X Search method: Advanced search that removes noise and picks up only “potential topics”

What you can learn from this article (results)

• How to change X search method to the purpose of “extracting winning themes”
• X Search Tips: Filters to reduce old/biased/noise
• How to use X search operator using "type"
• How to pick up “context” that makes it easier to reproduce Posts that increase with X search
• 30 minutes a week search routine and transcription template

Searching for X will become stronger if it becomes a process of retrieving popular posts in a reproducible form, rather than just looking for material. Here, we will create a one-stop process from "Search → Winning Theme → Next Idea → Draft" with a level of detail that even beginners can use.

First of all, conclusion: The purpose of search is not to “gather stories” but to “extract winning themes”

Conclusion: The goal of XSearch is not to generate more ideas, but to extract “winning themes (types that are easy to reproduce).” Reason: If you only collect ideas, you have to worry about starting from scratch every time, but it is easier to automatically progress from the next idea to the draft of a winning theme. Procedure: First, divide the “search purpose” into two, and from then on only adopt the movements on the winning theme side.

• Gathering stories: Seems interesting/popular, more notes (but it's hard to post)
• Extraction of winning themes: The basis for growth (what was solved by whom) remains as a set (easily posted)
• Goal form: Extraction of winning themes → Next draft → Draft

Misconceptions that beginners tend to make (common)

• Thinking that ``search = collecting random ideas'' ends up only increasing the number of saves.
• “Imitate the buzz” and get the context, target, and concerns wrong.
• No reproducibility remains because the reasons for growth are not verbalized

Correct goal (shortest route)

• Extraction of winning themes: Whose problems were solved and in what form?
• Next proposal: A state where you can write about the same problem from a different angle, specific example, or procedure.
• Draft: Hook (beginning) → Body → CTA is filled in all the way

If your search is stuck on "Measures to avoid running out of material", it will be easier to organize if you also read this first: How to eliminate running out of material

Figure 1: Difference in search purpose (gathering stories vs. extracting winning themes)

flowchart LR
subgraph A [collection of stories]
A1[Search] --> A2[Save interesting]
A2 --> A3 [Add more notes]
A3 --> A4 [Can't post]
  end

subgraph B [winning theme extraction]
B1[Search] --> B2[Identify growing posts]
B2 --> B3 [Extract whose problems and solution patterns]
B3 --> B4 [Next draft template]
B4 --> B5[draft]
  end

How to read the diagram: On the left is the flow of save increases'', and on the right is the flow of proceed to post''. The turning point is whether you can verbalize “whose problems” and “solution patterns” from the posts that have grown. From now on, the trick is to run only the process on the right side.

3 things that get you hooked on search (old/biased/noise)

Conclusion: The reason why X search doesn't work is usually one of (or a combination of) old'', biased'', and ``noise''. Reason: If the search results are out of sync, it will be more likely to be based on “feelings” rather than winning themes, slowing down decision-making. Procedure: Inspect each of the three as "Symptoms → Causes → Countermeasures" and reflect them in the search conditions (period, reaction, context).

1) Old: out of touch with the current context

Bottom line: Outdated search results can make it difficult to get responses even for the same topic. Reason: If the premise of the topic (function, fashion, environment) changes, the key points will also change easily. Steps: Set a time limit and prioritize "recent phrases" and "new complaints" even on the same theme.

• Symptom: Most of the posted examples are from several years ago/The premise is different when looking at them now.
• Cause: Period conditions not included/search word too broad
• countermeasure:
• Search again in the most recent time period (e.g. days to weeks)
• Separate by before and after events (release, financial results, season, campaign)
• Re-pick up "current troubles" under the same theme

2) Biased: Seeing only the same opinions

Bottom line: Bias is a sign that you are becoming an “extension of your own timeline.” Reason: If you only follow people from the same neighborhood and the same people, the winning theme will tend to lean toward “internal common sense.” Steps: We intentionally mix in posts from the opposite side (different occupations, different positions, different scales) to make the points at issue three-dimensional.

• Symptoms: Too many posts with the same conclusion/no novelty
• Cause: Only vocabulary related to following is used/search source is narrow
• countermeasure:
• Change position (e.g. operator → manager → user)
• Skipping across industries (e.g. B2B → Recruitment → CS)
• Search by replacing keywords with synonyms

3) Noise: Information that is detoured the more you pick it up

Conclusion: Noise is information that interferes with extracting winning themes, such as "irrelevant, low quality, controversial, and extremely biased" information. Reason: Even if the reaction is great, something that is difficult to reproduce (or has a high risk) is unlikely to become an investment asset. Procedure: Decide on the definition of noise first, and create “exclusion rules” from the search results.

• Symptoms: Too many radical posts/a lot of incitement/conclusions with little evidence
• Cause: Misunderstanding that "stretched = usable" / No filter
• Countermeasures (example of exclusion rules):
• Posts where it is unclear whose concerns are being held will be suspended.
• Lower the priority if there is no evidence (no experience, data, procedures)
• Things that are growing based on the premise of a flare-up or conflict should be handled carefully by comparing them with the operational purpose.

Figure 3: Noise removal filter (old/biased/noise) and countermeasure map

flowchart TD
S[Check first when looking at search results] --> O{What type of discrepancy? }

O -->|old| A[cut by period]
A --> A1 [Research most recently/before and after event]
A --> A2 [Prioritize new complaints/phrases]

O -->|Bias| B[Increase perspective]
B --> B1 [Search by vocabulary of other positions/jobs]
B --> B2 [Add opposing opinions]

O -->|Noise| C[Determine exclusion rules]
C --> C1 [irrelevant/low quality/flaming/extreme pending]
C --> C2 [Leave a post where you can see whose problems and solutions]

How to read the diagram: When the search results are questionable, first determine whether they are "old, biased, or noisy." Once you have determined, just rearrange the search conditions according to the countermeasures on the right. “Filtering before loading the contents” will save time.

Practical procedure: Pick up the “context” from the posted post as a starting point (reproducibility route)

Conclusion: If you want to increase the number of posts that grow, the order of X search is ** ``Posts that grow → context → analysis → next idea''**. Reason: Rather than imitating a single buzz, if you look at the context (replies, quotes, related topics), you will be able to understand why it stuck. Procedure: First, fix one “stretched post” and extract the material from around it and drop it into the template.

Figure 2: Reproducibility route (extended post → context → cut → next idea)

flowchart LR
P1 [Choose one post that grew] --> P2 [Gather context<br/>Reply/quote/related topic]
P2 --> P3 [Breakdown into sections<br/>Claim/Evidence/Feelings/Conclusion/Target]
P3 --> P4 [Next proposal<br/>Hook → Body → CTA]
P4 --> P5 [Convert to draft and post as candidate]

How to read the diagram: From left to right, a successful example is broken down into "materials". The point is to take the reason from the surrounding reactions = context, rather than the ``post itself that grew''. At the end, I put it into a template and connect it to the next draft.

Step1) Decide on one “post that grows”

Conclusion: At first, it is faster to pin only one post that has grown close to your area. Reason: Increasing the number of candidates can lead to confusion in comparison and shallow analysis. Steps: Select only one post from the posts found in your search with the following conditions:

• Selection check (immediate judgment with bullet points)
• Close to the theme you handle (it doesn't have to be an exact match)
• The reaction is above a certain level (emphasis is placed on whether there is “heat” rather than the number)
• I can imagine whose troubles (I can see the target)

Points where beginners fail (1)

• Choose only flashy numbers (Difficult to reproduce as it is out of sync with your audience)

Step2) Pick up the context (reply/quote/related topic)

Conclusion: The reason for the growth is more likely to be found in "reply/quote/chain topics" than in the main text of the post. Reason: The reaction often verbalizes the ``part of the bite'' in concrete terms. Instructions: Pick up the following three points and write them down (you don't have to read them all).

• Collection points (checklist)
• Reply: sympathy, counterargument, additional questions (what stuck)
• Quote: Misinterpretation/different case (where was expanded)
• Related topics: Topics talked about on the same day and in the same context (what is the premise)

Points where beginners fail (1)

• Save only the post and end (no context remains and cannot be used next time)

Step3) Decompose into sections (element decomposition: claim/evidence/emotion/conclusion/target)

Bottom line: For a winning theme, break down your posts into “elements” and make them reusable. Reason: Copying text in its entirety has low reproducibility, while elements can be replaced with other examples or different steps. Steps: Divide into the following five items and write out one sentence at a time.

• Element decomposition template
• Claim: What do you want to say? (1 sentence)
• Evidence: experience, numbers, examples? (1-2 pieces)
• Emotion: What did the reader feel? (Anxiety/Relief/Anger/Surprise)
• Conclusion: What should I do? (take action)
• Target: Who is it for? (Beginners/persons in charge/managers, etc.)

Points where beginners fail (1)

• Just claims, no evidence (“It seems like that” increases and trust doesn’t build up)

Step4) Next draft (template creation: Hook → Body → CTA)

Conclusion: Sub-proposals can be mass-produced by ``solving the same problem using a different route.'' Reason: Even if the problem (demand) is fixed, changing the order of explanation, examples, and procedures makes it easier to create a new post. Steps: Fill out Hook → Body → CTA with minimal structure.

• Next draft template (can be used as is)
• Hook: Describe your problem in one word (e.g. "For those who spend time searching for X")
• Main text: 3 steps (e.g. "Period → Reaction → Narrow down by context")
• CTA: Next move (e.g. "Determine preservation standards")

Points where beginners fail (1)

• Hook is abstract (Focus on “concrete problems” rather than “Let’s do our best”)

Advanced: Design that narrows down conditions (period, reaction, context)

Conclusion: The key to advanced X searches is to narrow down your search by ``period, response, and context'' before keywords. Reason: Using only keywords tends to increase noise and makes it difficult to meet the “conditions” necessary to extract winning themes. Steps: Narrow down your search using three axes, and finally apply the X search operator (search auxiliary symbol) by “type”.

Narrow down by period (recent/event/season)

Conclusion: Just by cutting the period, you can reduce "old" at once. Reason: Even if the topic is the same, there is a tendency for the most recent problems and phrases to take priority. Procedure: First, use three different periods.

• Typical usage
• Recent: Pick up the latest complaints and latest success stories
• Before and after the event: Pick up updates and trending posts
• Season: Pick up “demands that occur every year” such as the end of the fiscal year, recruitment season, busy season, etc.

Narrow down by reaction (equivalent to likes/replies/quotes/saves)

Conclusion: Looking at the “type of response” gives us more hints as to why there has been growth. Reason: Each reaction tends to have a different meaning, such as likes = sympathy, replies = questions or additions, quotations = context of spread... Procedure: Observe which reactions are more common than the size of the numbers, and reflect that in your next proposal.

• Points to see (checklist)
• Like: Does the population of people with the same problem seem to be large?
• Reply: What is difficult to understand/What do you want added?
• Quote: Which area did it hit?/Is there a different interpretation?
• Equivalent to saving: Are there any “procedures, templates, or checklists” that you would like to refer back to later?

Narrow down by context (“whose problem” x “solution pattern”)

Conclusion: If you fix the winning theme as “whose problem” x “solution pattern”, it will not waver. Reason: If the target is misaligned, the same content will be hit differently and reproducibility will tend to deteriorate. Steps: Tag your post suggestions with these two axes.

• Tagging example
• Whose problem is this: Beginners in SNS operation/Operation agency/Small B2B staff
• Solution pattern: Procedure/Checklist/Failure example → Workaround/Template distribution

Based on this axis, accuracy can be improved by organizing hypotheses on the algorithm side: How to read the Japanese X algorithm

X search operators are "used by type rather than memorized" (5 to 8)

Conclusion: It is more practical to have frequently used "types" on hand than to memorize X search operators. Reason: Since specifications and usability may change, it is difficult to break down if you operate from "purpose → model". Procedure: First, as a representative example, copy and paste the following type and make minor adjustments (the range that can be used may differ depending on the environment).

• Type 1: View only winning posts from a specific account
• Example: from:account name keyword
• Type 2: Reduce blurring with phrase matching
• Example: "Phrase you want to match exactly"
• Type 3: Remove noise with exclusion
• Example: keyword -excluded word
• Type 4: Narrow down by link/image format (typical example)
• Example: keyword filter:links / keyword filter:media
• Type 5: Narrow down by language (typical example)
• Example: keyword lang:ja
• Type 6: Narrow down by period (typical example)
• Example: Keyword since:YYYY-MM-DD until:YYYY-MM-DD
• Type 7: Search for posts with a certain level of reaction (typical example)
• Example: Keyword min_faves:100 (It may not be possible to use it depending on the environment)
• Type 8: Pick up question needs (put them in questions)
• Example: keyword "how" OR "why"

Tips on how to use (checklist)

• Don't make it too complicated (up to 2 such as type 1 + type 3)
• If the search results are rough, check the "period/context" before looking at the operator.
• Save types in Notion/spreadsheet and reuse them

30 minutes a week search routine (reusable template)

Conclusion: Searching will last longer if you complete everything from "extraction to next draft to draft" in 30 minutes once a week, rather than "every day depending on how you feel." Reason: Frequent searches can waste time and make it difficult to produce posts. Procedure: Divide the 30 minutes into "search, extraction, and drafting" and finish with a deliverable (filling in the template).

Figure 4: Weekly 30 minute routine (time block)

flowchart LR
T0[0-5 minutes<br/>Decide your aim<br/>Whose problem/theme] --> T1[5-15 minutes<br/>Search (narrow down by period/reaction)]
T1 --> T2 [15-23 minutes<br/>Gather context<br/>Reply/quote/related]
T2 --> T3 [23-28 minutes<br/>Element analysis<br/>Claim/Evidence/Emotion/Conclusion/Target]
T3 --> T4 [28-30 minutes<br/>Drop one on the next draft template]

How to read the diagram: 30 minutes is fixed as "Search → Pick up → Disassemble → Next idea". The key is to stop your search time at 15 minutes. If you get to ``one template completed'' in the last two minutes, the next day's draft will be faster.

Can be run as is: distribution in minutes

Conclusion: A “fixed” time allocation is the best (it eliminates hesitation). Reason: Searching is a tempting and protracted task. Procedure: Turn on alarm condition.

• 0-5 minutes: Decide on your goal in one line (e.g. "X search tips for B2B personnel")
• 5-15 minutes: Search (narrow down by period → response → context)
• 15-23 minutes: Pick up the context (reply/quote/related)
• 23-28 minutes: Element decomposition (fill in 5 items)
• 28-30 minutes: Drop one line into the next draft template (Hook/text/CTA)

If you want to make your whole routine even more labor-saving, you might also want to consider: X operation workflow in 30 minutes a week

“Item template” for Notion/spreadsheet transcription

Conclusion: Fix the posting items so that you don't just save them. Reason: If there is a format, extracting winning themes becomes “work” and individuality decreases. Instructions: Create the columns below and fill them with one row per post.

• date
• Search query (type used)
• Post URL (or identifying information) that has grown
• Whose worries (target)
• Solution pattern (procedure/checklist, etc.)
• Element decomposition (claim/evidence/emotion/conclusion)
• Next hook plan (1 line)
• Main text of the next draft (3 bullet points)
• Draft CTA (1 line)
• Noise judgment (OK/Hold/Exclude + Reason)

Frequently asked questions (FAQ)

Conclusion: Points that tend to cause blockages can be summarized as no output,'' unbalanced,'' flame,'' time,'' and ``storage standards.'' Reason: Since the X search has a high degree of freedom, it is easy to get lost without standards. Steps: Follow the Q&A below to fine-tune your search conditions and template (deliverables).

Q1. No material (winning theme) comes up when I search.

Conclusion: Rather than increasing the number of keywords, it is a shortcut to narrow down to one problem first. Reason: If the target is wide, the results will be scattered and difficult to make into a usable form. procedure:

• Decide on one target (e.g. operation beginners)
• Decide on one problem (Example: Searching saves time)
• Search again after a different period (latest → before and after the event)

Q2. I get biased because I always get the same opinions.

Conclusion: Replacing the search vocabulary with “another way of saying it” reduces bias. Reason: Each neighborhood uses different words, which changes the population of search results. procedure:

• “Operation” → “Public relations”, “Marketing”, “Recruitment” and other related words
• Pick up opposing opinions (focus on quotes and replies)

Q3. I pick up something that seems to be flaming.

Conclusion: The more responses a post receives, the safer it is to put it on hold based on operational objectives and risks. Reason: Conflict-driven topics tend to be costly to reproduce. procedure:

• Define exclusion rules (flaming/personality attack/assertive incitement) first
• Put something on hold if you can't see "whose problems and solutions"

Q4. Time disappears when you start searching.

Conclusion: It is easy to improve the search time by setting it to "forced termination in 15 minutes". Reason: Searching is an endless process with no end point. procedure:

• Fixed 30 minute blocks (15 minutes for search, 13 minutes for extraction, 2 minutes for next plan)
• Deliverables end with "one template completed"

Q5. What should I save? (I don't know the storage standards)

Conclusion: Preservation is determined based on whether there is information that can be used to extract winning themes. Reason: Even if a post is large in number, it is difficult to become an asset if it cannot be broken down into elements. procedure:

• Target (who) can be seen
• I can see what my worries are.
• I can see the solution (what to do) Prioritize saving those that have all three.

Q6. I can't remember operators.

Conclusion: It is enough to copy and paste the "type" without memorizing it. Reason: Since the purpose is “narrowing down”, it is more stable to turn with the minimum set. procedure:

• Start with type 1 (from:) + type 3 (exclude)
• Add periods, languages, and response types when needed

Q7. Even if you imitate posts that have grown, they will not grow.

Conclusion: What to imitate is not the text, but ``whose problem x solution pattern''. Reason: The same expression can have different impressions depending on the context. procedure:

• Extract the reason from the context (reply/quote)
• Break things down and put evidence and procedures into your own context.

Summary: Make it “unstoppable operation” by searching → winning theme → draft

Conclusion: If you change the search method to "extracting winning themes", you will end up with a system where drafts pile up before you run out of ideas. Reason: Reproducibility remains if you do not end the search with “information gathering” and connect it with elemental decomposition → next draft template. Steps: From today onwards, please complete ``one extended post → context → element decomposition → one next proposal'' in 30 minutes.

Key points (3)

• The purpose of the search is not "gathering stories" but extracting winning themes
• The most addicting part is "old/biased/noise", so filter first
• For longer posts, the context is more important than the main text.

Next action (in order of least hesitation)

• First, fix the once-a-week pattern: X operation workflow that takes 30 minutes a week
• Start by organizing the situation when you are running out of ideas: How to eliminate running out of ideas
• Having a context hypothesis that is easy to grow: How to read the Japanese X algorithm

CTA: Search → Generate → Make drafts “less troublesome”

Even if you find a winning theme through search, it tends to stall if you manually break down the elements and create a template for the next proposal. TenguX can be used as a conduit to easily shorten the flow from Search → Generate → Draft using the same steps. People who say, ``I understand the work, but it's a pain to do it every week'' are likely to be a good match. https://tengux.com/

Read next article

• How to eliminate running out of material
• X operation workflow in 30 minutes a week
• How to read Japan's X algorithm