AI Visibility Fix Kit

A downloadable workbook and practical guides for checking whether AI search experiences mention or link to your business. Record a baseline, compare competitors and prioritise website improvements using the evidence you collect.

Get the Fix Kit See what is inside

$9.89, introductory price. Instant download from Gumroad.

Inside the actual workbook: choose ten questions from 100 prompt templates and replace the bracketed fields with your business details. Open the preview to inspect it.

What one baseline records.

Test rows for one baseline run

Choose ten buyer questions and run each across five agreed AI search experiences. Save the responses and sources, and record unavailable answers separately. Repeat the full set in a new run if you want a second sample.

100

Buyer-intent prompt templates to choose from and adapt manually

Default rows: ChatGPT Search, Perplexity, Gemini, Google AI Overviews / AI Mode and Claude

Business covered by the single-business licence

What you get

The download contains a nine-sheet Excel workbook, a start guide, an AI visibility playbook, website copy templates in Word and PDF, and a schema starter pack. You run the checks manually; the files do not query assistants or edit your site automatically.

Prompt library

One hundred buyer-intent prompt templates. Choose ten relevant questions and replace the bracketed fields with your own category, service and market.

Test log

One row per check across all five platforms, recording whether you were mentioned, whether you were linked, which competitors were named, and which sources the answer drew from. That last column is the one most people skip and the one that tells you the most.

Competitor comparison

Records appearances for up to three comparison businesses. Treat the results as observations from your chosen prompts and conditions, not a universal ranking.

Dashboard and action plan

A dashboard, readiness checks and a 30-day plan organise the observations. Any score is an internal workbook measure, not an official platform score or proof that a change caused a result.

File formats, use and support

The package is a ZIP download containing the Excel workbook (.xlsx), PDF guides, Word (.docx) and PDF copy templates, and JSON-LD schema examples. The workbook has nine sheets covering setup, prompts, test results, technical checks, website content, competitors, a dashboard and a 30-day plan.

The licence covers your own business or internal team and allows you to edit the templates for that business. Resale and redistribution are not included. Use suitable spreadsheet, PDF and Word-compatible software; if you import the workbook into another application, check its formulas and formatting before relying on the results.

You need access to the AI experiences you choose to test. Their availability, usage limits and any subscription charges are separate from this download. Support and the existing refund offer are available through hello@pixelpulsehq.com.

Created by Pixel Pulse Media. Read the worked example with clearly labelled sample data before interpreting a score.

What your result will look like

Record what the answer actually shows, then investigate plausible explanations. A missing mention does not by itself establish a technical fault or its cause. Record unavailable or untriggered answers separately.

Named and linked

You are in the answer layer for this question. Protect it, and track whether it holds.

Named, not linked

A mention and a website citation are different observations. Check which sources were used and whether your own page provides useful supporting detail.

Not named, competitors are

Review the cited evidence and the fit between your service and the question before choosing a change.

Nobody named, generic answer

The response may not be making recommendations. Review the question and context; this does not prove a lack of competitors or authority.

Described inaccurately

Check the cited information and correct obsolete or conflicting facts on the pages and profiles you control.

Four areas to investigate

Use these checks to investigate obstacles supported by evidence. They are not an exhaustive explanation of how an AI system selects an answer.

Crawlers cannot reach you

Check robots.txt, HTTP responses and firewall rules for the relevant search agent. Search and training controls are separate; a robots.txt allowance does not prove a crawler received the page.
Nothing states what you are

Check the business name, service descriptions, contact details and location. Structured data should match the visible page, but special AI schema is not a requirement or a guarantee of selection.
No plain answer on the page

Give readers a direct answer to the question, followed by scope, process, prices and supporting evidence. Use descriptive headings and useful internal links.
Nothing off-site confirms you

Look for accurate, relevant information in profiles and independent sources. Correct contradictions and earn useful coverage through real work; a directory listing does not guarantee a citation.

If the fixes turn out to be technical

Some of what the workbook surfaces is a ten minute change. Some of it is server configuration, schema across a whole site, or a content rebuild, and if nobody on your side does that work, the list will sit there.

If you would rather have the audit completed for you, we can scope it separately after reviewing your market, competitors and current website. You will receive the deliverables and commercial terms before deciding whether to proceed.

Run the kit first though. Get your score, and then you will know whether you need me at all. Plenty of people will not.

See what the audit covers

Where this came from

I run Pixel Pulse Media, a performance marketing practice working with clients in the UK, Europe and the Gulf. The Fix Kit was not invented as a product. It is the workbook I built for client work, cleaned up so you can run it yourself without paying me to do it.

SULECO
Eduvoyant
Corsnet
DJI Sri Lanka
Topcon
Dasis Aerospace

Brands I have worked with. They have not endorsed this kit. It simply came out of the work.

If it finds nothing, I will refund you

Run the full baseline, all fifty checks. If it does not surface at least one fixable reason you are being left out, email me and I will refund it. The kit costs less than lunch. What I am actually asking for is an afternoon of your time, so that is what the promise covers.

Honestly, who this is for

Buy it if

You sell something people ask for recommendations about
You can edit your own site, or you can ask someone who can
You have never checked whether assistants mention you
You would rather find out yourself than commission a report

Skip it if

You have no website you are able to change
You already track citation share month on month
You want someone else to do the work, in which case book the audit
You will not spend an afternoon on it, because unread it will not help you

Common questions

No. The kit records a sample and helps you investigate website gaps. It cannot prove why a system omitted your business, make the fixes automatically or guarantee organic placement. Improve the evidence and access you control, then repeat the checks.

Not to run the test. You paste prompts into five assistants and record what comes back, and the workbook scores it. Some of the fixes need site access, such as editing robots.txt or adding schema. If that is not you, the result gives you a list precise enough to hand to whoever maintains your site.

An afternoon for the first run, because fifty checks means ten questions asked in five places and the answers take a moment to read properly. Repeat runs are faster since the prompts are already written for your business.

Yes. The prompt library builds itself from your category and location, so it works in any market the assistants serve. Answers are geo-influenced, which is why the workbook asks you to test from the location your buyers are actually in rather than your own.

They overlap heavily but the target differs. SEO competes for a position in a list of links. This competes to be the source a generated answer draws from and names. A page can rank third and never be cited, and a page can be cited without ranking on the first page at all.

Run the full baseline before changing the site. Ten questions across five experiences makes 50 checks. A second full run adds another 50 responses, for 100 across two runs. Save each run separately and compare the same conditions; a single answer cannot establish a trend.

Covered above, and it is a promise from me rather than a policy page. Email hello@pixelpulsehq.com, tell me you ran the baseline and it turned up nothing, and I will send the money back.

Find out where you stand today.

Complete a documented baseline, save the responses and identify the next checks your website needs. Compare repeated runs using the same questions and conditions.

Get the Fix Kit Read the method first

$9.89, introductory price. Refunded if it finds nothing.