---
title: "One Person, Four Hours a Week, No New Software: Find the AI Prompts That Skip Your Brand"
description: "A 30-day, four-hours-a-week playbook for building a buyer-prompt list, checking AI answers, tracing competitor citations and fixing the gaps you can reach."
url: "https://groas.com/post/one-person-thirty-days-no-new-software-m"
image: "https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/b1e62f0c-2589-40d3-a28b-c8e952cec60e.png"
published: "2026-10-18T05:27:26.764Z"
modified: "2026-10-18T05:27:26.823Z"
---

October 18, 2026 · 9 min read

# One Person, Four Hours a Week, No New Software: Find the AI Prompts That Skip Your Brand

[Alexander PerelmanHead Of Product @ groas](https://groas.com/author/alexander-perelman)[LinkedIn](https://www.linkedin.com/in/alexander-433793253/)

![Cartoon: a fortune teller's crystal ball shows a rival plumber's van while the overlooked plumber calmly logs it all in a notebook beside an egg timer.](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/b1e62f0c-2589-40d3-a28b-c8e952cec60e.png)

In this article

1. [1. Week 1: Spend four hours building a 40-prompt inventory](#1-week-1-spend-four-hours-building-a-40-prompt-inventory)
2. [2. Week 2: Spend four hours logging actual answers](#2-week-2-spend-four-hours-logging-actual-answers)
3. [3. Week 3: Spend four hours separating missing answers from blocked ones](#3-week-3-spend-four-hours-separating-missing-answers-from-blocked-ones)
4. [4. Week 4: Spend four hours fixing two gaps and checking the result](#4-week-4-spend-four-hours-fixing-two-gaps-and-checking-the-result)
5. [Day 30: Keep the list; lift the constraint only when the work demands it](#day-30-keep-the-list-lift-the-constraint-only-when-the-work-demands-it)

One person, four hours a week, thirty days, and no new software: that is the constraint. If an AI answer names your competitors and skips you, do not buy a dashboard yet. Pull the buyer questions, sort them by intent, inspect the answers and fix what you find. It is search-term mining with a new interface.

I spent years opening Google Ads search terms reports instead of trusting a tidy keyword list. The useful work was finding what people actually typed, separating purchase intent from expensive noise, and changing the account accordingly. A monitoring platform could show me more charts. It could not make that judgment for me.

The same rule applies here. A visibility tracker can count mentions and flag a change, but it cannot tell you whether you tested a question buyers ask, whether your page answers it, or whether a crawler can reach it. **Build a list you understand before you pay to monitor it.**

## 1. Week 1: Spend four hours building a 40-prompt inventory

**Cost: four hours. Buys: a fixed list of buyer questions**, not a collection of flattering prompts about your brand. Cut the urge to test “Do you know \[My Company]?” A polite answer to that question tells you little about what happens when a buyer asks for options.

Pull language from sources you already have:

- **Google Search Console:** queries that earn impressions but few clicks.
- **Google Ads search terms:** longer phrases that suggest someone is evaluating a purchase.
- **Support searches and tickets:** the problems people describe in their own words.
- **Sales notes:** comparisons and objections that come up before a decision.

Copy the questions into a spreadsheet. You do not need a perfect export or a new taxonomy. You need wording close enough to what a buyer might put into ChatGPT, Gemini, Perplexity or Google AI Overviews.

![Search-term printouts marked in red beside cards labelled Category, Comparison and Problem.](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/8ad14d22-767e-411d-a3f0-2a10a39d26d8.png)

Sort each prompt into one of three buckets:

1. **Category:** “What is the most reliable fleet tracking hardware for small trucking companies?”
2. **Comparison:** “Samsara vs Motive for under twenty vehicles” or “best alternatives to ServiceTitan for solo plumbers.”
3. **Problem:** “How to stop Google Ads click fraud on emergency plumbing campaigns.”

Aim for roughly fifteen category prompts, fifteen comparison prompts and ten problem prompts. The [prompt sheet of 60 buyer questions](https://groas.com/post/the-prompt-sheet-60-buyer-questions-to-p) can help if you are staring at an empty column, but use it to frame questions from your market, not to substitute generic ones for your own records.

**Cap the inventory at forty.** Cut duplicates, brand-name vanity checks and questions whose answers would not affect a purchase. Forty is enough to reveal patterns and small enough to maintain. You will not test all forty this month; the list is the queue from which you will choose the first ten. If you cannot explain why a prompt made the cut, remove it before Week 2.

## 2. Week 2: Spend four hours logging actual answers

**Cost: four hours. Buys: a comparable baseline for ten prompts.** Select the ten questions with the clearest purchase intent and the strongest connection to what you sell. Save the other thirty. Trying to run every prompt repeatedly across every engine this week would turn a manageable check into a second job.

Create six columns: `Prompt`, `Engine`, `Brand Position`, `Mentioned Brands`, `Cited URLs` and `Source Type`. Run each of the ten prompts once in ChatGPT with web search active, Gemini, Perplexity and Google AI Overviews where available. Record the answer you get, not the answer you expected. Copy cited URLs while they are in front of you; reconstructing them from memory next week is a fine way to lose an afternoon.

Use the same position labels throughout:

- `Absent`: your brand does not appear.
- `Mentioned`: it appears in passing.
- `Cited`: the answer links to your domain.
- `Shortlisted`: it names you among the leading choices.
- `Recommended`: it picks you as the primary answer.

These are **logging labels, not a claim that every answer climbs a neat ladder**. An answer can shortlist a brand without linking to it, or cite a page without recommending its product. Keep the label that best describes the recommendation, and preserve the URL separately. For a competitor mention, note whether the cited source is its own page, a review, a directory or a discussion thread. If you would rather adapt a template than build formulas, the [AI visibility tracker framework](https://groas.com/post/build-your-own-ai-visibility-tracker-in) is a starting point.

![Five brand-visibility labels for AI answers: Absent, Mentioned, Cited, Shortlisted and Recommended.](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/9f720e64-8eb8-4dc8-8855-9b843d6eff7d.png)

Now choose **five of the ten prompts** that matter most and run each twice more across the same four engines, on different days if your schedule allows. That is forty initial checks plus forty repeat checks: eighty answers to inspect and log in four hours. If an interface is unavailable or an answer has no citations, record that rather than quietly replacing the test. The point is a baseline you can revisit, not a beautiful spreadsheet full of invented certainty.

Do not turn one screenshot into a victory report. [Identical prompts can produce different AI answers across sessions](https://www.visiblie.com/blog/does-chatgpt-give-the-same-answers). If your brand appears in one of three runs, write “one of three.” Do not call it permanent visibility. Keep the prompt wording fixed so a later change tells you something more useful than “I asked a different question.”

## 3. Week 3: Spend four hours separating missing answers from blocked ones

**Cost: four hours. Buys: a short fix list with a reason behind each item.** Filter the sheet for prompts where your brand is `Absent` and a direct competitor appears. Start with the repeatedly tested prompts. A gap seen once is worth noting; a gap seen across several runs deserves your first hour.

Open the URLs attached to those competitor answers. Do not assume the engine cited a product homepage. It may have drawn from a comparison, a directory, a review or a discussion that describes the buyer’s problem well. Write down what each page supplies that your available pages do not: a direct answer, a usable comparison, a clear product description or a detail the buyer needs to decide.

Then check whether your own answer exists. I use two labels here:

- A _content gap_ means you have no page that answers the buyer’s question directly.
- An _access gap_ means you have a relevant page, but a crawler may be unable to retrieve its content.

**Do not commission new copy to solve an access gap.** Check your `robots.txt` rules and your site’s security settings. Review server or edge logs if you have access to them. A direct request using a crawler’s user-agent string can be a useful quick check for a block or challenge, but a `200 OK` response from your machine does not prove that the real crawler can fetch, render and use the page. A `403 Forbidden`, `401 Unauthorized` or challenge page is a reason to investigate. If you manage crawler rules, note that [OpenAI distinguishes its training crawler from OAI-SearchBot](https://developers.openai.com/api/docs/bots); do not treat every bot setting as the same switch.

![An AI search crawler blocked by a 403 error at the CDN edge, beside a route to an answer page.](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/9ba5e254-917d-46fb-be8d-3e9a8b795a72.png)

For a content gap, put the cited page beside yours and look past the design. Does it answer the question under a useful heading? Does it state the relevant parameters plainly? Can a reader tell what the product does and who it is for without decoding marketing language? A table may help with a genuine comparison, but do not invent pricing tiers, specifications or a verdict just to fill cells.

End the week with **at most two fixes** you can make yourself in Week 4. Cut the rest from this month’s workload, not from the inventory. If the strongest gap depends on an independent review or community discussion, a new brand-owned paragraph may not solve it. Log that limitation and choose a fix you can actually ship.

## 4. Week 4: Spend four hours fixing two gaps and checking the result

**Cost: four hours. Buys: two published answers, an access check and a small retest.** Cut the plan to write five new pages. Five rushed pages would leave no time to check whether they say anything useful or whether anyone can fetch them.

For each of your two selected gaps, improve an existing relevant page or add a focused answer where it belongs. Give the section a heading that matches the buyer’s question. Lead with a direct answer, then supply the comparison details or operating limits you can support. If a short table makes the choice easier to understand, use one. If it does not, write two clear paragraphs and stop. **Answer-first is a writing decision, not permission to make up precision.**

![A bloated text draft beside a concise, structured answer card.](https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/be6b7cad-a407-4fed-b98c-a52522f7352f.png)

Publish early enough in the week to leave room for a check near day thirty. Confirm that the pages load, that the answer appears in the page content, and that your crawler-access checks do not reveal an obvious block. Do not mistake publication for a citation. You control the page; you do not control when an engine retrieves it or what it chooses to recommend.

Use the final hour to rerun **the two fixed prompts**, three times each across the four engines you tested in Week 2. Enter the new results beside the old ones. If the brand moves from `Absent` to a mention or citation, keep the URLs and note which runs changed. If nothing moves, check whether the updated page has had a fair chance to be retrieved before rewriting it again. Either outcome is useful: you have a recorded result, not a screenshot looking for applause.

## Day 30: Keep the list; lift the constraint only when the work demands it

At day thirty, you have spent at most sixteen hours and bought no new software. You have forty buyer prompts, hands-on results for the ten you prioritised, repeated checks for five, a record of who appears in the answers, and two targeted fixes with a small before-and-after check. You do **not** have a permanent ranking, complete coverage of every AI surface or proof that two page edits caused a citation. Those are not defects to hide in a slide deck. They tell you what the next round of work must measure.

Keep the weekly schedule if the list still fits. Rotate the untested prompts into your checks, revisit the ones that matter commercially, and log changes in the same sheet. The list is the asset. The tool is optional.

When the four-hour ceiling lifts because the prompt set, paid-search account and content workload have outgrown one person, stop doing repetitive checks by hand. A passive dashboard can tell you a mention disappeared; it still leaves someone to investigate the gap and make the change. [groas](https://groas.com/) is built for the work beyond that alert: specialized models execute search campaign and content work continuously, while a named human strategist sets direction and remains accountable. I would start with the spreadsheet anyway. Once you know which buyer questions matter and why you are missing from them, you can hand off execution without paying for charts you cannot read.

## Frequently Asked Questions

### How do I find the AI prompts where my brand is being skipped?

Pull real buyer language from sources you already have: Google Search Console queries that earn impressions but few clicks, Google Ads search terms, support searches and tickets, and sales notes. Sort them into category, comparison and problem prompts and cap the list at forty.

### Should I test AI tools with questions about my own brand name?

No. The article advises cutting the urge to test prompts like 'Do you know \[My Company]?' because a polite answer to that question tells you little about what happens when a buyer asks for options. Test the questions buyers actually ask when evaluating a purchase instead.

### Which AI engines should I check, and how do I record where my brand appears?

Run each prompt once in ChatGPT with web search active, Gemini, Perplexity and Google AI Overviews where available, and log results in a spreadsheet with columns for prompt, engine, brand position, mentioned brands, cited URLs and source type. Use the labels Absent, Mentioned, Cited, Shortlisted or Recommended, and keep the label that best describes the recommendation while preserving the URL separately.

### Why repeat the same prompt multiple times when checking AI answers?

Identical prompts can produce different AI answers across sessions, so a single screenshot is not proof of visibility. If a brand appears in one of three runs, record 'one of three' rather than calling it permanent visibility, and keep the prompt wording fixed so a later change tells you something meaningful.

### What is the difference between a content gap and an access gap?

A content gap means you have no page that answers the buyer's question directly, while an access gap means you have a relevant page but a crawler may be unable to retrieve its content. Do not commission new copy to solve an access gap; check your robots.txt rules, security settings and server or edge logs instead.

### How many pages should I fix in the first month, and how should I write them?

Fix at most two gaps rather than writing five rushed pages. Improve an existing relevant page or add a focused answer where it belongs, with a heading that matches the buyer's question, a direct answer up front, and only comparison details you can support. Then rerun the two fixed prompts three times each across the four engines to check the result.

### Do I need paid AI visibility monitoring software to start?

No. A passive dashboard can tell you a mention disappeared, but it still leaves someone to investigate the gap and make the change, and it cannot judge whether you tested a question buyers ask or whether a crawler can reach your page. The article recommends starting with a spreadsheet and only lifting the four-hour constraint when the prompt set and content workload outgrow one person.

## Pay For Results, Not For Hours

Businesses buy the outcome, agencies resell it, and groas answers for it either way.

[See If You Qualify](https://groas.typeform.com/to/xC1bQNUT)

## Structured data

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://graph.groas.com/entity/groas","name":"groas","url":"https://groas.com/","logo":{"@type":"ImageObject","url":"https://groas.com/icon-512.png","width":512,"height":512},"sameAs":["https://www.linkedin.com/company/groas/"]},{"@type":"WebSite","@id":"https://groas.com/#website","name":"groas","url":"https://groas.com/","inLanguage":"en-US","publisher":{"@id":"https://graph.groas.com/entity/groas"}},{"@type":"WebPage","@id":"https://groas.com/post/one-person-thirty-days-no-new-software-m#webpage","url":"https://groas.com/post/one-person-thirty-days-no-new-software-m","name":"One Person, Four Hours a Week, No New Software: Find the AI Prompts That Skip Your Brand","description":"A 30-day, four-hours-a-week playbook for building a buyer-prompt list, checking AI answers, tracing competitor citations and fixing the gaps you can reach.","inLanguage":"en-US","isPartOf":{"@id":"https://groas.com/#website"},"about":{"@id":"https://graph.groas.com/entity/groas"},"primaryImageOfPage":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/b1e62f0c-2589-40d3-a28b-c8e952cec60e.png"},"breadcrumb":{"@id":"https://groas.com/post/one-person-thirty-days-no-new-software-m#breadcrumb"},"dateModified":"2026-10-18T05:27:26.823Z","mainEntity":{"@id":"https://groas.com/post/one-person-thirty-days-no-new-software-m#article"}},{"@type":"BreadcrumbList","@id":"https://groas.com/post/one-person-thirty-days-no-new-software-m#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://groas.com/"},{"@type":"ListItem","position":2,"name":"Blog","item":"https://groas.com/blog"},{"@type":"ListItem","position":3,"name":"One Person, Four Hours a Week, No New Software: Find the AI Prompts That Skip Your Brand","item":"https://groas.com/post/one-person-thirty-days-no-new-software-m"}]},{"@type":"BlogPosting","@id":"https://groas.com/post/one-person-thirty-days-no-new-software-m#article","url":"https://groas.com/post/one-person-thirty-days-no-new-software-m","mainEntityOfPage":{"@id":"https://groas.com/post/one-person-thirty-days-no-new-software-m#webpage"},"isPartOf":{"@id":"https://groas.com/blog#blog"},"headline":"One Person, Four Hours a Week, No New Software: Find the AI Prompts That Skip Your Brand","description":"A 30-day, four-hours-a-week playbook for building a buyer-prompt list, checking AI answers, tracing competitor citations and fixing the gaps you can reach.","datePublished":"2026-10-18T05:27:26.764Z","dateModified":"2026-10-18T05:27:26.823Z","image":{"@type":"ImageObject","url":"https://pub-87da24ecbbfc4c3bad6875f3aa013712.r2.dev/generated-images/b1e62f0c-2589-40d3-a28b-c8e952cec60e.png"},"author":{"@id":"https://groas.com/author/alexander-perelman#person"},"publisher":{"@id":"https://graph.groas.com/entity/groas"},"keywords":"four hours, prompts, four, week, hours, new, answer, brand","wordCount":1817,"inLanguage":"en-US"},{"@type":"Person","@id":"https://groas.com/author/alexander-perelman#person","name":"Alexander Perelman","jobTitle":"Head Of Product @ groas","description":"Ex Goldman Sachs and Ex Stanford Computer Science","image":"https://groas.com/media/blog/4f17cac3a81acc224cc1b2cfab8f94bbc295a3089d13e8421125771ef8d4064f.jpg","worksFor":{"@id":"https://graph.groas.com/entity/groas"},"sameAs":["https://www.linkedin.com/in/alexander-433793253/"]},{"@type":"FAQPage","@id":"https://groas.com/post/one-person-thirty-days-no-new-software-m#faq","mainEntity":[{"@type":"Question","name":"How do I find the AI prompts where my brand is being skipped?","acceptedAnswer":{"@type":"Answer","text":"Pull real buyer language from sources you already have: Google Search Console queries that earn impressions but few clicks, Google Ads search terms, support searches and tickets, and sales notes. Sort them into category, comparison and problem prompts and cap the list at forty."}},{"@type":"Question","name":"Should I test AI tools with questions about my own brand name?","acceptedAnswer":{"@type":"Answer","text":"No. The article advises cutting the urge to test prompts like 'Do you know [My Company]?' because a polite answer to that question tells you little about what happens when a buyer asks for options. Test the questions buyers actually ask when evaluating a purchase instead."}},{"@type":"Question","name":"Which AI engines should I check, and how do I record where my brand appears?","acceptedAnswer":{"@type":"Answer","text":"Run each prompt once in ChatGPT with web search active, Gemini, Perplexity and Google AI Overviews where available, and log results in a spreadsheet with columns for prompt, engine, brand position, mentioned brands, cited URLs and source type. Use the labels Absent, Mentioned, Cited, Shortlisted or Recommended, and keep the label that best describes the recommendation while preserving the URL separately."}},{"@type":"Question","name":"Why repeat the same prompt multiple times when checking AI answers?","acceptedAnswer":{"@type":"Answer","text":"Identical prompts can produce different AI answers across sessions, so a single screenshot is not proof of visibility. If a brand appears in one of three runs, record 'one of three' rather than calling it permanent visibility, and keep the prompt wording fixed so a later change tells you something meaningful."}},{"@type":"Question","name":"What is the difference between a content gap and an access gap?","acceptedAnswer":{"@type":"Answer","text":"A content gap means you have no page that answers the buyer's question directly, while an access gap means you have a relevant page but a crawler may be unable to retrieve its content. Do not commission new copy to solve an access gap; check your robots.txt rules, security settings and server or edge logs instead."}},{"@type":"Question","name":"How many pages should I fix in the first month, and how should I write them?","acceptedAnswer":{"@type":"Answer","text":"Fix at most two gaps rather than writing five rushed pages. Improve an existing relevant page or add a focused answer where it belongs, with a heading that matches the buyer's question, a direct answer up front, and only comparison details you can support. Then rerun the two fixed prompts three times each across the four engines to check the result."}},{"@type":"Question","name":"Do I need paid AI visibility monitoring software to start?","acceptedAnswer":{"@type":"Answer","text":"No. A passive dashboard can tell you a mention disappeared, but it still leaves someone to investigate the gap and make the change, and it cannot judge whether you tested a question buyers ask or whether a crawler can reach your page. The article recommends starting with a spreadsheet and only lifting the four-hour constraint when the prompt set and content workload outgrow one person."}}]}]}
```