Case study: ProCoders — copy (EN)

IT outsourcing · custom development

ProCoders: zero to 2,217 Top 10 keywords in 41 months

Lead: Ahrefs showed exactly zero. Not “low” — zero traffic, zero positions, zero keywords. The largest project in the agency’s history: 973 tasks, 5,720 hours, a team of ten.

• 434 keywords in the Top 3, from nothing
• 30,632 keywords ranking, from nothing
• More than half the traffic non-branded

Small print under the bar:

Ahrefs data, March 2020 to July 2023. Ahrefs estimates traffic from its own index, so traffic figures are an estimate rather than a counter reading; positions are measured directly.

Our largest project by time invested. For comparison: EPAM Anywhere ran to 1,860 hours, Omegarender to 3,295.

What the client brought

ProCoders is a Ukrainian company supplying development teams on an outsourcing and outstaffing basis. The market is the US and the English-speaking world, where thousands of similar companies run near-identical pages.

March 2020. Ahrefs on the domain:

Why that matters

A flat zero in Ahrefs is rare. Usually there’s some tail — the brand name, a few accidental queries, leftovers of old content.

Here there was nothing. Not “weak positions” — an absence from organic search as a matter of fact.

Build an organic channel from nothing, in a market where nobody knows the client and every competitor uses the same words.

Which meant:

• work out what the company is actually good at before writing a word about it;
• build a site structure that didn’t exist;
• create content, a link profile and analytics from a blank page;
• count inquiries, not traffic.

We didn’t guess at directions. The project ran on our WEDGE framework, which answers one question: where should a client’s limited resource go to return the most.

W — Where the client can win

The first thing we produced wasn’t keyword research or an audit. It was a competence map: five tabs the client fills in themselves — solutions, verticals, languages and frameworks, technologies, services.

Each row carries three scores from 1 to 3: how easy this is to sell, how much the company earns on it, how the market is moving. The client scores the first two. We score the third.

What the ProCoders map showed:

And the services:

Three things became visible at once, and none of them from outside.

Front-end markup sold beautifully: over 10,000 projects in ten years, eight specialists on staff. But the market scored 1 — it’s narrowing. Building content there would have meant investing in a direction heading down.

Full outsourcing has a growing market — scored 3. And earnings of 1, with the note “very undesirable.” Without the map, the logical move would have been to go there.

The most interesting entry is a vertical no external analysis would find: distilling and breweries, 3/3/2. The client’s own justification in the map: “we understand production processes in detail, because we launched our own craft brewery in Kharkiv; we work closely with distillers in Kentucky.”

No niche analysis tells you that a Kharkiv software company should go into whiskey production. That’s the point of step W: the client knows things about themselves the agency cannot.

E — Evaluate whether we can get in

A second screen over the ranked list: competition in search, available traffic, whether the site as it stands can reach the direction at all.

At ProCoders the site was empty — Ahrefs showed zero. That cuts both ways: nothing to compete with, but also no old structure and no bad links in the way.

The output isn’t one direction but a queue. Markup and cost plus didn’t pass: falling market, low earnings. Travel and Entertainment scored 1/1/1 — not touched at all.

D — Decode who signs the contract

For the chosen directions we worked out who decides on outstaffing, what they already know, what worries them, what they need to see to take the next step.

Running alongside, in two parts, was a separate exercise — developing the company’s positioning.

The result went straight into the briefs: specifications were written for a specific person rather than for an algorithm.

Later, once ChatGPT existed, we built the project its own agent for pre-publication content QA. The persona stopped being a document and became a stage in the process.

G — Guide them to a decision

The content plan was built from the competence map, not from search volume. That distinction matters: one way you write about what you can sell, the other way about what gets googled.

And here the project produced something we’ve seen almost nowhere else. In 2022 two spreadsheets appeared:

• the pages that produced relevant inquiries;
• which funnel stage each lead was at, according to the content on the page.

So we knew more than “traffic went up.” We knew which page produced an inquiry, and how far along the decision the person leaving it was. After that, reoptimization stopped being guesswork: through 2022 and 2023 it runs monthly across three to six articles, chosen by contribution to inquiries.

E — Expand the coverage

The site structure was built on three axes: services → verticals → technologies. Each axis got its own keyword research, recommendations and briefs.

Then the intersections: not “outstaffing” alone and not “Node.js” alone, but the specific combination. Plus links: checking outreach links that had dropped off, chasing webmasters about removed ones — monthly, for years.

Not only organic. Google Ads, remarketing, a Facebook ad account, banner recommendations, Core Web Vitals, an internal linking scheme, an analysis of video queries. Across 41 months this wasn’t an SEO retainer. It was an outsourced marketing department.

Scale: 973 tasks and roughly 5,720 hours over 41 months, the largest project by time in the agency.

Pace: around 140 hours a month, peaking at 307.6 hours in December 2022. For comparison, a typical project of ours runs 40 to 60 hours a month.

A team of ten. The largest single share was 17%; the rest spread wide. At 300 hours a month there’s no other way to run it — content, links, technical work, ads and analytics all move in parallel.

Rhythm: a monthly cycle — project review, topic selection, keyword research, briefs, article review, implementation check, outreach, report.

The traffic peak we deliberately keep out of the headline

This part calls for honesty, because the numbers are public and easy to check.

The highest traffic on this project came not at the end of the work but in November 2022:

It reads like a decline. Then look at what that peak was made of: 60.7% of the November 2022 traffic was queries containing other companies’ names. It has the shape of a one-off spike from a single comparison page.

Positions, meanwhile, kept climbing the whole time: keywords doubled, Top 10 nearly doubled, and the non-branded share went from 38% to 52%.

That’s why positions are the headline metric here, not traffic. We could have put “10,834 visits” at the top and explained nothing. It would have looked better and been less honest.

A niche where every site says the same thing

IT outsourcing is a market where hundreds of companies describe themselves identically. You don’t escape that by producing more content. You escape it by writing about your own competences specifically, instead of about “quality development.”

Which is exactly why the project opened with a competence map rather than a keyword list.

Positions, Ahrefs:

Estimated traffic, Ahrefs: 0 → 6,961 visits a month.

Caption:

Per Ahrefs. Traffic there is an estimate from its own index; positions are measured directly. The November 2022 peak and what it consisted of are covered in “What was hard”.

Link profile: 995 referring domains, 5,023 backlinks.

Work delivered: 973 tasks, ~5,720 hours.

Inquiries. The project maintained a table of pages that produced relevant inquiries, and a table matching leads to funnel stage. The figures themselves are the client’s commercial information.

We chose a direction rather than accumulating volume. The competence map showed the investment belonged in outstaffing and the Node.js stack, not in markup, which sold more easily. The content plan followed what the company could sell.

A structure on three axes. Services, verticals, technologies. Each axis kept expanding for years and kept opening new entry points.

Inquiries tied to pages. We knew which page produced a lead and at what funnel stage. That turned reoptimization from a ritual into a decision.

Starting from zero is an advantage, not a sentence. No old mistakes, no bad links, no structure anyone is attached to. Zero to 2,217 Top 10 keywords in 41 months.

A content plan built on competences beats one built on search volume. The second gives you traffic. The first gives you inquiries.

Easiest to sell isn’t most profitable. Markup sold better than anything else here, but its market was narrowing. Outstaffing was harder to sell and paid more. You only see that when both axes sit in the same table.

Traffic and positions can move in opposite directions. November 2022 had the highest traffic here and half the positions of eight months later. Look at what the traffic is made of, not at how high the curve goes.

Links need maintaining. A standing process for spotting dropped links and chasing webmasters. Dull work, and without it the profile quietly melts.

Page-level inquiry attribution changes how you treat content. Once you can see which article produced a lead, reoptimization stops being a ritual.

The active phase ended in July 2023. In 2025 the client came back to working with us.

Sound familiar?

A site with no history, a competitive English-speaking market, a need for inquiries. We’ve built this channel from zero before.

Питання