03 / 06 Idaproject website
Role
Product designer: case catalogue, filters, clients, home page, products; built the products page in code myself
Team
brand designer, senior designers, art director, founders, developer, CDO
Company
Idaproject, 2026

An agency website that shows the work, not just a client list

Idaproject works with 64% of the top 100 residential developers in Russia, but its old site did not show a single project. We built the site around the work: a project catalogue with filters, client pages with numbers, and a product ecosystem I built in code. In the first months after launch, visit-to-lead conversion grew 28% while traffic fell 19%.

Before

The old site was one long page: big numbers, products as banners and clients as rows with years of work, tasks and services. Not a single screen or project. The company had also grown from a contractor into a set of its own connected products, with nowhere to show them as a system.

Research before design

  • I asked colleagues about previous attempts to redesign the site and why they failed.
  • I looked at who comes to us and what objections they have. Visitors are marketers and brand managers, who care about visuals and metrics, and commercial or IT directors, who care about timelines, budget and process. Common objections to the old site: no case studies, only similar developer sites, no clear support after launch.
  • With the brand designer I went through the brand system: principles, photo style, layout and the core metaphor.

Each decision below answers one of these objections.

Key decisions

1. A project catalogue with a fact in every title

The catalogue came first. We set the grid: large projects alternate with rows of smaller ones. Each card title carries a fact from the project, such as "+29% to unit price" or "2× qualified leads", so visitors see results instead of logos. We went through several rounds with senior designers, the art director and the founders.

2. Filter by project type

Clients look for what they need: a developer website, an agent portal, a buyer portal, a manager dashboard. So the main filter is project type, and the second is the client, out of 280. This answers the "only similar websites" objection directly.

Next to the filter, one switch moves between clients, cases and products. All three sections share one template: title, filter, list. Visitors see the same work from three angles without losing their place.

3. Clients with numbers and a project strip

Each client has a logo, a short description, years of work together, number of tasks and products, and a strip of projects. Projects come in two kinds: a full case study to read, or a preview that links to the live site, to a request form, or is marked as private under NDA. Visitors see the full scale of work even when a case cannot be shown.

Every project needed a cover. By launch, a colleague and I had designed about 200 covers in one to two months, each on brand and each different from the next.

4. The home page alternates clients and products

After the first screen come a few clients, then product slides, then clients again, then products. The home page shows both experience and the ecosystem.

5. A product screen with a lot of information and no overload

One screen had to fit visuals, name, key metrics, benefits, description and buttons. I put the product visual in the center and a list of benefits on the right, linked to it: pick a benefit and the visual changes. Metrics sit under the name instead of getting lost in the text.

New products keep appearing, and large screens will stop fitting at some point. So there is a second view: a compact card grid that shows the whole line-up at once.

6. One hour of vibe coding instead of 4 days of development

The products are connected and sell each other, so they orbit a shared center. I used to hand the developer After Effects animations, and he had to guess the parameters. This time I vibe-coded the whole page in an hour: the developer got a working page instead of a video and only had to connect the code to the site. One hour of my work saved him about 4 days.

Results

In the first months after launch, year over year:

  • visit-to-lead conversion for services +28%;
  • traffic −19%, with the same number of leads as a year before.

In development: one hour of vibe coding saved about 4 days of front-end work on the products page.

Time on site and page depth also grew, which is expected: the old site had a single page.

Next case REFS REFS: 3,072 interface references with meaning-based search and 130 animations turned into prompts for AI agents. My own product, from idea to launch.