Which facts does an AI assistant need before it will recommend a manufacturer?
It needs the facts a buyer would ask about, written where it can read them. A purchase manager asking Gemini for "Fe 500D TMT bar suppliers in eastern India who can deliver 500 tonnes a month" is setting a filter on grade, location and volume. If your site does not state all three, the assistant has no grounds to include you, however well you fit.
Google's guidance points the same way. Its guide to generative AI features asks for "valuable, non-commodity content", and its helpful content questions ask whether a page gives "a substantial, complete, or comprehensive description of the topic". For a factory, the specific facts of your plant are the part no competitor can copy.
How often are these facts missing from factory websites?
Often enough to matter. EbizIndia's published study of 30 Indian manufacturer websites, read as text in October 2026, found production capacity stated as a number on 14 of the 30. Approvals or vendor registrations were named on 6. A statement about export or countries supplied appeared on 15. Past projects or supplies were described in detail on only 2.
Plant location with town and state did better, at 25 of 30. So owners are willing to publish facts about their plant; the rest has mostly never been written up for the website.
What should each fact look like on the page?
Each fact should be a specific number or name in a sentence or table, not an adjective. The table below sets out what to state, with the weak version common on factory websites alongside a stronger one.
| Fact | Weak version | Plain-text version |
|---|---|---|
| Capacity | "Large-scale production facility" | "Installed capacity of 1,20,000 tonnes a year across two rolling mills" |
| Standards and approvals | "Quality certified" | "Made to IS 1786. BIS licence number listed below. ISO 9001:2015 certified." |
| Tolerances | "Precision engineered" | "Bore tolerance held to H7; face flatness within 0.02 mm" |
| Materials and grades | "High-quality raw materials" | "Bodies in WCB and CF8M; trims in SS 316" |
| Industries served | "Serving diverse sectors" | "Supplied to refineries, water treatment plants and sugar mills" |
| Export markets | "Global presence" | "Regular shipments to the UAE, Kenya and Bangladesh" |
| Order sizes | "Bulk orders welcome" | "Minimum order 25 tonnes; most orders between 100 and 1,000 tonnes" |
| Lead times | "Timely delivery" | "Standard sizes dispatched in 7 to 10 days; special sizes in 4 to 6 weeks" |
| Location | "Pan-India" | "Plant at Durgapur, West Bengal; depots in Kolkata and Bhubaneswar" |
The numbers in the right-hand column are examples, not benchmarks. Use your own, and use the same figure everywhere it appears: if the home page says 1,20,000 tonnes and the About page says 1 lakh, an assistant has two answers and no way to choose. That problem has its own guide, on spec numbers that disagree across your site.
What does a before-and-after product page look like?
Take a made-up product page for TMT bars. The before version is typical of many steel websites:
Before: "We are a leading manufacturer of premium quality TMT bars with superior strength and ductility. Our state-of-the-art plant ensures consistent quality. Available in all sizes. Contact us for the best price."
A buyer learns nothing they could not assume, and an assistant has nothing to quote. The after version says the same things as facts:
After: "Fe 500D TMT bars to IS 1786, in diameters from 8 mm to 32 mm, supplied in 12 m lengths or cut to order. Rolled at our Durgapur plant, installed capacity 1,20,000 tonnes a year. Minimum order 25 tonnes. Standard diameters dispatched in 7 to 10 days across West Bengal, Odisha and Jharkhand. Mill test certificate supplied with every consignment. Used in high-rise residential, bridge and industrial building projects."
That paragraph answers grade, standard, size range, location, capacity, order size, lead time, delivery area and use. Each answer is a sentence an assistant can lift and repeat with confidence. Nothing in it is confidential, and none of it needs a designer.
Where do these facts usually hide, and how do I get them out?
They hide in the brochure, the certificate scans and the sales team's heads. Typically the capacity figure exists only in a PDF, the approvals only as logos, and the lead times only in quotation emails. Each needs moving into page text.
- Brochure facts: copy every number from the PDF catalogue onto the matching web page. Why ChatGPT struggles with your PDF catalogue explains what PDFs hide.
- Certificate logos: type the standard, certificate number, issuing body and validity date beside each logo or scan.
- Sales knowledge: ask your sales head for the five questions buyers ask on every call. The answers are your missing facts.
- Plant facts: list main machines, testing equipment and shift pattern on a Plant or Infrastructure page.
Why these facts carry so much weight is explained in your website now has two visitors: the assistant reads first, and the buyer arrives only if it found enough to recommend you.
A checklist to hand to whoever edits your site
Give this list to your developer or the person who updates the website. Each line is a sentence or table row that should exist in visible page text, not only in an image or a download.
- Installed capacity, as a number with a unit and period, on the home page and Plant page.
- Every standard you make to and every approval or vendor registration you hold, by name and number.
- Tolerances you routinely hold, per product line.
- Materials and grades, per product.
- Size, pressure, weight or rating ranges, as an HTML table.
- Industries you supply, named.
- Countries you export to, named.
- Minimum and typical order sizes.
- Lead times for standard and special items.
- Plant address with town and state, and any depots or branch offices.
Once the facts are on the page, structured data can label them for search engines, but it cannot stand in for them. For more on how assistants pick between suppliers once they can read you, see how AI assistants decide which companies to recommend. EbizIndia also explains what buyers assume when capacity is missing from a website, and the AI Visibility section collects the rest of this series.
Questions founders ask
Is it safe to publish my production capacity?
For most manufacturers, yes. Competitors usually know your rough size already, and a buyer who cannot find a capacity figure may assume you are smaller. Publish installed capacity as a round annual or monthly figure; you do not need to share current utilisation or order book.
We make to drawing. How do I state tolerances and sizes without a fixed catalogue?
State the envelope you can work within: the size range, weight range, materials and the tolerance you routinely hold, for example a machining tolerance in microns or a casting weight range in kilograms. Buyers with custom parts mainly want to know whether their job falls inside what your plant handles.
Should these facts go on the home page or on product pages?
Both, at different depths. The home page and an About or Plant page should carry company-level facts such as capacity, approvals, location and export markets. Each product page should carry its own sizes, grades, standards, tolerances and lead time, so an assistant answering a product question finds the answer on the page it reads.
Will adding structured data do the same job?
No. Structured data helps search engines label facts that are already on the page; it does not replace them. Google says no special schema is needed for its AI features. Write the facts in visible text first, then mark them up as ordinary housekeeping.