Case Studies

What data-driven AI search looks like

What actually happens when AI-search decisions are made from measured demand, and what it costs when they are made from opinion. Each case study is written from real work, shows the method end to end, and is free to read online.

Latest

A voice-of-customer radar that runs itself

What happens when weekly customer-voice monitoring stops being a manual chore: comments scored by an LLM, a dashboard rebuilt on schedule, and an integrity gate that blocks bad deploys.

August 2026

Roughly 18,000 public comments, scored weekly, published only when proven

A European automotive brand gets a weekly read on what customers say in public. Two LLM judges score every comment on a 4-tier sentiment scale and tag it against a closed 13-topic taxonomy; one cron job runs the whole chain for cents per week, and an integrity gate re-checks every published number against the database before anything deploys. The loudest finding: the most negative topic was a brand-trust question rather than any product attribute.

Read the case study

~18,000

public comments scored and topic-tagged, as of 10 August 2026

topics

sentiment tiers

Weekly

unattended runs

Published July 2026

The 9% problem

What happens when 400 well-written AI prompts steer a content strategy, and nobody has measured the demand they are supposed to represent.

July 2026

400 AI prompts, measured against 182 days of real search demand

An agency delivered a 400-prompt workbook to steer a brand's AI-search content. We matched it against six months of Google Search Console data: roughly 5 million impressions, 5,000 distinct queries. The clusters the set covered carried roughly 9% of measured demand. The bulk of what people actually ask, prices, availability, technical data, and one topic with no page at all, was absent. The replacement: a per-page question engine that produced 14 content drafts, every one traceable to measured demand.

Read the case study

~9%

of real query impressions covered by the 400-prompt set

~5M

impressions

5,000

queries

drafts

In preparation

Next write-ups from the same engagements

These cases are delivered and measured; the write-ups are in progress. Each will appear here when published.

Write-up in preparation

The baseline before opinions: a multi-market AI visibility audit

How a measured baseline replaced opinion in a multi-market brand's AI-search strategy, and how a fix list ordered by impact came out of it.

Write-up in preparation

Why AI assistants love press portals

What makes press-portal content the kind AI assistants cite: server-rendered pages, factual answers, question-shaped headlines. The lessons, ready to copy for any content hub.

Get new case studies by email

New case studies go out through the AI Visibility Report, our free fortnightly briefing on AI search. Double opt-in, unsubscribe any time.

Subscribe free Scan your page free

Want numbers like these for your brand?

The AI Search Visibility Baseline Audit measures where you stand across the AI surfaces your buyers use and delivers a fix list ordered by impact. From €4,900, two weeks from kickoff to dashboard.

Book a discovery call See the full service