---
title: "One Marketer, Five Hours a Week, 60 Days: A Citation Playbook for AI Search"
description: "1. Set the limit: one marketer, five hours a week, 60 days One marketer, five hours a week, 60 days, and no new tools: that is enough to test whether your…"
url: "https://groas.com/post/one-marketer-five-hours-a-week-60-days-m"
image: "https://groas.com/media/blog/1e64786a5d8c725fd4f49335ef476be9760730a15450e52f52c0d063ba869caa.png"
published: "2026-10-07T05:28:03.372Z"
modified: "2026-10-07T05:28:03.372Z"
---

[AI For Google Ads](https://groas.com/category/ai-for-google-ads) · October 7, 2026 · 9 min read

# One Marketer, Five Hours a Week, 60 Days: A Citation Playbook for AI Search

[DavidFounder & CEO @ groas](https://groas.com/author/david)

![Ten seedlings growing in terracotta pots beside five bare pots labelled CONTROL, with a kitchen timer and notebook: a small, controlled test of targeted effort.](https://groas.com/media/blog/1e64786a5d8c725fd4f49335ef476be9760730a15450e52f52c0d063ba869caa.png)

## 1. Set the limit: one marketer, five hours a week, 60 days

One marketer, five hours a week, 60 days, and no new tools: that is enough to test whether your business can earn citations in ChatGPT and Perplexity. It is not enough to publish fifty posts, chase every adjacent topic, and wait for a slide deck to tell you whether any of it worked.

**Getting cited is a budget-allocation problem, not a content-volume problem.** I run this the way I would set up a new Google Ads account: a small set of tight targets, a control group, and a weekly review. Spraying effort across generic topics has the same appeal as spraying ad spend across thousands of loose keywords. You get plenty of activity. You may not get the buyer.

Five hours a week gives you 40 working hours across eight weeks inside the 60-day window. Spend them in order:

1. **5 hours** to choose 10 buyer questions and check whether crawlers can fetch your pages.
2. **20 hours** to build one answer page per question, including comparison and pricing detail where it belongs.
3. **5 hours** to earn relevant off-site mentions.
4. **10 hours** to test the pages against five unchanged control questions.

That budget cuts new subscriptions, bulk rewrites of old meta descriptions, and generic explainers before they eat the calendar. The first purchase you make with your time is clarity about what buyers actually ask.

### 2. Weeks 1–2: Spend five hours on questions and access

#### Find 10 questions in conversations you already have

Spend roughly 2.5 hours in Week 1 pulling questions from closed-lost sales notes, prospect calls, and your Google Ads search terms report. Look for the friction just before a decision: _How much does implementation cost for a 50-person team?_ _Why does this unit fail after two years?_ _What is the difference between standard compliance and SOC 2 Type II for payment gateways?_

Pick 10. Not 100. A broad prompt such as “what is commercial refrigeration” invites an established reference source to answer it, and the resulting citation is unlikely to tell you much about buyer demand. A specific buying question gives you a page to build and a prompt to test.

I think of these as exact-match targets, not because an AI engine works exactly like a Google Ads match type, but because **each page needs to resolve one identifiable decision**. If a question needs a price, a limitation, or a comparison, record that before you write. Those details determine the page. Otherwise you will spend your two-hour writing slot producing a polished introduction to a question you never answered.

##### Use the next 2.5 hours to check what a bot receives

Before drafting, inspect the raw HTML your server returns. An analysis of [569 million crawler requests](https://crawlmouse.com/blog/can-ai-crawlers-see-javascript) reported that major AI search crawlers, including `OAI-SearchBot` and `PerplexityBot`, fetch raw HTML rather than executing client-side JavaScript. If the answer appears only after a browser runs your app, a crawler may receive a thin shell instead of the passage you wrote.

Start with the [45-minute curl and CDN audit](https://groas.com/post/can-chatgpt-actually-read-your-site-a-45). Then use the rest of this access slot to check:

- **Raw HTML:** Does a page response contain its answer text?
- **`robots.txt`:** Are the search crawlers you want allowed to fetch it?
- **Server responses:** Do bot requests receive a page, or a 403 or rate limit?
- **Sitemap:** Have you submitted the XML sitemap containing the pages you want found?

The bots are not interchangeable. [OpenAI’s crawler documentation](https://developers.openai.com/api/docs/bots) distinguishes `GPTBot`, used for training, from `OAI-SearchBot`, used to surface search results. [Perplexity’s crawler documentation](https://docs.perplexity.ai/docs/resources/perplexity-crawlers) also describes access issues caused by security configurations. In our edge logs, AI search crawlers repeatedly request `/robots.txt` and the homepage root before content. Check those responses rather than assuming a page that loads in your browser is available to a bot.

This is why access comes before writing. **Do not spend 20 hours on answers a crawler cannot read.**

#### 3. Weeks 2–5: Spend 20 hours on 10 answer pages

You have about two hours per question. Write one page for each of the 10 questions you selected, with the answer in its first two sentences. Then supply the facts that make the answer usable: a number you can stand behind, a relevant constraint, or a comparison that helps someone choose.

The placement matters. An [analysis of 680 million generative citations](https://www.leapd.ai/blog/ai-visibility/how-chatgpt-google-ai-overviews-and-perplexity-source-information-in-2026) reported that 44.2% came from the first 30% of a page’s text. [Research on more than one million AI Overviews](https://www.xseek.io/blogs/articles/ai-overviews-what-length-actually-wins) found many summaries in the 100-to-300-word range. Neither finding requires you to make every page short. It gives you a clear editing rule: **put the answer where a hurried reader and a retrieval system can find it.**

![Illustration of a page with its key metrics concentrated in the top third.](https://groas.com/media/blog/c4099bec1c33856b0ec69032423d5319b63a11f1b1973cd52f4bcb6cdc0488c9.png)

After the opening answer, use the page space in this order:

1. Give the facts that support the answer.
2. Explain the conditions under which that answer changes.
3. Show the relevant alternatives, costs, or limitations.
4. Remove background material that does not help with the decision.

Specificity is worth more than an adjective. A [study across 10,000 queries](https://arxiv.org/abs/2311.09735) reported higher AI citation visibility for pages using concrete statistics and attributed quotes, and lower visibility for keyword stuffing. The mechanism is plain enough: a precise, attributable passage gives an assistant something useful to quote. “Our software saves significant time” gives it a claim it cannot inspect. Use your own verified numbers where you have them; do not manufacture a benchmark to make a paragraph look quotable.

Make three of the 10 pages comparison or pricing answers if those are among your buyers’ questions. That means **three pages within the 20-hour budget, not three bonus assignments**. In an [analysis of 1,400 citations from commercial queries](https://www.reddit.com/r/seogrowth/comments/1qr3q0t/we_analyzed_1400_chatgpt_and_perplexity_citations/), comparison matrices, teardowns, and pricing breakdowns accounted for a large share of citations. A table can help when the question genuinely calls for side-by-side trade-offs. Include the conditions under which your offer is not the right fit. A buyer needs those limits, and a page that omits them reads like an ad pretending to be a reference.

I would cut a broad “ultimate guide” before I cut one of these answers. The guide consumes the same scarce hours while making the decision harder to find.

#### 4. Weeks 5–6: Spend five hours where buyers discuss the problem

Allocate 2.5 hours a week to relevant discussions, not a miniature PR campaign. Find specific technical or buying questions on industry forums, applicable GitHub discussions, or trade communities. Answer the question on its own terms. Where useful, point to the facts or benchmarks already published on your answer pages.

The purpose is **external context for a claim you have already made**, not a pile of links. The [680-million-citation analysis](https://www.leapd.ai/blog/ai-visibility/how-chatgpt-google-ai-overviews-and-perplexity-source-information-in-2026) found limited overlap between the domains cited by ChatGPT and Perplexity. A single on-site publishing plan may not reach every place an assistant looks for context. Relevant public discussions give your answers a chance to be encountered outside your own domain.

Keep the constraint intact. No journalist-pitch spreadsheet, no hour spent turning one answer into five lightly rewritten posts, and no drive-by reply that exists only to drop a URL. If a discussion does not address one of your target questions, leave it alone. Five useful hours beat five hours of what a report would call “community engagement.”

#### 5. Weeks 7–8: Spend 10 hours testing against a control

Here is the lab-notebook part. It is a test you run, not a result I am claiming to have produced for you.

**Question:** Do the 10 answer pages gain citations while comparable questions you left alone do not?

**Setup:** Take your 10 target questions and select five commercial questions in the same niche for which you made no page changes and published no off-site answers. These are the controls. Use a plain spreadsheet. In Weeks 7 and 8, run all 15 prompts once a week in ChatGPT Search, Perplexity, and Google AI Overviews. Keep the wording and time of day consistent, and use clean browser profiles with history and personalised memory turned off. The [Prompt-Trigger Test](https://groas.com/post/the-prompt-trigger-test-find-out-which-b) lays out the same basic discipline: do not mistake a general change in search results for a win by your new page.

![Open lab notebook showing 10 target questions beside five control questions.](https://groas.com/media/blog/6268a4e7950333101a0949c8e515831603bf1bb775179f4af98c20043332580a.png)

**Controls:** Do not edit the five control topics or discuss them off-site during the sprint. Keep the prompt wording fixed between runs. You cannot freeze an AI engine’s index, so the control questions help you spot changes happening without your intervention.

**Measure:** For each prompt and engine, log:

- **Citation presence:** Did it link to your domain, and to which page?
- **Passage match:** Did it use the answer near the top of that page, or cite you for something else?
- **Control movement:** Did your domain appear on any of the five unchanged questions?
- **Other cited sources:** Did a source replace yours, or did the answer stop showing citations?

**Expectation:** I expect a tightly answered page that a bot can fetch to have a better chance on its matching question than on an untouched control. The mechanism is the combination you paid for: a specific target, accessible HTML, and a quotable answer near the top. I would not treat a particular engine’s pickup date as a deadline. Index updates and answer selection can move independently of your publishing schedule.

Read the two runs as evidence, not a verdict stamped in ink. If a target earns a citation in one run and loses it in the next, inspect the other sources before rewriting. If targets improve while controls stay flat, keep working that page pattern. If targets and controls move together, do not credit the new pages for the whole change. If neither moves, check access and the quality of the answers before commissioning more content. **Either result changes what you do next; neither justifies publishing 50 more posts on instinct.**

#### 6. If the limit drops to two hours a week, cut the scope

At two hours a week, you have 16 hours across eight weeks. Do not perform the 40-hour plan badly. Pick four buyer questions instead of 10 and cut the off-site work entirely. Spend the first two hours choosing those questions and checking that the raw page HTML and crawler responses work. Use the remaining 14 hours for four answer pages, roughly 3.5 hours each, with comparison or pricing detail where the question demands it.

This version does not buy the same test coverage. Log the four prompts as you can, but do not present a smaller, less controlled check as proof of a citation lift. **Cut breadth before you cut the answer or the access check.** Four pages that resolve actual buying questions are a better use of the constraint than 10 thin posts assembled to hit a publishing target.

#### 7. When the time constraint lifts, stop doing the checklist by hand

I would run this sprint myself if I had one marketer, 40 hours, and no software budget. It gives you a set of buyer questions, pages built for those questions, and a record of whether citations appeared. It also shows the limit of manual execution. Prompts, competitors, search results, and paid-search performance do not wait for a weekly spreadsheet review.

If your search programme has grown beyond that cadence, do not turn the five-hour checklist into a larger pile of manual checklists. That is the point to move execution to [groas](https://groas.com/): specialised models handle campaign and search-visibility work continuously while a human strategist sets direction and guardrails. Keep the questions and the habit of measuring outcomes. When the constraint lifts, let the repetitive execution go.

## Frequently Asked Questions

### How many hours does it take to test whether ChatGPT and Perplexity will cite my site?

The plan allots 40 working hours to one marketer over 60 days: 5 hours to choose 10 buyer questions and check crawler access, 20 hours to build one answer page per question, 5 hours to earn relevant off-site mentions, and 10 hours to test the pages against five unchanged control questions.

### Where should I find the buyer questions to target for AI search citations?

Pull questions from closed-lost sales notes, prospect calls, and your Google Ads search terms report. Pick 10 specific buying questions rather than 100 broad ones, because a broad prompt such as what is commercial refrigeration usually gets answered by an established reference source and tells you little about buyer demand.

### How do I check whether AI crawlers like OAI-SearchBot and PerplexityBot can actually read my pages?

Inspect the raw HTML your server returns, since AI search crawlers fetch raw HTML rather than executing client-side JavaScript. Check that pages contain their answer text, that robots.txt allows the crawlers you want, that bot requests get a page instead of a 403 or rate limit, and that you have submitted an XML sitemap with the pages.

### How should I structure an answer page so AI engines quote it?

Put the answer in the first two sentences, since an analysis of 680 million generative citations found 44.2% came from the first 30% of a page's text. Then give the supporting facts, the conditions under which the answer changes, and relevant alternatives or limitations, using concrete statistics you can stand behind rather than vague claims.

### Should I include comparison or pricing pages in my answer pages?

Yes. Make three of the 10 pages comparison or pricing answers if those are among your buyers' questions, inside the 20-hour budget rather than as bonus assignments. Include the conditions under which your offer is not the right fit, since a page that omits its limits reads like an ad pretending to be a reference.

### What kind of off-site work helps get cited by ChatGPT and Perplexity?

Spend the 5 hours answering specific technical or buying questions in industry forums, applicable GitHub discussions, or trade communities, pointing to facts or benchmarks already published on your answer pages. The purpose is external context for claims you have already made, not a pile of links or a miniature PR campaign.

### How do I test whether my new pages actually earned the AI citations?

Run the 10 target questions plus 5 unchanged commercial control questions once a week for two weeks in ChatGPT Search, Perplexity, and Google AI Overviews, with fixed wording, consistent timing, and clean browser profiles. Log citation presence, passage match, control movement, and other cited sources. If targets and controls move together, do not credit the new pages for the change.

### What should I do if I only have two hours a week instead of five?

Pick four buyer questions instead of 10 and cut the off-site work entirely. Spend the first two hours choosing those questions and checking raw HTML and crawler responses, then use the remaining 14 hours for four answer pages of roughly 3.5 hours each. This version does not buy the same test coverage, so do not present it as proof of a citation lift.

## 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-marketer-five-hours-a-week-60-days-m#webpage","url":"https://groas.com/post/one-marketer-five-hours-a-week-60-days-m","name":"One Marketer, Five Hours a Week, 60 Days: A Citation Playbook for AI Search","description":"1. Set the limit: one marketer, five hours a week, 60 days One marketer, five hours a week, 60 days, and no new tools: that is enough to test whether your…","inLanguage":"en-US","isPartOf":{"@id":"https://groas.com/#website"},"about":{"@id":"https://graph.groas.com/entity/groas"},"primaryImageOfPage":{"@type":"ImageObject","url":"https://groas.com/media/blog/1e64786a5d8c725fd4f49335ef476be9760730a15450e52f52c0d063ba869caa.png"},"breadcrumb":{"@id":"https://groas.com/post/one-marketer-five-hours-a-week-60-days-m#breadcrumb"},"dateModified":"2026-10-07T05:28:03.372Z","mainEntity":{"@id":"https://groas.com/post/one-marketer-five-hours-a-week-60-days-m#article"}},{"@type":"BreadcrumbList","@id":"https://groas.com/post/one-marketer-five-hours-a-week-60-days-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":"AI For Google Ads","item":"https://groas.com/category/ai-for-google-ads"},{"@type":"ListItem","position":4,"name":"One Marketer, Five Hours a Week, 60 Days: A Citation Playbook for AI Search","item":"https://groas.com/post/one-marketer-five-hours-a-week-60-days-m"}]},{"@type":"BlogPosting","@id":"https://groas.com/post/one-marketer-five-hours-a-week-60-days-m#article","url":"https://groas.com/post/one-marketer-five-hours-a-week-60-days-m","mainEntityOfPage":{"@id":"https://groas.com/post/one-marketer-five-hours-a-week-60-days-m#webpage"},"isPartOf":{"@id":"https://groas.com/blog#blog"},"headline":"One Marketer, Five Hours a Week, 60 Days: A Citation Playbook for AI Search","description":"1. Set the limit: one marketer, five hours a week, 60 days One marketer, five hours a week, 60 days, and no new tools: that is enough to test whether your business can earn citations in ChatGPT and Perplexity. It is not enough to publish…","datePublished":"2026-10-07T05:28:03.372Z","dateModified":"2026-10-07T05:28:03.372Z","image":{"@type":"ImageObject","url":"https://groas.com/media/blog/1e64786a5d8c725fd4f49335ef476be9760730a15450e52f52c0d063ba869caa.png"},"author":{"@id":"https://groas.com/author/david#person"},"publisher":{"@id":"https://graph.groas.com/entity/groas"},"articleSection":"AI For Google Ads","keywords":"hours, five, five hours, week, answer, marketer five, page, questions","wordCount":1918,"inLanguage":"en-US"},{"@type":"Person","@id":"https://groas.com/author/david#person","name":"David","jobTitle":"Founder & CEO @ groas","image":"https://groas.com/media/blog/363be9ad5255816654dac065d01bf1c6063110b400bb38cc199570267232027c.jpg","worksFor":{"@id":"https://graph.groas.com/entity/groas"}},{"@type":"FAQPage","@id":"https://groas.com/post/one-marketer-five-hours-a-week-60-days-m#faq","mainEntity":[{"@type":"Question","name":"How many hours does it take to test whether ChatGPT and Perplexity will cite my site?","acceptedAnswer":{"@type":"Answer","text":"The plan allots 40 working hours to one marketer over 60 days: 5 hours to choose 10 buyer questions and check crawler access, 20 hours to build one answer page per question, 5 hours to earn relevant off-site mentions, and 10 hours to test the pages against five unchanged control questions."}},{"@type":"Question","name":"Where should I find the buyer questions to target for AI search citations?","acceptedAnswer":{"@type":"Answer","text":"Pull questions from closed-lost sales notes, prospect calls, and your Google Ads search terms report. Pick 10 specific buying questions rather than 100 broad ones, because a broad prompt such as what is commercial refrigeration usually gets answered by an established reference source and tells you little about buyer demand."}},{"@type":"Question","name":"How do I check whether AI crawlers like OAI-SearchBot and PerplexityBot can actually read my pages?","acceptedAnswer":{"@type":"Answer","text":"Inspect the raw HTML your server returns, since AI search crawlers fetch raw HTML rather than executing client-side JavaScript. Check that pages contain their answer text, that robots.txt allows the crawlers you want, that bot requests get a page instead of a 403 or rate limit, and that you have submitted an XML sitemap with the pages."}},{"@type":"Question","name":"How should I structure an answer page so AI engines quote it?","acceptedAnswer":{"@type":"Answer","text":"Put the answer in the first two sentences, since an analysis of 680 million generative citations found 44.2% came from the first 30% of a page's text. Then give the supporting facts, the conditions under which the answer changes, and relevant alternatives or limitations, using concrete statistics you can stand behind rather than vague claims."}},{"@type":"Question","name":"Should I include comparison or pricing pages in my answer pages?","acceptedAnswer":{"@type":"Answer","text":"Yes. Make three of the 10 pages comparison or pricing answers if those are among your buyers' questions, inside the 20-hour budget rather than as bonus assignments. Include the conditions under which your offer is not the right fit, since a page that omits its limits reads like an ad pretending to be a reference."}},{"@type":"Question","name":"What kind of off-site work helps get cited by ChatGPT and Perplexity?","acceptedAnswer":{"@type":"Answer","text":"Spend the 5 hours answering specific technical or buying questions in industry forums, applicable GitHub discussions, or trade communities, pointing to facts or benchmarks already published on your answer pages. The purpose is external context for claims you have already made, not a pile of links or a miniature PR campaign."}},{"@type":"Question","name":"How do I test whether my new pages actually earned the AI citations?","acceptedAnswer":{"@type":"Answer","text":"Run the 10 target questions plus 5 unchanged commercial control questions once a week for two weeks in ChatGPT Search, Perplexity, and Google AI Overviews, with fixed wording, consistent timing, and clean browser profiles. Log citation presence, passage match, control movement, and other cited sources. If targets and controls move together, do not credit the new pages for the change."}},{"@type":"Question","name":"What should I do if I only have two hours a week instead of five?","acceptedAnswer":{"@type":"Answer","text":"Pick four buyer questions instead of 10 and cut the off-site work entirely. Spend the first two hours choosing those questions and checking raw HTML and crawler responses, then use the remaining 14 hours for four answer pages of roughly 3.5 hours each. This version does not buy the same test coverage, so do not present it as proof of a citation lift."}}]}]}
```