Get your products found in ChatGPT, Perplexity and Google AI Overviews, without the snake oil
There is no secret trick. AI search reads the same web pages search engines do. We check what it can read about your company today, then build the product pages and facts it needs to cite you. Below is a real test we ran on 1 November 2026, misses included.
What actually helps
- A page per productSpecs as real text, not locked in a PDF
- Structured data that matches the pageProduct and FAQ JSON-LD, checked
- Cited by credible sitesDirectories, associations and trade press
Buyers now ask ChatGPT, Perplexity or Google's AI Mode which suppliers make a part, before they ever open a search result. If the answer names your competitor and not you, the reason is usually plain: the AI could not find clear, readable facts about your products, or no search index had them yet.
AI search optimization, also sold as GEO (generative engine optimization), is mostly ordinary work done carefully: pages a crawler can read, one per thing a buyer looks for, facts stated the same way everywhere, and other sites that mention you.
Krira Digital is a small software studio in India that builds catalog websites for manufacturers. We work with US companies remotely. We audit how AI search sees your company, tell you what is missing, and build the missing pages once. We don't sell monthly GEO packages or promise mentions.
What Google says you need (and what it says you don't)
Google's own page on AI features (updated 10 December 2025) says there are no additional requirements to appear in AI Overviews or AI Mode, and no special optimizations are needed. You don't need new machine-readable files, AI text files or special schema markup.
What it does recommend is ordinary good practice: let Google crawl your pages in robots.txt, put important content in text, and make sure your structured data matches the visible text. ChatGPT and Perplexity search the web too and show the pages they cite. So the work is the same everywhere: clear pages, one per thing a buyer looks for, that other credible sites link to.
Why Bing matters if you care about ChatGPT
OpenAI's help page on ChatGPT search says it sometimes uses outside search providers: it rewrites the user's question into search queries and sends them to those providers. Microsoft Bing is the provider most often reported. OpenAI does not say how much of each answer comes from Bing.
So a page that Bing has not indexed is hard for ChatGPT search to find. Many manufacturers check Google Search Console and never open Bing Webmaster Tools. That is a cheap gap to close.
- Google's guidance, updated 10 Dec 2025
- AI features and your website
- How ChatGPT search finds pages
- OpenAI help: ChatGPT search
- Catalog we built
- ptpcnctoolings.com: 1,432 model pages
- PTP's llms.txt
- ptpcnctoolings.com/llms.txt
- Our own llms.txt
- kriradigital.com/llms.txt
- Free check of your site
- Krira website checker
AI visibility audit checklist
| How to check it yourself | What PTP's pages show | |
|---|---|---|
| Crawler access | Open yoursite.com/robots.txt. Look for Disallow rules that block Googlebot, Bingbot, GPTBot, OAI-SearchBot or PerplexityBot. Check your firewall or CDN bot settings too. | All allowed; four AI crawlers named |
| Can it be indexed? | View the page source. Look for a robots meta tag with noindex, and a canonical link that points to the page itself. | index, follow; self canonical |
| Text before scripts | View source (not Inspect) and search for a spec value, such as a dimension. If it is missing, crawlers may not see it. | Full spec table in the HTML |
| A page per product | Pick five part numbers. Does each have its own address, with the code in the title and heading? | Yes: /p/<ordering-code>, code in title |
| Product structured data | Paste a product page into Google's Rich Results Test. Check that name, sku and specs match what is on the page. | Product JSON-LD, sku and mpn = code |
| Sitemap | Open your sitemap.xml. Count product URLs and compare with your catalog. | 1,432 of 1,432 models |
| In the index | Search site:yoursite.com plus a part number in Google and in Bing. Check Search Console and Bing Webmaster Tools for indexed pages. | Not found in our 1 Nov test |
| Outside mentions | Search your company name and a product in quotes. Who else names you: distributors, directories, trade associations? | Few found; this is PTP's main gap |
The free website checker at /check covers the basics in under a minute: security warning, phone layout, broken pages and Google title.
Why AI search skips most manufacturers
The fix list, in order
What you get from the audit
A short report in plain English: the questions we asked and the answers we got, with dates; a table like the one above for your own part numbers; the list of gaps on your site, in order of effort and value; and a fix plan with the pages to build. You can hand the plan to your own web team, or we build it.
The fixes are usually not exotic. A page per product family and per part number. Spec tables as text. An FAQ page that answers what buyers ask. Comparison pages for the choices buyers weigh. Structured data that matches the page. A clean sitemap, submitted to Google and Bing. If your catalog is still a PDF, this is where a catalog website comes in.
Proof: a tooling catalog built for search and AI
For PTP CNC Toolings, a tooling maker in India, we turned a printed catalog into a website of 1,432 models. Every model has its own page at its ordering code, with its full dimension table as text, an interactive drawing, Product and FAQPage JSON-LD, and a request-for-quote button. The site also has a sitemap and an llms.txt file that says what the company makes and where each model page lives.
We also built PTP a public MCP server, so ChatGPT, Claude and other assistants can look up its catalog directly when a user connects it. Our own site, kriradigital.com, has an llms.txt file and FAQPage JSON-LD on its pages. We add llms.txt because it is cheap, not because it is proven: Google says it doesn't need one. Our test above shows the limits plainly, and we won't claim results we can't show.
An AI-readable product catalog
GEO agency pricing in 2026: what agencies publish
| Published price | Source | |
|---|---|---|
| Agency retainers, small business | $1,500 to $5,000 a month | WebFX GEO pricing guide, dated 15 Oct 2026 |
| Agency retainers, mid-size to enterprise | $5,000 to $50,000+ a month | WebFX GEO pricing guide |
| WebFX's own GEO plan | From $3,000 a month | WebFX, same page |
| Mersel managed GEO | Growth $1,800 a month; Scale $3,200 a month; Enterprise custom | mersel.ai/pricing, read 1 Nov 2026 |
Prices as each firm publishes them, read on 1 November 2026. We don't publish a monthly GEO plan: we quote the audit and the fixes as fixed prices after a short call.
A monthly GEO retainer, or fix the site once?
| Monthly GEO retainer | Audit and fixes | |
|---|---|---|
| What you pay for | Ongoing reports and content | A fixed audit, then the pages that are missing |
| llms.txt and "AI schema" | Often sold as the main fix | Added if cheap; Google says neither is required |
| What changes on your site | Blog posts, often | Product pages, spec tables, FAQ pages and structured data |
| Guarantees | Sometimes promised | None. AI answers vary by user and by day |
| Who owns the result | Depends on the contract | You: pages and code in your name |
If an agency already gives you real pages and honest reports, keep it. Ask it what Google's AI features page says.
Why a one-time fix beats a monthly retainer for most manufacturers
A manufacturer's catalog changes slowly. Once every product has a readable page, correct structured data and a place in the sitemap, the technical work is done until the catalog changes. Paying every month for the same checks rarely adds much.
What does keep moving is outside: other sites mentioning you, and indexes picking up your pages. You can track that yourself with the three checks below, for free. If you later want help with trade-press articles or distributor pages, buy that as a fixed piece of work too.
How to measure progress
- 1Fixed prompts, same day each monthWrite ten buyer questions, such as "Which pull stud fits a BT40 MAS-BT spindle with through-coolant?" Ask them in ChatGPT, Perplexity and Google AI Mode, signed out. Record who is named and which pages are cited, with the date.
- 2GA4 AI referrersIn Google Analytics 4, add a report on session source for chatgpt.com, perplexity.ai, copilot.microsoft.com, gemini.google.com and claude.ai. ChatGPT often tags links with utm_source=chatgpt.com.
- 3Search Console and BingWatch indexed product pages and impressions for part numbers in Google Search Console. Google counts AI Overviews and AI Mode traffic in its normal Web report. Do the same in Bing Webmaster Tools.
How it runs
- 1Free first checkRun your site through our free checker, and send us five questions your buyers ask and ten part numbers.
- 2AuditWe check crawl access, text, structured data, index presence and coverage, and sample AI answers. You get the report, a test table like ours and a fix plan.
- 3FixesWe build the missing pages on your site, or a new catalog section, and re-run the same questions and codes after a few weeks.
Questions
What is GEO (generative engine optimization)?
It is a newer name for making your company easy for AI tools to find, read and cite. In practice it is mostly good SEO done carefully: crawlable pages, facts as text, structured data that matches the page, and mentions on other credible sites.
What is a GEO agency, and do manufacturers need one?
A GEO agency sells AI search optimization, usually as a monthly retainer. Most manufacturers need a one-time fix more: a page per product, specs as text, and their sitemap in Google and Bing. After that you can measure progress yourself.
How much does AI SEO cost?
Published agency prices run from about $1,500 to $50,000+ a month (WebFX, October 2026); Mersel lists managed GEO from $1,800 a month. We quote a fixed price for the audit and a separate fixed price for any pages we build.
How do I get my business cited by ChatGPT?
ChatGPT searches the web and cites pages that answer the question clearly. Give each product and service its own page with the facts as text, let crawlers in, make sure Bing has indexed you, and get named by credible sites in your industry. There is no form to submit and no paid shortcut.
Does ChatGPT use Bing?
OpenAI says ChatGPT search sometimes sends rewritten queries to outside search providers, and Bing is the one most often reported. OpenAI does not say how much each answer relies on it. Getting indexed in Bing is cheap, so do it.
How does Google decide what to show in AI Overviews?
Google says it uses the same systems as normal search, and there are no extra requirements to appear. Pages need to be indexed and eligible to show a snippet. Good, crawlable pages with clear text are what count.
Do I need an llms.txt file?
Not for Google: its AI features page says you don't need AI text files. Some AI tools may read llms.txt, and it costs little to add, so we put one on our own site and on the PTP catalog. Our 1 November test shows it is no shortcut: PTP has one, and its codes still did not show up.
Does schema markup help with AI search?
There is no special AI schema. Normal structured data, such as Product, Organization and FAQPage, helps search engines understand a page, as long as it matches the visible text. Markup on a thin page won't make it worth citing.
What does an AI visibility audit check?
Ours checks crawler access, what text a crawler sees, which products have their own pages, structured data, sitemap and index presence in Google and Bing, and how your company facts read across the web. We also ask fixed buyer questions in ChatGPT, Perplexity and Google AI Mode. Start with the free website checker.
How long does it take to show up in AI answers?
New pages must be crawled and indexed first, which usually takes days to weeks, and longer for a new site with few links. AI answers then change as sources change, and they vary by user. Plan on weeks to months, and measure with the same questions each month.
Can you guarantee my company will be mentioned by ChatGPT?
No. Nobody can. AI answers vary by user, wording and day, and no one outside OpenAI, Google or Perplexity controls them. We fix what is in your control: your pages, your data and how clearly your facts read.
Why can't ChatGPT find the products in our PDF catalog?
A PDF is one file for hundreds of products, often with specs as images. A buyer asks about one part, and there is no page for that part to cite. A page per model with specs as text fixes that, as on the PTP CNC Toolings catalog we built. See catalog websites for manufacturers, and for a direct route into ChatGPT and Claude, our MCP server for product catalogs.
Related pages
A fixed quote for the audit and the fixes
Run the free checker, then send us your site, five questions your buyers ask and ten part numbers. We reply with a fixed price for the audit, and a separate fixed price for any pages we build, in writing. Payment in stages.
Tell us what you need.
Website, catalogue, software or app. Message us and we reply within one working day.
Tap one. Your message is already written: just press send.