Case study: CYLIND — copy (EN)

3D visualization · English-speaking markets

CYLIND: 2 to 774 Top 10 keywords in 25 months

Lead: A site search barely noticed: 18 ranking keywords, 57 visits a month. The keyword research had already been bought before we arrived. We didn’t redo it. We rebuilt it.

• 397 keywords in the Top 3, up from one
• 795 keywords ranking, up from 18
• 97.9% of traffic is non-branded

Small print under the bar:

Ahrefs data, July 2024 to August 2026. Ahrefs estimates traffic from its own index, so traffic figures are an estimate rather than a counter reading; positions are measured directly. The curve for this window is visible in Ahrefs to anyone with an account — go check it.

What the client brought

CYLIND produces 3D visualization: architectural renders, interiors, virtual staging, walkthrough animation. The market is English-speaking, concentrated in cities where property actually gets built and sold.

In July 2024 the site had 18 ranking keywords and 57 visits a month. Two of those keywords sat in the Top 10. One sat in the Top 3.

What was already done

This one is unusual: the client didn’t arrive empty-handed.

• The keyword research was already bought and paid for. We didn’t start our own from scratch. We re-clustered what existed against the actual shape of demand.
• Analytics, Search Console and Tag Manager were already running. No first month spent setting up measurement.
• Verticals and ICP were documented.

What was missing

A structure that turned all that research into pages. And any content that would lead people to those pages.

One more condition: the working language of the project was Russian while every published word was English. Every brief passed through a double language filter.

The client needed inquiries for visualization work — from developers, architecture practices and real estate agencies.

Which meant:

• turn existing research into a working site structure;
• win search results in specific cities rather than in general;
• build domain authority from nothing;
• do all of it without throwing away what the client had already paid for.

We also agreed a RACI matrix up front — who is responsible for what, on both sides. Unusual for a project this size, and it removed most of the later “whose job was that” conversations before they could happen.

The client didn’t arrive empty-handed: the keyword research was already bought and paid for. The question was different — where, out of all that, should a limited resource go. That’s what our WEDGE framework answers.

W — Where the client can win

The competence map was filled in at the start: services — architectural renders, interiors, virtual staging, walkthrough animation; verticals — developers, architecture practices, real estate agencies.

Each row carries the client’s scores for sellability and earnings, and our score for market movement.

We don’t dispute the client’s scores. How strong the studio is at virtual staging is the studio’s own judgement. We bring what they don’t have: where the competitors already stand and whether the direction is reachable.

E — Evaluate whether we can get in

The second screen produced the project’s defining decision.

“3d rendering” is a wall. Large studios with decades of authority, plus the marketplaces. A site with 18 ranking keywords isn’t getting in there in a year, or two.

“3d architectural rendering in Los Angeles” is another matter. A specific city, a specific market, an entirely different competitive set.

So the output wasn’t “promote 3D rendering” but a queue of cities where property is actively built and sold.

7.1. Rebuilt someone else’s research instead of starting over

The easy move would have been to say the research was wrong, start fresh, and bill for it. We didn’t.

The client had already paid for that work. We took it and re-clustered it: regrouped queries around actual user intent and around the pages that were worth building.

What came of it: the project started with work rather than with re-spending a budget on something already bought.

D — Decode who signs the contract

Visualization gets commissioned by different people with different fears: a developer looks at timelines and volume, an architecture practice at image quality, a real estate agent at speed and price.

Personas don’t sit in a document. The project has two GPT agents: one adjusts the brief to the persona before it reaches a writer, the other checks the finished text against the marketing personas. This is how we work on every project since ChatGPT appeared.

Client meetings run to a written agenda, and each one covers not just positions but inquiries: how many, from where, how many proposals went out, and why the lost ones were lost.

G — Guide them to a decision

Proofreading as its own step. Every month the project carries a separate Proofreading task — not an SEO check, a language pass over the English copy.

The content is English, the team works in Russian. Without a dedicated step you get text that is technically correct and still reads wrong.

E — Expand the coverage

A geo intersection. We built a network of geo pages and articles for cities where property is actively built and sold: Toronto, New York, Los Angeles, Chicago, San Francisco, Las Vegas, Dallas, Florida, London, Sydney, Dubai.

Then the service clusters: virtual staging, exterior rendering, walkthrough animation, hotel rendering, interiors.

What came of it: those 397 Top 3 keywords are mostly city and narrow-service phrasings, where winning was possible, rather than a head-on fight with the whole market.

Measuring what most people don’t. We built the client a Looker Studio dashboard reporting monthly on organic traffic and key events from GA4, clicks, impressions and average position from Search Console, and LLM traffic as its own block.

The team: six specialists and a project manager, with hours spread evenly. No stage depended on a single person.

The rhythm: a monthly cycle — project review → content plan topics → keyword research → briefs → SEO review of the copy → marketing review → proofreading → implementation check → internal linking recommendations → outreach → next month’s plan → report.

Client meetings run to a written agenda. Each one covers not just positions but inquiries: how many, from where, how many proposals went out, and why the lost ones were lost.

We told the client the extra money wasn’t needed

In August 2026 we discussed raising the outreach budget by $200. Our position, recorded in the meeting notes: there is no pressing need for it right now — let’s revisit later.

An agency turning down extra money looks odd for about as long as it takes to look at it from the client’s side.

SEO lands. Sales doesn’t.

This is the uncomfortable part, and it sits in the meeting notes in plain words: the hypothesis that the product misses what those clients expect — SEO lands, sales doesn’t.

Traffic is right. Inquiries arrive. Then the deal doesn’t close, because what the client sells isn’t quite what the market we brought expects.

What we’re doing about it:

• a two-month window to test the hypothesis;
• an experiment with product positioning — shifting the presentation from “projects for construction approval” to “projects for sale”;
• a request for the client’s rejection data, with the actual message texts, so we work from reasons rather than guesses.

We could have left this out. But a case study where SEO “delivered inquiries” and stops there explains nothing about how the work runs.

Results aren’t linear

Also recorded in the meeting notes: the non-linearity of effort against outcome. The month with the heaviest workload isn’t necessarily the month with the biggest gain, and the reverse happens too.

Not an excuse. A working reality that is better agreed on at the start.

Part of the site’s authority was switched off

The /gallery/topworks page used to rank for core commercial queries but was closed to indexing by the time we analysed it. Its authority was doing nothing for anyone.

Positions, Ahrefs:

Estimated traffic, Ahrefs: 57 → 5,344 a month.

Caption:

Per Ahrefs. Traffic there is an estimate from its own index; the position figures above are measured directly.

Link profile: 307 dofollow domains, 723 dofollow links.

Inquiries. Conversions are tracked in GA4 and then in the client’s CRM, broken out by project type and by reason for loss. As of August 2026, 19 key events a month from organic. Deal-level figures are the client’s commercial information.

Work delivered: 378 tasks, 1,348 hours, 107 of those tasks content and copy.

We didn’t redo what was already paid for. The client’s keyword research got re-clustered, not replaced. The budget went into structure and content instead of a second study.

The direction came from the second screen, not from taste. “3d rendering” is a wall of large studios. Eleven cities with their own pages are real markets with visible competition. That’s what step E produces: not one direction but a queue of reachable ones.

Responsibility agreed on day one. A RACI matrix removed most “whose job was that” questions before they could come up.

Someone else’s keyword research is an asset, not garbage. Before you rebuild it, check whether regrouping it works. It often does, and it saves the client a month and a budget line.

In visual niches, cities beat categories. “3d rendering” is a wall. “3d architectural rendering in Sydney” is a workable query with competition you can see.

Non-branded share matters more than total traffic. Going from 1.8% to 97.9% means the traffic was earned rather than arriving on the company name. Without that figure, any multiple proves nothing.

Traffic can be right while sales are wrong. If inquiries come in and deals don’t close, the channel may not be the problem. Worth testing rather than blaming traffic quality.

English copy needs its own proofreading step. Not editing, not an SEO check — a language pass, as its own stage with its own hours.

The project has been on hold since August 2026. Before the pause we had agreed the next growth points: re-optimizing the home page and Portfolio, strengthening the Toronto and Dallas pages, building a Calgary page, and widening the content plan into adjacent niches — senior living visualization among them, where demand exists and competition is thin.

Sound familiar?

A narrow visual niche, an English-speaking market, a need for inquiries. We’ve run this before.

Питання