Traffic for a niche B2B chemical manufacturer, in a global market of ~50 companies

Starting point — page 6 of Google, 300 visits a month

The usual argument for SEO is volume, and this project had none of it. The company makes metal-organic precursors, substances used to synthesize more complex chemical compounds, and runs laboratories in the US and Europe. The market is narrow enough that the end customers worldwide number 50 to 60: universities, or technology manufacturers of the Intel and Samsung sort. Not 50,000. 50.

The site was the main source of leads. The company had a pool of regular clients and almost no new ones arriving. Traffic before the work began ran at about 300 users a month, and the reason was position: for the main queries the site sat on page 6 of the results, while the competitors held the top and took everything that came with it.

The brief — product pages into the US top, traffic 10x

In a narrow B2B market, fast traffic growth is a secondary goal. Nobody is out there searching for a chemical compound with next-day home delivery. Position is what matters: the site has to be at the top for product and commercial queries so a buyer finds their aluminum chloride on the first try.

Diagnosis — 3 problems keeping the catalog off page one

The working principle is straightforward: look at whoever is already winning and do it better. My reference points were the chemical suppliers ranking in Google. I went through their sites, their structure and their traffic sources, and came away with 3 findings.

Their catalogs were better structured

Competitors offered advanced product search: by industry, by substance type. Easier for a visitor to use, and Google reads that ease as a ranking signal.

What I proposed: draw an ideal mental map of the site, then rework the structure wherever the map showed a gap.

The semantics were underworked

A chemical substance carries several names at once: the trivial one (sulfuric acid), the formula (H₂SO₄), and the CAS number (7664-93-9) that catalogs it under the American standard. Competitors had separate pages for CAS numbers and those pages were already at the top of Google. The client's equivalents were somewhere past page 7.

A second problem is typical of Western results. Alongside the other suppliers, informational platforms explaining the periodic table were sitting in slots a commercial site wants.

What I proposed: collect the semantics by hand for every product and fill the product pages with keywords covering each variant of the name. I also widened the page content, adding a safety section and links to related products.

The site had almost no links

Nobody had ever done link building, so the profile held only what had turned up on its own. Against these competitors, nowhere near enough.

What I proposed: a link-building strategy running on crowd marketing and outreach.

A fair number of small technical problems turned up too, of the kind that accompany any SEO project. They are worth fixing and are not worth writing about; what follows is what actually decided the result.

Work — manual semantics and a card template for 1,270 products

Why not push the new template across 1,270 pages at once? Because a wrong template applied 1,270 times is 1,270 pages to redo. I started on a single substance and worked the structure out there: collected its semantics, tested different placements of the keys in the tags and in the copy, and only once those pages reached the top-5 did I scale the pattern across the catalog.

Every name a buyer might use

The client supplied the product names, formulas and synonyms, and I parsed the competitors for the variants they carried and we didn't. Then I worked every card by hand, checking each of many thousands of names. There is no tool that does this part.

Here is how a buyer in the US might search for lithium bis(trimethylsilyl)amide:

Any one of those strings is a real query typed by a real chemist, and each was worth a place on the page.

A product-card template the client could repeat

Fitting all the naming variants in meant rebuilding the card itself. I sent the client's developers a page mockup and an implementation specification, and they worked through the catalog pages to that algorithm. What the card had to account for:

• URL — https://<site>/product-type/.
• Title — %Product type% — Products | %Brand%.
• Description — %Product type% products in the catalog.
• H1 — %Product type%.
• Structure — h1, then copy, then the product table.
• A guide to the copy: what it covers and how long it runs.
• A guide to the product table.
• Synonyms and CAS number.
• Safety information.
• Certificate of analysis.
• External identifiers.
• A “Download specification” button.
• A “Buy” button.

Scientific citations, and tags generated at scale

I picked external links to scientific papers for each product together with the client. Links to expert material inside a product card raise the page's authority for the crawlers, which is what E-A-T is asking for in a technical field.

With thousands of products in play, I generated the tags from a template carrying the naming variants and the formula:

After updating the tags I tracked every product card on its own: how it ranked, what position it held. Then I analyzed the results for the priority product queries separately, to push those pages higher still.

Then the loop ran for the rest of the engagement. The client prepared catalog pages to the template, by hand at first and then through a plugin once we automated it, and sent me 50 to 70 pages a month. For each batch I collected the semantics manually, made the corrections, tested meta tags and key placements, and chose the best variant. 1,270 pages went through that loop.

Work — links, including the ones student grants earned

Links still count in Google's ranking, and quality counts at least as much as number. The field made link building slow: I found every suitable site by hand against Ahrefs ratings, with the competitors' profiles as the guide.

One channel was specific to this client, and it is my favorite kind: the company already had it and nobody had thought to use it. Its customers include universities, and it funded study grants for students. The universities published those scholarship programs on their own sites, which Google trusts a great deal, with a link back.

Work — catalog navigation rebuilt around custom orders

Competitor analysis pointed at UX as the other growth point. Chemical catalogs are large and the names are punishing, so the search has to work. Competitors filtered by the industry a substance is applied in, and by its type: foams, alloys and so on.

My first proposal missed something specific about this client. Alongside finished products the company makes custom compounds to each customer's requirements. Competitors could offer a tidy search over a fixed list; this catalog needed 3 kinds of search at once: by substance, by requirement, by industry.

On recommendation the client implemented a compromise: a hub page organized by application area (protective coatings, microdevices, nanomaterials), plus a separate page linking the compounds the industry uses most often. A buyer could then reach a product without being confined to the shape of the existing catalog.

There was an ordinary catalog as well, and it brought an unexpected difficulty with it.

Work — making a JavaScript catalog visible to Google

The audit turned up a rendering fault. The general catalog pages ran on JavaScript, so the crawler saw no content and could index none of it. On my recommendation the client installed a prerender service, Prerender.io, to serve the search bot an HTML version of the page.

The prerender improved indexation, and I celebrated slightly early. Google could see the first page of the catalog and nothing behind it, 20 substances out of 1,270, because the pagination was dynamic and ran on the same JavaScript.

The solution was deliberately unglamorous: one static HTML page listing every product.

It passed internal weight through to the dynamic catalog pages and to the new CAS pages the client had run through the prerender. Together that made the catalog visible to Google.

Result — 90% of product pages on page one

Within a year the site reached about 2,000 visits a month, and by the end of the active period of work it hit the planned 2,500–3,000.

Monthly organic traffic rose 10x. 2,000 or 3,000 visits is a rounding error in most contexts; in a segment with 50 buyers it is the segment.

The result that mattered to the business is the ranking one: 90% of product pages now appear on the first page of the results, and frequently in the first 5.

Setting 2021 against 2020, the most striking movement was in impressions — which is to say in the site's visibility. Unique users rose 8x, and by the client's account it showed up in sales.

What transfers to another narrow market

In Western markets, even in the narrowest and most specialized field, buyers look for suppliers in Google. That makes search the strongest channel there is for reaching new clients, as long as the work goes deep into the specifics of the field.