Case study: OroCommerce — copy (EN)

B2B eCommerce platform · product company

OroCommerce: 2.55× the traffic on fewer ranking keywords

Lead: An unusual kind of case. The client didn’t arrive at zero — they arrived with a large, established site: 24,905 keywords and 2,720 visits. Over four years we doubled the traffic and reduced the number of ranking keywords.

• Top 10 keywords: 1,009 → 1,586
• Positions 4–10: 631 → 1,141, the biggest shift
• Hundreds of contacts a month, more than half from organic search

Small print under the bar:

Ahrefs data, October 2019 to September 2023. Ahrefs estimates traffic from its own index, so traffic figures are an estimate rather than a counter reading; positions are measured directly.

Our second-largest project after ProCoders. Across 48 months, only two came in under 40 hours.

What the client brought

OroCommerce is a B2B eCommerce platform. A product company, not an agency: documentation, a glossary, whitepapers, comparison pages, years of blog archive, and a site written by many hands over a long time.

October 2019:

Where the difficulty was

This is not an empty site. You can’t walk in and start writing articles — everything has been written already. Twenty-five thousand ranking keywords means the site is enormous, and most of those keywords deliver nothing.

Projects like this are harder than starting from zero. Zero has no inertia.

Grow organic traffic on a large product-company site, without the option of simply adding content, because there was already too much of it.

Which meant:

• lift what already ranks close to the top;
• remove what pays nothing back;
• work with documentation, the glossary and comparison pages, not just the blog;
• open European markets with language versions.

On a site with 25,000 ranking keywords the question isn’t what to write. It’s what to touch at all. Our WEDGE framework answered it.

W — Where the client can win

A competence map for a product company looks different from one for an agency: instead of languages and frameworks, the products themselves and the directions inside them. OroCommerce, OroCRM, OroMarketplace — each with its own audience and its own market.

The map is adaptive: tabs that make no sense for a given business simply don’t get filled in.

E — Evaluate whether we can get in

Here the second screen worked differently from usual. The site was large and old — 24,905 ranking keywords. The problem wasn’t getting into a direction; it was that most of those keywords delivered nothing.

The tracker states it plainly: “Prioritizing pages for reoptimization based on niche analysis and analytics data.”

So the screen was applied not to new directions but to pages that already existed: which deserve the resource and which don’t.

The result looks like failure until it’s explained. Ranking keywords went down on purpose: 24,905 → 20,627. Minus 3,254 at position 51+. That isn’t a loss. It’s a cleanup.

Traffic doubled over the same period.

D — Decode who signs the contract

Step D is recorded here month by month, in a task named “Build the persona and JTBD” — March, April, May, June, July.

And separately: “Prepare the CTO (CIO) persona.”

For a B2B platform that’s the decisive role: replacing an eCommerce system isn’t a marketing decision.

G — Guide them to a decision

The tracker has a task that describes step G better than any explanation could: “Blog analysis — finding articles for reoptimization, establishing where each topic sits in the funnel.”

Every article got a place in the funnel, and what happened to it next followed from that.

Plus an assessment of a topic’s potential before committing to it, as its own task. Cheaper than writing the article and finding out nobody wanted it.

E — Expand the coverage

The most interesting part of this project is where we expanded.

Other people’s roundups, not our own pages. For four straight years we negotiated with editors at third-party publications for guest and sponsored placements, so that OroCommerce would appear in other people’s comparisons of B2B platforms.

The work ran monthly and split by region: separate target research and editor outreach for the US, for the UK, and for the French region. Without a break from 2019 to 2023.

Why that matters now. In 2025–2026 it became clear that AI answers lean on exactly this. A “best B2B eCommerce platforms” piece on an independent publication carries more weight than a company’s own page saying the same thing.

And the inquiries bear it out: the roundup “9 Best B2B eCommerce Platforms with Comparison Table”, our own comparison page and the content built around the independent Gartner ranking were three steady sources of contacts across the whole period.

Language versions as their own axis. In 2021, recommendations for the German and French regions, then verification that the language versions were actually implemented, as a recurring monthly task. A language version that ships without verification works at half strength.

Documentation and the glossary. At a product company traffic doesn’t live only in the blog: a separate audit of doc.oroinc.com, a glossary with rewritten terms, whitepapers, comparison pages, the CTAs on commercial pages.

Link hygiene. Alongside the building: link profile analysis with a disavow list, hunting down low-quality links, cleaning the target list, tier-2 work, and HARO — answering journalist queries to earn media mentions instead of buying links.

Volume: 851 tasks and 4,587 hours across 48 months — our second-largest project.

Consistency: around 95 hours a month for four straight years. Only two months in the whole engagement dropped below 40 hours. The peak was 154.9 hours in March 2022.

Team: nine specialists, the largest share at 24%. On a site this size there’s no alternative — technical work, content, links and language versions all run in parallel.

The peak came before the finish

The highest traffic on this project was around 11,000 visits in spring 2023, a few months before the work ended. By September 2023 it was 6,946.

We don’t hide that, because the chart is public. We also don’t build the headline on it: fluctuation on a site this size is normal, and the durable result of four years shows in positions rather than in one spring’s peak.

Large sites resist change

On a site with 25,000 ranking keywords, every change touches something else. That’s why the project carries so many tasks for verifying implementations and analysing redirects — more than for producing the recommendations themselves.

That’s the defining difference when working on a legacy site. Writing a recommendation is quick. Confirming it didn’t break the neighbouring section takes time.

A product site is harder than a services site

A services company has few pages. A product company has documentation, a glossary, whitepapers, competitor comparisons, integrations, release notes. Each page type follows its own rules and needs its own treatment.

Positions, Ahrefs:

Estimated traffic, Ahrefs: 2,720 → 6,946 visits a month, 2.55×.

Caption:

Per Ahrefs. The falling total keyword count alongside rising traffic is the result of a deliberate cleanup, explained under “What we did”.

Inquiries. In the final stretch of the engagement the site was bringing in hundreds of contacts a month, more than half of them from organic search. The flow was steady: no spikes, no troughs, across six months.

What matters most is the shape of those contacts:

Nearly half of all contacts are demo requests. For a SaaS platform that’s the most valuable conversion there is: someone wants to see the product, not read about it.

Link profile: 2,041 referring domains, 23,795 backlinks — the largest profile across our projects.

Work delivered: 851 tasks, 4,587 hours.

The resource went where it paid back. Pages were prioritized by niche analysis and analytics data rather than by traffic potential. Positions 4–10 grew 81% — which is where most of the gain lives.

We cleaned without sentiment. Minus 4,278 ranking keywords, minus 3,254 at position 51+. On a large site, removing matters as much as adding.

We worked every page type. Documentation, glossary, whitepapers, comparisons, integrations — not just the blog.

Fewer keywords isn’t automatically worse. Here the count fell 17% while traffic rose 2.5×. On a large site, ranking keyword count isn’t a quality metric.

Most traffic gains live at positions 4–10. It’s the cheapest movement available: lifting a page from eighth to third beats launching a new one from nothing.

Assess a topic’s potential before committing to it. On this project that was its own task. Cheaper than writing the article and finding out nobody wanted it.

Presence in other people’s roundups is an asset that matures slowly. We were placing OroCommerce into third-party comparisons from 2019, region by region. Those are the pages AI answers now cite. Anyone starting this in 2026 is queuing behind everyone who started in 2019.

HARO works where links can’t be bought. Answering journalist queries earns mentions that guest posting never produces.

Verify a language version after it ships. A recurring monthly implementation check isn’t a formality.

The last full working month was September 2023; the project closed formally in January 2024.

The reason was straightforward: the client needed a US-based contractor. There were no complaints about the work. Later changes to the site were made without our involvement.

Sound familiar?

A large site with history, a lot of pages, and traffic that won’t grow despite all that content. That’s our profile — we ran exactly this for four years.

Питання