this chapter solves

You want to earn traffic money from a game guide site. The whole thing is like opening a snack shop: first decide what snack you sell, then decorate the store.

Most beginners build the site right away — that’s like opening the shop only to discover nobody on this street eats what you sell. Game guide traffic follows a rule: after a new game takes off, the 2 to 8 weeks are the golden window. That month-plus of searches can be 60% to 80% of the game’s lifetime search volume. Pick the right window and even ordinary content ranks; pick the wrong one and nobody reads you no matter how well you write.

So lesson one is not building a site. It’s picking the product. Judge two things:

Will this game take off? (does anyone search for it)
When it does, are there still open seats on page one of Google’s results? (is the competition crowded)

Two iron rules so you don’t pick forever: a score of 60 or above only advances a candidate to demand and competition validation; it does not authorize a build. A crowded SERP is a hard veto. From starting to deciding, spend at most 2 days. Picking a game is racing the clock, not finding excuses for a high-scoring candidate.

What you’ll have when this chapter is done

One clear decision: this game — build or drop
A “first 8–12 pages” list containing only pages that really exist for this game and have evidence, in priority order

A few words to know (just for this chapter)

Search volume: how many people search a term on Google each day. More people = more potential customers.
SEO: the whole toolkit for getting Google to rank your pages higher. You’ll be using it throughout this book.
SERP: the Google search results page itself. The top 10 spots on page one are the golden storefronts — most clicks land there.

Gate 1: List your candidates (30 minutes)

Find new games from the places below; gather 10 to 20 candidates before moving on. Pick 3 to 5 from each:

Where to look	What to watch
The newest page on itch.io (pick Play in browser)	New games every day, the least competition
Steam’s new-release and wishlist charts	High game quality, but more people watching too
Roblox’s Discover page (switch to the Rising / Up-and-Coming sort)	The densest crowd of players searching “codes”
Recent breakouts on YouTube gaming channels	Over 50K views in 7 days means people really watch

Candidate pool too small? The repo doc docs/sourcing.md is the full playbook: 9 sourcing channels (how to read SteamDB’s trending charts, spotting outlier videos from small creators on social media, sweeping game-aggregator sitemaps…), plus a decision log table — game name | build / skip / maybe | launch date | site link. Anything you can’t call goes down as “maybe”, and gets a second look the next day. Twenty minutes of channel-surfing a day yields a steady 3 to 5 new candidates: sourcing becomes a pipeline, not luck.

Gate 2: Score every candidate (5 minutes each)

Four scoring items, 100 points max, 60 or above advances to the next gate:

Is there enough to write (hard gate): does the game have enough mechanics, items, and bosses to write guides for? Pure match-three mini-games have no guide demand — eliminate outright.
Is there popularity evidence: comment counts, subscriber counts, concurrent players; is the author still updating (still updating = a longer window)?
Will players search: when players want a guide, will they search “game name + wiki / guide / codes”?
Is English content missing: few English guides is both popularity evidence and your opportunity.

Gate 3: Confirm people really search (the most important gate)

Open Google Trends, click “Compare”, and enter two terms: your game name codes and a baseline term whose heat you roughly know (don’t know one? Use roblox codes — it’s a perennial big term).

Read the curve: your game’s line must be going up to be worth building. The best move is to copy the curve data and feed it to your AI for judgment (prompt at the end of this chapter).

Do not treat a single 100 as proof. Google says Trends is sampled and normalized relative interest: low-volume terms may show 0, and isolated spikes may be statistical noise. Check both past 7 days and past 30 days in the target market against a like-for-like benchmark. Insufficient data, a single spike, or mostly-zero observations means drop/watch for now, not a test site. Compare search terms with search terms and topics with topics. See Google’s Trends data FAQ and term/topic comparison guide.

A common trap: huge YouTube views but no Trends searches — viewers watch on YouTube and leave; they never come searching for guides.

Two-source rule: popularity evidence must come from at least two unrelated places (e.g. Trends + YouTube). If only one source is hot, watch it for a week before deciding.

Gate 4: Check how crowded Google’s results page is (5–10 minutes, hard veto)

Google game name wiki and game name codes, and read page one:

What you see	Verdict
Fewer than 10 results, or mostly PDFs, forum threads, YouTube videos	Open seats — build now
A Fandom (an old wiki site) exists, but the content is thin and updates are slow	Winnable — take its spot with fuller content
Fandom or Game8 (a pro gaming site), thick content and daily updates	Drop it, next candidate

Do not merely count domains. Audit at least two intents (wiki plus codes or guide) and record at least five page-one results one by one: result type, content depth, update cadence, and intent fit. Any one of these means “too crowded”:

Five or more dedicated sites on page one;
Three or more major guide publishers on page one;
Four or more audited results with thick content and a correct intent match;
Any thick, correctly matched strong competitor that updates daily or frequently.

One more signal that’s easy to miss: a full page one ≠ no room. But an intent gap cannot be asserted by feel. It overrides the competition veto only when at least five results were audited and at least three results, representing 60% or more of the audit, clearly answer the wrong intent. Writing only intent gap: true without result URLs and judgments counts as no gap. How to check, plus a real-world case, is in the “intent fit” section of docs/sourcing.md.

One common trap: Google Keyword Planner’s Competition measures advertiser competition, not organic ranking difficulty; generic third-party KD is also an estimate. Neither can override a result-by-result SERP audit. See Google’s official metric definition.

Operational hard gates (eliminate first, score second)

All four must pass. A high raw score cannot override a failure:

Search demand: both 7-day and 30-day Trends windows were checked; data is sufficient and not a one-off spike; the recent segment is at least 15% above the preceding segment; at least half of 7-day observations are non-zero.
Independent two-source proof: one source is Trends and the other is platform, sustained YouTube creator, or community behavior. Google Suggest, related queries, and Google Search are not independent of each other.
Competitive opening: at least two query intents and five page-one results were audited. Too much competition without a reproducible intent gap is an immediate drop.
Content depth: at least eight independent useful pages, with at least four backed by autocomplete, related-query, SERP-gap, or community-question evidence. Inferred pages cannot pad the gate.

Only after all four pass may the score distinguish BUILD from a 30% TEST. TEST is not a consolation prize for a failed hard gate.

Let AI run the four gates for you (recommended)

Organize the materials below and feed them to an AI (ZCode / Claude Code / Codex — any one; the ChatGPT web app works too). Replace everything inside <> with yours, then copy and send the whole block:

You are a game wiki selection analyst. Evaluate through the four-layer funnel: discovery → scoring → demand validation → competition validation.
**Input:**
Game name: <game name>
Google Trends data: <paste the comparison of "{game name} codes/wiki" vs your baseline term>
Top 10 Google results: <paste the result lists for "{game name} wiki" and "{game name} codes">
Other signals: <YouTube views / Roblox concurrent players / community discussion>
**Task:**
1. Scoring table (0-100, ≥60 passes): wiki content depth (hard gate, eliminate outright if insufficient) / social proof / title searchability / English content
2. Demand hard gate: audit US 7-day and 30-day windows; exclude insufficient data, low-volume zeros, and one-off spikes; verify Trends plus a non-Google independent source
3. SERP hard gate: cover at least two query intents and audit at least five results one by one; list URL/type/depth/update cadence/intent fit. 5+ dedicated sites, 3+ major publishers, 4+ thick matching results, or one thick frequently updated strong site = too crowded
4. Intent-gap exception: valid only when at least five results were audited and at least three, representing 60% or more, clearly mismatch intent
5. Content hard gate: at least eight independent pages and at least four with search or community evidence
6. Verdict: list PASS/FAIL for every hard gate first; any FAIL must be drop/watch and cannot be overridden by total score
Analysis only — do not write any files.

Note the last line, “Analysis only — do not write any files” — at the selection stage the AI only gives advice; it never touches your repo.

Produce your “first 8–12 pages” list

Once you decide “build”, immediately have the AI turn the verdict into a day-one writing plan:

Based on the selection verdict, plan the first-day 8-12 pages. Game name: <game name>; verified search terms: <list>; systems that actually exist in the game: <list>; unanswered community questions: <list>.
Choose only from verified, real intent clusters: wiki/guide, beginner/progression, how-to/quest, item/weapon/build, boss/map/location, codes, tier/value/trade, calculator/tool.
Create codes only when the game has a redemption system and codes can be verified. Create tier/value/drop-rate pages only when versioned data and ongoing maintenance are available. Never invent bosses, trading, or tool demand to pad the launch.
Output a table: priority / page / slug / category / suggested title / target keyword / evidence type / fact source / update cadence / materials I need to provide.
Mark pages lacking data as "waiting on my input" and ask me with a clear list. Plan only — do not produce pages.

Do not force a fixed ten-page recipe onto every game. A first batch usually includes one overview/navigation page, real beginner blockers, and specific how-to questions already visible in search or the community, followed by item, quest, map, or tool pages only when the game supports them. A new game’s weapons, characters, and mission names may have no third-party volume yet, so publishing first can create a window; bigger publishers may later displace it. Each page therefore still needs reproducible facts, screenshots/testing, or tool value rather than relying on being early.

If you get stuck

“I don’t know which game has potential”: go back to Gate 1 and spend 10 minutes on each of the four sources — quantity first, quality later.
“Two games both score above 60 — which one?”: pick the one with the emptier Google results page. An open SERP is worth more than heat.
“I can’t read Trends”: look at one thing only — is your game’s line above or below the baseline, and is it rising or falling. Rising = people search.
“I still can’t decide”: the default move is to drop it and look at the next candidate. Remember the two-day limit.

✅ Acceptance criteria (all must hold)

You can say it off the top of your head: which game, why (which two pieces of popularity evidence), what the competition looks like (the shape of the results page)
The 7-day and 30-day Trends evidence is complete and is neither insufficient data nor a one-off spike
At least two search intents and five page-one results were audited; any claimed intent gap meets the three-result and 60% evidence threshold
You hold an 8–12-row writing list, each row with a title, target query, evidence type, and fact source
☐ From starting selection to making the decision took no more than 2 days

Next step

The game is chosen and you still have nothing — the next chapter spends 30 minutes installing 6 tools (a one-time install, yours forever), and the chapter after that gets a working site onto your computer. Go to Chapter 2 · Before You Set Off: Install the 6 Tools