B2B · Packaging manufacturing

From a new domain to the most-cited site on ChatGPT: a B2B packaging manufacturer

An Indian maker of bottle closures for water and beverage brands launched a new website in mid-2026 with no search history, no backlinks and no presence in AI answers. Geology ran SEO and GEO together from the first week: an index cleanup, a reference library for the engineers and buyers who specify closures, product pages built on plain facts, and weekly tracking of ChatGPT and Google's AI Overviews. Four months later the site draws steady search demand, and ChatGPT names the client on 8 of 11 buyer questions and cites its site more than any other.

Client

About B2B packaging manufacturer

A B2B packaging manufacturer in India making metal and plastic closures for packaged drinking water, soft drinks and other beverages. Its customers are bottling plants and beverage brands that buy in bulk and qualify a closure supplier before switching.

A closure has to match the bottle's neck finish, hold the drink's pressure and show tampering, so the people who choose a supplier are often production and quality engineers, with purchasing signing the order. They research specifications long before they ask for quotes.

Challenge

The Challenge

The client came to Geology with a brand-new domain. Google had seen it for barely a week, it had no links pointing to it, and none of the 11 buyer questions we tracked produced an AI answer that named it. Every one of those answers named other suppliers.

The brief was simple to state and hard to do fast: make a new manufacturer findable by buyers who do not know its name yet, in Google search and in the AI answers those buyers now read first.

A new site cannot win on authority, so the plan had to win on usefulness. Closure buyers search for standards, sizes and specifications. A site that answers those questions clearly earns Google's trust page by page, and gives AI assistants concrete, quotable facts about who makes what.

Approach

Our approach

01

Week one: clear the leftovers out of the index

The new site launched on a purchased theme, and the theme had brought its demo content along. Of the 64 URLs in the sitemap, 49 were template leftovers, and Google had already indexed some of them, including demo blog posts. We removed the junk, submitted a clean sitemap in Search Console, and asked Google to index the core pages it had not yet found. On a domain with no history, every crawl spent on a demo page is a crawl the real catalogue does not get.

02

Build the reference library buyers actually use

Bottling plants and beverage brands do not search for a closure supplier by name. Their engineers and buyers look up cap size charts, neck-finish standards, how to measure a cap, which closure types suit which drink, and what a given closure format is. We wrote that reference layer first: explainers, size and specification sheets, and comparison pages, each one written to answer the question in its opening lines. Those reference pages earned the site's first clicks and still draw most of its non-branded demand.

03

Product and category pages that state facts plainly

Next came category pages for each closure family and product pages for each line, with the specifications a buyer checks before asking for samples: materials, liner options, tamper-evidence, pressure ratings, and the bottle types each closure fits. We added structured data so search engines and AI assistants could read the catalogue as a set of products, not a block of marketing copy.

04

Fix who AI assistants think makes the products

Our first AI check found a problem specific to this client. The names of its two flagship product lines are also used by unrelated companies abroad, so when a buyer asked who makes them, ChatGPT and Google pointed to those companies. We corrected the tracker so it stopped counting those answers as wins, then published pages that tie each product line to the manufacturer in plain language: who makes it, where, and to what standard, plus an about page and a plant page that AI answers could quote.

05

Weekly AI tracking on 11 cold buyer questions

Every week we ran the same 11 non-branded buyer questions through ChatGPT and Google AI Overviews, covering closure types, tamper-evident and liner options, bulk buying and regional sourcing. We recorded two things separately: whether the answer named the client, and whether it cited one of its pages. The split mattered. Google cited the site for weeks before ChatGPT named it once, which told us the content was readable and the missing piece was outside confirmation.

06

Outside confirmation from directories

Google's AI Overviews in this category lean hard on B2B directory and marketplace listings. We built and worked through a list of the directories Indian packaging buyers use, so the client's name, products and plant details appear consistently in the places the answers already trust.

Outcome

Results

90x+
Search impressions
15x+
Search clicks
80x+
Impressions on searches without the brand name
9x+
Distinct searches the site appears for
0 → 8 of 11
Buyer questions where ChatGPT names the client
0 → 2 of 9
Google AI Overviews that name the client

Search figures come from a fresh Google Search Console pull on the client's verified domain property. They compare the first four weeks the site appeared in Google (late June to late July 2026) with the latest four weeks (to mid-November 2026). All multiples are rounded down. Because the site was new, the starting point is close to zero, and the multiples show how far it has come rather than a lift on an established base.

In the first four weeks, one search put the site on page one of Google. In the latest four weeks, about three in four of the searches it appears for put it on page one. Six times as many of its pages now show up in search, and most of the growth in pages earning clicks came from the reference library.

AI figures come from the weekly tracking reports, comparing the first week measured on the final question set (late July 2026, 11 non-branded buyer questions) with the latest week (mid-November 2026). ChatGPT went from naming the client on 0 of 11 questions to 8 of 11, and it cites a page of the client's site on every answer that names it. The client's domain is now the site ChatGPT cites most across those questions, at twice the next. Google's AI Overviews, which lean on B2B directories in this category, went from naming the client in 0 of 11 Overviews to 2 of 9 (Google did not show an Overview on the other two questions). That second engine is where the directory work continues.

Get started

Launching or relaunching a manufacturer's site?

Geology specializes in insurance, financial services and B2B. A free audit shows how ChatGPT and Google's AI Overviews answer your buyers' questions today, and which pages they would need to cite you.