Traffic collapsed after leaving Lovable
A Firebase migration replaced ranked URLs with 404s, broad redirects and conflicting canonicals. See how the loss appeared in Search Console and what recovery requires.
Read the case studyWebsite, search & AI visibility case studies
Evidence beforeempty promises.
These case studies document the technical failures, content changes and visibility signals behind real AI and search outcomes. The useful part is not the headline. It is the mechanism that produced it.
{}Evidence review board
5 studies live
Review standard
Show the mechanism. Separate signal from outcome.
01Starting conditionBefore
02Failure mechanismWhy
03Change madeWork
04Signal observedAfter
What changed. What broke. What worked.
Migration→Citations→Technical audits→Measurement
Selected work & analysis
A result only matters when the mechanism is visible.
Each piece looks past the easy headline. It records the starting condition, the change, the signal and the practical lesson another website can use.
Six months of AI visibility work
A practical look at which changes produced citation signals for a B2B client and which popular tactics did not move the result.
Read the case studyWhat an OpenClaw audit found
A client-site audit connecting crawlability, page structure, machine-readable signals and the fixes most likely to improve visibility.
Read the case studyAuthority on platforms you do not own
Why strong third-party content can fail to strengthen the main website, and how to reconnect distributed expertise to the owned domain.
Read the analysisTrack visibility every week
A measurement framework for separating rankings, citations, crawl signals, branded demand and actual business outcomes.
See what to trackStart with the failure mechanism
If traffic, indexing or citations changed, the first job is to find the exact layer that broke before producing more content.
Explore website auditsEvidence standard
A useful case study explains why.
A screenshot of a result is not enough. The work should show the starting condition, isolate the likely mechanism and separate a promising signal from a commercial outcome.
01
Diagnose
Show the cause ↗Find the mechanism
Trace the result back through URLs, crawl paths, page structure, content changes, proof signals and timing.
02
Measure
Track the layers ↗Separate signal from outcome
Rankings, citations and crawl changes matter. They become useful when connected to branded demand, referrals and business results.
What we record
From starting condition to practical lesson.
The aim is not to make every project look perfect. It is to document what another website can learn from the work.
- 01↗
Starting condition
Record the website state, technical constraints, content structure and known visibility signals before the change.
- 02↗
Change and mechanism
Show what changed, why it mattered and which technical, content or authority layer the work was designed to influence.
- 03↗
Signal and lesson
Track what moved, avoid claiming more than the evidence supports and turn the result into a practical next step.
Straight answers
Evidence without the inflated claims.
What makes these case studies useful?
Each analysis focuses on the starting condition, the technical or content mechanism involved, the change made and the visibility signal that followed.
What counts as an AI visibility signal?
Useful signals include crawl and indexation changes, search visibility, AI citations, branded demand, referral behaviour and business outcomes. No single metric explains the whole system.
Does AI visibility work guarantee citations?
No. Citations cannot be guaranteed. The work improves the clarity, evidence, technical structure and discovery paths that make a business easier for search and AI systems to understand.
Can you audit a visibility or traffic problem?
Yes. We audit design, content, traffic, indexing, redirects and technical structure, then identify the failure mechanism and prioritise the right repair.
Start with the mechanism
Find what changed.
Fix what broke.
If traffic, indexing or citations moved, diagnose the exact layer before publishing another batch of content.
Discuss an audit Find the mechanism. Protect what works. Repair the right layer.