CHALK LABS
AI search, SEO and GEO visibility
AEO / GEO / SEO Growth Proposal
Prepared for [Company name]
- Prepared for
- [Contact name]
- [Company name]: [website]
- Prepared by
- Rahil Jain, Founder
- Chalk Labs
[Month Year]
Executive summary
[Company name] sells [what you sell] to [who buys it]. Those buyers now ask ChatGPT, Gemini, Perplexity, Claude and Google AI Mode for recommendations before they open a website. The brands those assistants name and cite shape the shortlist.
We checked how [Company name] appears for the questions buyers ask. The findings are in the next section. In a finished proposal, this paragraph states the single most important finding in one or two plain sentences, for example where the company is visible and where it is not.
Chalk Labs proposes a three-month AI SEO programme covering AEO (answer engine optimisation), GEO (generative engine optimisation) and SEO. This document sets out the evidence, the plan month by month, and three ways to work with us.
About this template. Text in [square brackets] is a placeholder. Every finding and spot-check marked Example is there to show what a finished proposal looks like. None of it is a real reading.
Where [Company name] stands today
The findings below show how [website] and the [Company name] brand appear today. Each one carries a label that says how we know: Checked live means we fetched it ourselves, AI research means an AI model gathered it from public sources and a person reviews it, and Example means it is shown here only to illustrate.
Findings
| Finding | Status and source |
|---|---|
| AI crawlers can read the site Example: robots.txt allows OAI-SearchBot, PerplexityBot, Claude-SearchBot and Googlebot. In a finished proposal we list every bot we checked and what the file says about it. | GoodExample |
| No llms.txt file Example: nothing was found at [website]/llms.txt. It is a proposed convention, not a ranking factor, and no AI company has said it is required. We list it as a low-cost tidy-up. | GapExample |
| Structured data on content pages Example: the home page has Organization markup, but article and FAQ pages have none. Structured data helps machines read who wrote a page and what it answers. | GapExample |
| Pages built to answer buyer questions Example: the site has [number] pages and none is built around a question such as the best [your category] for [buyer role]. When no page answers a question, assistants cite whoever built one. | GapExample |
| Key text visible without JavaScript Example: we could not tell whether the main text appears in the raw HTML. We would test this in Month 1 because some crawlers do not run scripts. | UnknownExample |
| Authority compared with rankings Example: [X] referring domains but only [Y] ranking keywords. A strong link profile with few rankings means trust is not yet turning into traffic. In a finished proposal we show the third-party tool reading and name the tool and date. | GapExample |
Spot-checks in AI answers
A spot-check records what an AI assistant answered to a buyer-style prompt on the date shown. These three rows are examples. In a finished proposal each prompt is run and the answer is recorded. Not checked means we could not run it, not that the answer was negative.
| Prompt | Brand named | Site cited | Cited instead |
|---|---|---|---|
| What is the best [your category] for [buyer role]?Example row | No | No | [Competitor A], [Industry publication] |
| [Your product] vs [Competitor A]: which is better for [use case]?Example row | Yes | No | [Comparison website] |
| How do I [problem your product solves]?Example row | Not checked | Not checked | None recorded |
What your buyers ask
Buyers do not type one keyword. They ask a question, and the assistant then runs several related searches to build its answer. The searches below are examples of what an assistant may run for [Your market]. In Month 1 we record the real ones and turn each into a page.
What is the best [your category] for [buyer role] at a [company size] company?
Related searches behind it
How do I [problem your product solves] without [common pain]?
Related searches behind it
[Your product] vs [alternative]: which fits [use case]?
Related searches behind it
The page groups we would build
- What is [your category]?
- What is [key term in your field]?
- How to choose [your category]
- How to [problem] step by step
- [Your product] vs [Competitor A]
- Best [your category] for [use case]
- [Your category] for [industry]
- [Use case] with [your product]
- [Your category] in [Your market]
- [Your category] rules in [region]
What changes after this engagement
The aim is that [Company name] is found and cited for the questions its buyers ask, and that the visits it earns turn into enquiries. We measure progress in four ways and report each one every month, whether it is up or down.
- Pages. Every long-tail question we target gets its own page, built to answer it directly near the top.
- Visibility. [Company name] is named and cited more often in AI answers for the prompts in the spot-check table, tracked on ChatGPT, Perplexity, Gemini and Google AI Mode.
- Search. More impressions and clicks in Google Search Console, first for the long-tail set and later for the head terms, which move more slowly.
- Enquiries. Visits from these pages lead to calls and form fills that you can count.
- We report movement as it happens. We do not guarantee any ranking, citation or lead number.
What we are solving for
- Questions with no page built to answer them. When no page on [website] answers a buyer's question directly, Google and AI assistants cite whoever did build one.
- Authority that does not turn into visits. Links, press and reviews show that [Company name] is credible. If few pages target the questions buyers ask, that credibility does not become traffic. We check this by comparing link data with ranking data.
- No view of AI citations. Many teams track rankings but cannot say how often AI assistants name them. We set up a monthly check on the same prompts so change is visible.
- A thin set of indexed pages. Product pages and a blog are good trust assets, but a handful of URLs cannot cover the long tail of use-case, comparison and integration questions.
- No repeatable system. Without a routine for testing, measuring and rewriting pages, results stall. The three-month plan builds that routine.
How AI search (AEO and GEO) works
Traditional SEO tries to place one page among ten blue links. AEO (answer engine optimisation) and GEO (generative engine optimisation) aim at something different: being the source an AI assistant reads and names when it writes an answer. This is how that works, in the order it happens.
1. A person asks a full question
Buyers ask ChatGPT, Gemini, Perplexity, Claude or Google AI Mode in whole sentences, such as which option fits their situation. A question carries far more context than a short keyword, and it is rarely typed the same way twice.
2. The assistant splits it into searches
Instead of searching once, many AI search tools break the question into several narrower web searches and run them together: features, price, alternatives, reviews, how-to steps. Google calls this query fan-out in its Search Central documentation for AI Overviews and AI Mode. Other assistants do something similar, but the details differ and change often.
3. Pages are retrieved
Those searches return pages from a search index. A page can only be used if the crawler can fetch it, the page is indexed, and its text is readable. Not every crawler runs JavaScript, so the key text should be in the page's HTML. For Google's AI features, a page must be indexed and eligible to show in Search with a snippet.
4. Passages are chosen and an answer is written
The model reads the retrieved text and picks passages that answer the question clearly. Pages that give a direct answer near the top, use specific facts and have clear headings are easier to use than long pages that circle the point.
5. Sources are cited and brands are named
Many assistants show source links and name companies inside the answer. Being named and being cited are different results, and a brand can have one without the other. We track both.
6. What we do about it
We map the questions your buyers ask and the searches behind them. We build one page for each clear intent, make sure crawlers can read those pages, and re-run the same checks every month so you can see what moved.
The first pages we would build
Each page targets one clear question, so it can be found and cited for that question.
- DefinitionWhat is [your category]? A plain-English guideTarget query: what is [your category]
- How-toHow to choose [your category] for [use case]Target query: how to choose [your category]
- Comparison[Your product] vs [Competitor A]: which fits [buyer role]?Target query: [your product] vs [competitor a]
- Industry[Your category] for [industry]Target query: [your category] for [industry]
- Region[Your category] in [Your market]Target query: [your category] in [your market]
- Integration[Your product] with [integration]Target query: [your product] [integration] integration
- Use case[Use case]: how [Company name] solves itTarget query: [use case] with [your category]
- How-toHow to [problem your product solves]Target query: how to [problem]
Engagement structure
The programme runs for three months. Each month has one clear deliverable and builds on the month before, so progress adds up instead of restarting.
- Month 1Research & Build
Full brand and SEO audit, keyword and demand research, and 100 buyer queries ranked by intent. With Strategic Execution or Deep Execution we also write and publish the first batch of answer pages. With Strategic Consulting we deliver the audit and the three-month strategy for your team to run.
- Month 2Optimise & Convert Authority
Check how pages perform in Google and in AI answers. Fix on-page issues, schema and page structure. Use internal links to send the site's existing authority to the new pages.
- Month 3Rewrite & Scale
Rewrite the pages that underperform, using the Month 2 data. Set up the next batch of pages and carry forward what worked.
Payment and terms
- Strategic Consulting is a one-time fee. Strategic Execution and Deep Execution run as a three-month programme with monthly pre-payment; each month has a clear deliverable and builds on the last.
- On the monthly plans, you can pause or stop at the end of any month with no further pre-payment owed.
- Pricing is in USD. Currency conversion and local taxes are the client's responsibility.
- We commit to the work and report every result. We do not guarantee rankings, citations or lead counts, because search engines and AI models are third parties.
Package options
Three options, depending on how much of the work you want Chalk Labs to do. The one-time plan gives your team the audit and the plan to run. The monthly plans follow the three-month structure above.
Package 1: Strategic Consulting
$1,000 one-time
We hand you the plan. You execute.
What is included
- Full brand and SEO audit
- Keyword and demand research
- 100 real buyer queries, ranked by intent
- A three-month AI SEO strategy
- Brand onboarding and a knowledge base
- AI author humanization framework
Package 2: Strategic Execution
$3,000 per month
Everything in Consulting, and we do the work.
What is included
- Everything in Strategic Consulting
- Technical SEO, done for you
- AI-visibility audit across ChatGPT, Perplexity, Gemini and Google AI Mode
- 100 answer pages, one per buyer query, written and published
- Content refresh and optimisation
- 25 branded content placements
Package 3: Deep Execution
$4,000 per month
Everything in Execution, scaled up.
What is included
- Everything in Strategic Execution
- Advanced technical SEO
- 200 answer pages
- Continuous content expansion
- 40 branded content placements
Next steps
- Choose the package that fits your stage and budget.
- Review this proposal and confirm that you want to go ahead.
- Join a kickoff call to agree the priority use cases, markets and integrations for the first batch of pages.
This draft was prepared from public information. It is not a guarantee of rankings, AI citations or lead numbers. Results depend on your market, your competition and where you start. Search engines and AI models are run by third parties and change often. All prices are in US dollars.