Week 3 · Days 15 to 21. About 7 hours 30 minutes of focused work, free like the rest of the plan.
Week 3 · Days 15 to 21
Three moves that put your pagesinside the fan-out.
Every day below carries its own timing, tools, steps, worked example, pitfalls and completion test; the graded evidence behind the plan lives on the full plan page.
Days 15 to 162 hoursEnds with: A dated retitling batch plus briefs for the gaps.
Read the fan-out queries and align your titles
Fan-out queries are the machine's private vocabulary, and matching it is cheaper than outranking anyone.
Have this open
Your tracker's fan-out queriesOr the day's prompt, which generates plausible ones.
An export of your titles and headingsThe matching exercise needs both lists side by side.
How to do it
Collect the sub-queries your tracker logs for each tracked question; where you have no tracker, generate plausible ones with the day's prompt and treat them as hypotheses.
Cluster them by intent and count repeats: a modifier that appears across many questions, near me, cost, 2026, best, is a title ingredient, not a coincidence.
Retitle in batches of five and note the date; scattershot renaming makes attribution impossible.
Give any orphan sub-query with real repeat volume its own page brief rather than stretching an existing page over it.
Do it with your AI assistantAI prompt
Role: you are a search engine generating fan-out queries.
Context: a buyer asked an AI assistant: [paste one of your category questions]. The market is [category]. My current page titles and headings: [paste them from your site].
Task: 1) write the 15 sub-queries the assistant would most plausibly search before answering: variants, comparisons, modifiers and adjacent questions, phrased the way a search engine expects; 2) match each sub-query against my titles and headings.
Rules: no duplicate sub-queries; for every unmatched one, say whether an existing page should absorb it or a new page is needed, and name the page.
Output: the 15 sub-queries numbered, each followed by the word MATCHED plus the title, or the word GAP plus the recommendation.
Worked example · Marnwell Kitchens, an invented brand
Marnwell's tracking shows the engines repeatedly searching kitchen fitting cost calculator UK, a phrase no Marnwell page contains. The cost guide gains that heading and a worked pricing table beneath it; a fortnight later the guide is the page Copilot cites.
What goes wrong
Retitling every page at once, so no change can be attributed to anything.
Chasing a sub-query that appeared once, which is noise wearing a keyword's clothes.
Stretching one page across unrelated sub-queries until it answers none of them well.
You are done with days 15 to 16 when
Sub-queries are clustered and counted, not just collected.
Five titles or headings are rewritten and the date is noted.
Every orphan query with real volume has a page brief.
Days 17 to 194 hoursEnds with: One published, dated, honest alternatives page.
Build comparison and alternatives pages worth citing
The alternatives page is bottom-of-funnel oxygen: buyers ask it daily, and honesty is its ranking strategy as well as its ethics.
Have this open
Public pricing for six or more rivalsAnything you cannot verify is marked, not asserted.
Your own honest limitsThe section that makes the page citable.
Choose the rival with the most alternatives searches, not your closest competitor; you are borrowing their demand.
Name at least six real options including yourself, each with a genuine strength, a genuine limit and public pricing where it exists.
Write the section your lawyers will query twice: when the rival remains the right choice, and when YOU are the wrong one. Those two sections are why the page gets cited.
Corroborate every claim with something checkable and date the page visibly.
Do it with your AI assistantAI prompt
Role: you are an honest industry analyst drafting a comparison page.
Context: the page is titled [competitor 1] alternatives, for [audience]. It includes [brand] and at least six real tools: [list them].
Task: outline the page. For each tool: who it fits best, one genuine strength, one genuine limit, and public pricing if known. Add a section on when [competitor 1] remains the right choice, and one on when [brand] is the wrong choice.
Rules: every claim must be checkable; mark anything uncertain with the word VERIFY rather than asserting it; no superlatives; no tool wins everything, because a page where one does is a page nothing cites.
Output: the full outline with headings phrased as buyer questions, roughly 60 words of notes per tool.
Worked example · Marnwell Kitchens, an invented brand
Brayford Interiors alternatives lists seven fitters. Marnwell concedes Brayford wins on eight-week lead times and admits Marnwell declines jobs under 6,000 pounds. The concessions are exactly what an assistant quotes when a buyer asks for the honest comparison.
What goes wrong
Winning every row, which is the fastest way to be ignored by both readers and models.
Omitting pricing because it is awkward, when pricing is the most-asked question about any product.
Naming fewer than six options, which halves the citation odds the measured data shows.
You are done with days 17 to 19 when
Six or more real options are named, yours among them.
Both honest sections exist: when they win, and when you are wrong.
Every claim is checkable and the page carries a visible date.
Days 20 to 211 hour 30 minutesEnds with: Three published posts carrying real findings.
Turn your expertise into crawlable LinkedIn text
LinkedIn is where B2B expertise becomes crawlable text under a named person, which is what retrieval and buyers both trust.
Have this open
A named author with a real roleThe company page alone carries less weight.
This week's strongest materialA number or a finding, not a product update.
How to do it
Repurpose the week's strongest material into three posts under a named author with a real role, not the company page alone.
Lead each post with the number: the finding in the first two lines decides whether anyone, human or machine, reads line three.
Post natively rather than linking out first; put the link in the close once the finding has landed.
Do it with your AI assistantAI prompt
Role: you are a LinkedIn ghostwriter for [named person, role at brand].
Context: the audience is [audience]. Source material: [paste the piece or its key points].
Task: write three native posts from it, three different angles: one contrarian take, one how-we-did-it, one surprising data point.
Rules: a specific number or first-hand finding in the first two lines of each; 120 to 200 words per post; plain text; at most three hashtags; end each on a genuine question; mention [brand] at most once per post; no engagement-bait phrasing and no emoji walls.
Output: the three posts, each under its angle as a one-word label.
Worked example · Marnwell Kitchens, an invented brand
Marnwell's fitter writes: we measured 40 kitchen quotes across Leeds this spring; the spread between highest and lowest was 2.4x for the same room. The post outperforms every brochure the company ever ran, and the figure starts appearing in answers about kitchen quote ranges.
What goes wrong
Posting from the company page only, which reads as an advert and travels as one.
Opening with context so the finding arrives after the fold.
Leading with a link, which costs reach and buries the point.
You are done with days 20 to 21 when
Three posts are written under a named person with a real role.
Each opens with a specific number or first-hand finding.
All three are published or scheduled.
What this week leaves youYou end week 3 speaking the retrieval layer's own language, with comparison pages built to be quoted.
The workbook
The workbook turnsreading into doing.
A printable day-by-day checklist, the 48-prompt starter pack organised by buyer intent, the week 4 outreach template and the baseline tracking sheet. Email one carries the workbook link; seven more follow, paced to the plan’s weeks across thirty days, then the newsletter’s occasional notes.
Questions
Asked about this week, answered straight.
Including the ones where the honest answer costs a sale.
How do I see fan-out queries without a tool?
You approximate. The day 15 prompt generates the likeliest sub-queries for any question, and Google's own related searches plus autocomplete confirm which phrasings are real. A tracker replaces the guesswork with the queries actually logged during answer runs, which is the difference between a hypothesis and a reading, but the approximation is enough to start retitling.
Will naming competitors on my site just advertise them?
The buyer reading an alternatives page already knows the competitor; that is why they searched. What the page decides is whose framing the comparison happens in, and measured citation data says entity-rich pages naming six or more options take nearly double the citations of pages naming none. The risk of the page is not advertising rivals; it is dishonesty, which stops it being cited at all.
How many comparison pages should I build in week 3?
One, properly. A single honest, corroborated, dated alternatives page against your most-searched rival beats four thin ones, and it becomes the template the later pages copy. Add the next rival's page in the weekly rhythm after day 30, one at a time, each earning its own research.
Every step of this week, with the graded evidence behind it, sits in the full 30-day plan.
Start here
See where you show up in AI answers today.
Add your site and RankX AI suggests the prompts to track, monitors the keywords that matter and audits your pages, with your first results minutes after you finish setup.