Start with someone trying to do something. Imagine they want to meditate for five minutes without downloading an app. Their search is a request for help with that task. Describe that need before looking at what search already offers.
01A real need
01 A real need
Imagine someone wants five quiet minutes without downloading an app. Describe that need before planning the page.
Person: someone who does not want an app.
Task: time five minutes now.
02A search
02 A search
They might search for “meditation timer no download.” This is a teaching hypothesis. We do not yet know whether people actually search this way.
Suggested query: meditation timer no download.
Search results: not checked.
03A result appears
03 A result appears
A relevant page may appear. Being accessible, readable by search, stored, and shown are separate events.
Fictional result title: Free meditation timer, no download.
Description: Choose a duration and start a timer.
This is not a live search result. No impression, ranking, or click data is shown.
04A page visit
04 A page visit
If the title and description fit the need, someone may click. An appearance does not guarantee a click, and a click does not prove successful use.
Planned page: choose a duration, then find “Start timer.”
A page sketch is not a working tool.
05A task completed
05 A task completed
The page should let the visitor finish the timer. Opening it is only a beginning; successful use needs a separate check.
Planned behavior: choose five minutes → start → finish.
Status: not built or tested.
No working timer, real search, or traffic result is provided here.
Do this one thing now
Write four lines on your own paper or in your own document: whom it is for, the task, possible help, and what is still unknown. If you completed the introductory exercise, reuse those four lines.
Here is a completed example
- 01For
- People who do not want to download an app.
- 02Task
- Time five minutes now.
- 03Possible help
- A timer they can use in the browser.
- 04Still unknown
- Whether people actually search this way and whether existing results already meet the need.
Continue with your situation
I do not have a website
Use the fictional timer to write the same four lines. A real search is optional; without one, record “search results: not checked.” No account is needed.
I have a website
Replace the example with the person your existing page helps. Save the four lines. Add a source for a real question if you have one; otherwise keep the hypothesis label. A search result is not a user interview.
Keep it on your own paper or in your own document. No upload is needed. Mark work you have not done “not checked”; do not guess.
How do I know I am done?
You can explain all four lines in your own words, label the unverified need as a hypothesis, and honestly state what is unknown. No website or account is needed.
What can wait?
You do not need search volumes, paid tools, or dozens of topics. If you have no idea yet, fill in the example first.
Open when you need more
After writing, want to see which pages already exist?
Optional: search for the words you think they would use and open one ordinary result without an ad label. Separately record its address and what it offers. One observation does not establish demand or count as a user interview.
How is my suggested search different from real demand?
The words a person actually searches are a query. Terms you use to plan a topic are often called keywords. Your sentence is a guess for now. Real questions, interviews, or authorized search reports can add evidence, but one question still cannot tell you monthly search volume.
Complete work checklist, technical guides, templates, and sources
After the small exercise, consult this for your real project as needed. It retains the full scope of the original six-stage workflow.
Editorial sources reviewed · Not tested end-to-end on a live site
Entry requirement
A website direction is enough; existing traffic and paid keyword tools are not required.
Goal
Choose one evidenced question you can credibly answer on a useful page.
Concepts first
- Query: words entered into search. Intent: the task behind them; a query can have several intents.
- Evidence of need, search volume, and commercial value are different. One real question does not establish monthly demand or willingness to pay.
Search intent
The task someone wants to complete with a search.
Do not confuse: Your sales goal is not necessarily their goal; intent is a hypothesis.
See examples and evidence →Keyword / query
A query is what someone searches; a keyword is a term you use to plan content.
Do not confuse: Synonyms need not each get a page; repetition is not the goal.
See examples and evidence →Long-tail keyword
A less frequently searched query, often expressing a more specific need.
Do not confuse: Long-tail does not just mean more words, or guarantee easy rankings or conversions.
See examples and evidence →SERP
The full results page, which can include organic results, ads, and other features.
Do not confuse: An ad is not an organic ranking; one screenshot is not everyone’s result.
See examples and evidence →Ranking
How the engine selects and orders results for a query.
Do not confuse: There is no permanent site-wide rank across all queries and regions.
See examples and evidence →Inputs
- One sentence describing your audience and the experience, material, or product you can offer.
- Existing sites can bring a recent Search Console query export; new sites can start with interviews, support questions, or public discussions.
Actions
01.1Collect real questions
Collect 5–10 questions from accessible customer conversations, public communities, or your own practice notes. Record wording, source or traceable reference, date, audience, and context. Remove personal identifiers. Mark unsupported ideas as hypotheses.
Output: A question list separating observations from hypotheses.
01.2Inspect search results
Search three candidate questions in the target language and region. Record the date, query, and the formats of several organic results: guide, tool, product, comparison, or local service. Open the pages to inspect their answers. Separate ads from organic results. Treat this as a small observation, not a comprehensive ranking study.
Output: An intent hypothesis, page format, and unmet question for each candidate.
01.3Choose one feasible question
For each candidate ask: Does someone actually face this? Can I provide useful steps, examples, or evidence? Can one page deliver it? Does it fit my site? Mark yes/no/unknown and choose the clearest feasible opportunity rather than inventing a precision score.
Output: One primary question, rationale, and a deferred list.
01.4Record data limits
If using Google Trends, record region, period, and relative interest; it is not an absolute query count. Keep provider, geography, period, and date for third-party estimates. Without reliable data, leave volume unknown, not zero, and never ask AI to fabricate it.
Output: An opportunity card containing evidence and unknowns.
Tools
- A browser and notes or spreadsheet suffice. Google Trends is optional; Search Console needs property access and available data.
Tasks to delegate within this stage
Source reviewed · Live site NOT_RUN
Source reviewed · Live site NOT_RUN
Deliverables
- Opportunity card fields: audience, context, primary question, intent, sources, dates, region/language, distinctive help, volume with provenance or unknown, and next step.
Copy this stage’s blank template →
Acceptance criteria
- At least one traceable need signal, a specific question, and a reason for your page to exist. Unsupported ideas can remain exploratory but cannot be labeled validated opportunities.
Stop conditions
- No evidenced need or credible ability to answer: research further or find a qualified reviewer before producing pages. Stop on sensitive specialist topics beyond your competence.
Next step
Take the opportunity card to stage 2 and design one page.
Official sources
- FAQ about Google Trends data ↗
Trends uses normalized relative interest, not absolute search volume. · 2026-10-10 - Creating helpful, reliable, people-first content ↗
People-first content, original value, and reliability. · 2026-10-10
Stage structure, checklists, output templates, and diagnostic sequence are original First Crawl teaching design, not a Google-certified curriculum or ranking formula.
Understand AI help before choosing it
The diagram uses a page check to explain input → operation → output → your review. Lesson 1 only needs a four-line record, even without a website.
01Give a goal
01 Give a goal
Give the Agent a specific question and the material it may read. An Agent is the AI doing the task; a Skill supplies its steps.
Lesson 1 input: your four-line record.
Goal: separate a need, a hypothesis, and unknowns.
Scope: organize only the text provided.
02Check the evidence
02 Check the evidence
An Agent follows a Skill to organize material or run authorized checks. Without material, account access, or actual execution, it should record “not checked.”
Lesson 1 operation: read the four lines and identify assumptions.
For a separate search observation: confirm the query, language, region, and scope.
This node only opens an example. It does not run an Agent or a search.
03Bring back findings
03 Bring back findings
The output should separate actual observations from unchecked items, then suggest a next step.
- Simulated known
- The page opens.
- Simulated unknown
- Indexing has not been checked.
- Suggested next step
- An authorized owner reviews the indexing report.
This is the diagram’s simulated check. Lesson 1 still produces four lines; a guess does not establish real demand.
04You decide
04 You decide
Review the output before deciding on a suggestion. The diagram contains no execution, approval, or publishing action.
- Where did the material come from?
- What was actually done, and what is unchecked?
- Which problem does the suggestion address?
- Does it require more permission?
For lesson 1, explain your four lines yourself and keep hypotheses and unknowns. There is nothing to install now.
I understand this step. See how AI can help.
Optional. Check the required input and expected result first. Installing a Skill does not complete the work or authorize publishing, messaging, or account access.
- Research Search Results
Browse actual search results and competing pages to identify intent, evidence gaps, and original content opportunities.
- Search opportunity evidence
Turn audience problems into sourced opportunities without invented search volume.
The complete packages are source-reviewed; live-site execution is NOT_RUN. Private reports require existing access. Complete instruction packages are available in Chinese and English.
Continue to the next lesson
Next: turn this need into a page plan.