Semrush AI Visibility Index

As search shifts from ranked links to AI-generated answers, brands face a new and largely unanswered question, whether the models mention them at all. Semrush set out to measure it, charting AI visibility, how brands surface inside the answers large language models give, before the category had a yardstick. Reporting on a market this new carried an unusual requirement, one that shaped the build from the ground up. What began as a single report site became a partnership that now reaches into Semrush’s core marketing infrastructure.

Client-side React would have been the natural build, and for most of the last decade it would have carried little risk, because Google learned to render JavaScript. The crawlers behind ChatGPT, Claude and Perplexity never made that investment. They read raw HTML and stop. A site can still rank on Google while staying invisible to the AI engines now answering the questions, which is the gap most teams haven’t noticed yet. A report on AI visibility was the one publication that couldn’t afford to be on the wrong side of it.

We built it on Next.js and React, using server-side rendering and static site generation so the content arrived fully formed rather than assembling in the browser.

Apache ECharts handled the data visualisation, rendering the interactive charts that carry the report. One detail, tucked into the corner of the screen, showed the same thinking. Scroll to the bottom of a chart-heavy page and the underlying source data appeared, exportable to CSV. The site read as a clean visual story while remaining a working dataset underneath.

Semrush handed over the early design direction openly. They supplied several visual references for how the charts might work within The Index, then left the interaction model, the filter behaviour and the final execution to our judgement, treating the examples as direction rather than specification. That latitude shaped how people move through the data, from an orderly card layout in place of rotated cards, to animated prompts in the hero and a chart style aligned with Semrush’s existing reports.

Semrush wanted live chart data from the Index to appear on the main semrush.com homepage, which runs on an entirely different technology stack. We built a dedicated API endpoint that exposed the chart data cleanly, letting the flagship homepage consume it without inheriting the site’s frontend. Two codebases stayed independent while a single data feed flowed between them, the kind of structural choice that prevents breakages when either system changes later.

A full rebrand followed. New brand guidelines were applied across the site and delivered on schedule. Around the same time, we refactored the chart logic to support a new scaled measurement methodology. As the report library grew, detail-page content moved into a JSON-driven structure rather than a hand-built route per report, so new reports generate dynamically and the system scales with the editorial calendar instead of against it. From a download hub, a dedicated reports and new modules, we keep evolving with the site through our partnership.

For Semrush, the payoff is position. The Index now feeds the flagship homepage, putting an early, evidence-based read on AI visibility at the centre of the brand, in a category it helped define. New reports ship at the speed that Semrush needs to stay in a fast-moving field, leading the conversation and reaching the businesses trying to navigate it with data rather than opinion.