Say Hello

WHAT THE AUDITS
ACTUALLY FOUND.

Fourteen stories from real client engagements. Firewalls silently blocking AI crawlers. Sites vanishing from the datasets AI learns from. Schema contradicting the page it sits on. Scary charts that meant nothing, and flattering numbers that lied. Every number is real. Every client is anonymized. Every check is one you can run on your own site.

Case Studies
14
Real Engagements
3+
Invented Numbers
0
Checks You Can Run
All

BOTS & ACCESS

4 CASE STUDIES

What server logs reveal about the machines reading your site - and the silent blocks nobody configured on purpose.

№ 01

AI Bots Kept Knocking On Doors This Site Abandoned Years Ago

Five AI crawlers made 1,500 requests. Five reached content. The rest hit doors the company forgot it had.
6 min read
№ 02

The Firewall Blocked 2 In 5 Of Claude's Crawler Visits

Nobody decided to block Anthropic. Bot protection did it anyway - on the one crawler with no side door.
5 min read
№ 03

The Site That Vanished From AI's Favorite Dataset

A 'block low-value crawlers' cleanup swept Common Crawl into the same bucket as a scraper. Nobody noticed for two months.
7 min read
№ 04

Every Page On This Site Had The Same Birthday

The team published new material every month. The site's markup said nothing had ever changed.
6 min read

TECHNICAL FINDINGS

3 CASE STUDIES

The invisible layers - schema, duplicates, forgotten surfaces - where a site contradicts itself and machines notice.

№ 05

The Site That Was 16x Bigger Than Itself

One checkbox in the CMS multiplied the site by 16. Nothing in the tooling asked whether there were 15 translations to fill it.
5 min read
№ 06

The Page Says £8,000. The Schema Says 'Ask Us.'

The pricing model changed. The visible page got updated. The schema kept saying whatever it said the day it was written.
4 min read
№ 07

4.7 Stars On The Website, 1.0 Star On Google

An unclaimed profile is still your profile. Its statement is just being written by whoever bothers to show up.
5 min read

CITATIONS & CONTENT

4 CASE STUDIES

How pages earn citations in AI answers, and how companies destroy their most citable assets without noticing.

№ 08

The Article Google Ignored Became ChatGPT's Favourite Source

The most valuable AI asset this company owns spent months invisible for the price of a sitemap entry and some internal links.
4 min read
№ 09

The Site Published Real Prices - And AI Stopped Making Them Up

The question 'what does X cost' does not go unanswered because you declined to answer it. It gets delegated - to the least accurate sources on the internet.
4 min read
№ 10

The Most Citable Pages On The Site Were Deleted

A statistics roundup can be the deadest wood in the archive by pageviews and still be the most machine-quotable asset the company owns.
5 min read
№ 11

The Gap That Showed Up In Every Engine

Presence flatters. Absence under favorable conditions is hard evidence - and this gap showed up in every engine at once.
5 min read

MEASUREMENT & CRAFT

3 CASE STUDIES

Reading the data honestly - the scary charts that mean nothing and the flattering numbers that lie.

№ 12

We Fixed The Rankings. The Clicks Never Came.

Position 6 in 2026 is not what position 6 was when the CTR curves in everyone's models were fitted.
4 min read
№ 13

Impressions Fell 53% And Nothing Was Wrong

Google showed the site to half as many searchers and the site got the same number of visits. That is not decline. That is pruning.
5 min read
№ 14

47% Or 94%? Both Numbers Came From The Same Data

The self-serving surprise is the most dangerous number in a report, because nobody in the delivery chain has a motive to challenge it.
4 min read
Your Site Has Findings Too

STOP GUESSING
WHAT AI SEES.

Every case study above started the same way: an audit that read the logs nobody was reading. Fixed price, 7-10 day turnaround, and every finding comes with the check that proves it.