Case studies

Eight engagements. Proof, not promises.

Every number below is read from the client’s own GA4 property, Search Console or Bing Webmaster Tools, over the window named beside it. Nothing here is modeled, blended or rounded up.

GA4 client property Search Console export Named date range Limits stated on every case

Brands already being named inside AI answers

Serviced Fire EquipmentCase NineMary Vance NCOITOOSingenuityBenarrivati Travel ServicesThe Avenue Cookery SchoolJulia Nicole Wellness Coach
The eight cases

Eight cases. Open data.

Each row carries the export the figure came from, the window it covers, and the three numbers that moved most. Open a case for the baseline, what was changed in what order, and what it did not fix.

01
Professional services · Sep 2025 to Jun 2026 LLM SEO

1,402 ChatGPT sessions, with Google clicks up 38.7%

A firm with real search demand that ranked on page three for most of it. Technical recovery, service pages rebuilt around buyer questions and separate tracking for each AI source took average position from 25.4 to 17, and GA4 logged 890 form submissions.

+38.7%Google clicks
1,402ChatGPT sessions
890Form submissions
Read the case →
Analytics export for the professional services engagement GA4 · client property
02
UK cooking school · 12 months AEO + GEO

£3,422 from ChatGPT visitors, beside £278.6K from organic search

A school that sells its classes online, where organic search was already the biggest revenue channel. Booking pages were rebuilt around buyer intent, with entity work and separate measurement for each layer, and the school ranked first of the brands tracked in its AI visibility benchmark.

£278.6KOrganic revenue
£3,422ChatGPT revenue
#1AI visibility
Read the case →
GA4 channel report for the cooking school engagement GA4 · client property
03
Digital marketing agency · 3 months AEO + GEO

Impressions up 462%, with clicks doubled and a mention in Google AI Overviews

An agency with service pages and some Google visibility that AI assistants weren’t recommending. Entity work and pages aligned to the queries buyers use grew impressions from 7.45K to 41.9K and GA4 key events from 16 to 126.

+462%Impressions
2×Clicks
16 → 126Key events
Read the case →
Analytics export for the digital marketing agency engagement Search Console export
04
Criminal defence law firm · Australia · 90 days Technical SEO + AEO

35 call taps and 13 forms in 28 days, with no paid ads

A firm that relied on referrals, with one landing page for its whole practice. A page for each of the 18 kinds of charges it defends and 95 answer articles took Google clicks from 800 to 1.37K and AI feature impressions from 59 to 13K.

1.37KGoogle clicks, from 800
13KAI feature impressions
704Bing AI citations
Read the case →
Search Console export for the criminal defence law firm engagement Search Console export
05
Real estate buyer’s agency · Australia · 6 months SEO + AEO + GEO

Google clicks tripled and Copilot citations grew sixfold, with 29 search and AI leads

An agency on page four of Google with a small site and no lead tracking. A topical map, 19 new money pages and 90 in-depth articles took Google clicks from 591 to 1.82K and average position from 31.4 to 22.2, and GA4 credited 29 new leads to search and AI assistants.

3xGoogle clicks
6.8KCopilot citations, from 1.1K
29Search and AI leads
Read the case →
Search Console export for the real estate engagement Search Console export
06
US eCommerce brand · Shopify · 90 days SEO + AEO + GEO

Google organic sales 3.2x, and $1,263 from a new AI sales channel

A store that sold mostly through paid social, with no sales from AI. An audit, schema, buyer-intent articles in six topic hubs and Shopify’s agentic storefronts took Google organic sales from $992.55 to $3,226.38 in one quarter, and Bing cited the brand 1.7K times in AI answers.

$3,226Google organic sales, from $993
$1,263AI storefront sales
+30%Total store sales
Read the case →
GA4 export for the eCommerce engagement GA4 · client property
07
US investment fraud law firm · 90 days SEO + AEO + GEO

Form submissions from organic search up 91%, with AI assistant visits more than doubled

A firm whose guides ranked for research searches while its service pages missed the hiring ones. Eight pages rebuilt for hiring intent, technical fixes and investigation pages on live matters took organic form submissions from 82 to 157 and free case evaluations from 82 to 137.

+91%Form submissions
+67%Free case evaluations
2.2×AI assistant visits
Read the case →
GA4 inquiry events for the law firm engagement GA4 · client property
08
Research peptide supplier · US and Australia · month one SEO + AEO + GEO

8.03K Google impressions and 55 ChatGPT visits in a new site’s first month

A supplier launching a new site in late September 2026, with no search history. A topical map, money pages, research articles and entity work, all in place before launch, brought its first Google clicks within days and showed it in Google’s AI features from its third week.

8.03KGoogle impressions
55ChatGPT visits
171AI feature impressions
Read the case →
GA4 traffic acquisition for the research peptide engagement GA4 · client property
The pattern across all eight

What actually moved the numbers.

Eight engagements in very different businesses, and the same sequence did the work in every one. Read this way the page stops being a list of results and starts being a method you can check against your own site.

01

The foundation came first

Every case started with crawl access, indexation and how pages link together. For the law firm that meant going from one landing page to 24; for the professional services firm, recovering pages already ranking on page three; for the peptide supplier, a site built for search before it launched.

Law firm · professional services
02

Pages were rebuilt around buyer questions

Priority pages were restructured so each section answers one question in full: service pages for the firm and the agency, booking pages for the school, a page per charge and 95 articles for the law firm, 19 money pages and 90 articles for the real estate agency, buyer-intent articles in six topic hubs for the online store, and eight service pages rebuilt for hiring searches at the investment fraud firm.

Agency · impressions +462%
03

Entity work made the brand quotable

An assistant won’t name a business it can’t tell apart from its category. Clear entity signals and structured data were part of the work in all eight, before any claim of AI visibility.

Cooking school · #1 AI visibility
04

Measurement was split into layers

Search clicks, leads, AI visits and AI visibility were tracked as separate numbers, so visibility was never reported as leads. Where no baseline existed, the case says so.

All eight engagements

None of the eight is a content trick. They are structural, which is why the same sequence holds for a law firm and for a cooking school, and why a vertical with no published case study here is not one we are guessing at.

How we verify

Why you can check every figure.

AI search results are easy to overstate and hard to audit, so the measurement rules were set before the work started rather than chosen afterwards to suit the number.

The client’s own property

Figures are read out of the client’s GA4 property, not ours and not a reporting layer on top of it. Nothing is re-keyed into a slide on the way.

Exports, not screenshots

Position, impression and click figures come from a dated Search Console export, so the same query set can be re-run later and checked against what we published.

A named date range

Every claim carries the window it was measured over. A 28-day count is labeled as 28 days, never annualized or projected forward.

Recorded in the live account

Two of the eight cases include a screen recording made inside the client’s own accounts, so you can watch the numbers being read rather than take our word for it.

What we don’t do

No modeled attribution. No blended “AI-influenced revenue” totals built from assumptions about how a buyer found you. No figures from our own analytics standing in for a client’s. And no annualizing a short result into a yearly one: if a number only existed for 28 days, that is the number we publish.

In their words

Hear it from the clients.

Three clients on camera, talking about the work in their own words.

EdenFounder · Hypa
Benjamin WilkisonMarketing Manager · Leaf Blocker
Niv ShalgiCOO · The Avenue Cookery School
Questions, answered

Questions about the proof.

Are these figures verified, or estimates?
Verified. Each one is read from the client’s own GA4 property, Search Console or Bing Webmaster Tools, over the window named on the card. Nothing here is modeled, blended or rounded up, and two of the cases include a recording in which the figures are read out of the live account.
Why are the measurement windows different lengths?
Because the engagements were. A 28-day lead count is published as 28 days and a 12-month revenue figure as 12 months. Normalizing them to a common window would mean estimating, and we would rather publish the real range than a tidier one.
Can I see the analytics myself?
On a call, where the client has agreed to it, we screen-share the account the figure came from. We don’t hand over raw exports from another company’s property, which is the same protection you would want applied to yours.
Why is revenue shown for only two of the eight?
Because only the cooking school and the eCommerce brand have purchase data in GA4 and Shopify. The research peptide supplier sells online too, but its sales tracking isn’t live yet, and the other five are lead-generation businesses. For them we publish form submissions, call taps, sessions and impressions rather than inferring a revenue figure.
How is a sale attributed to ChatGPT?
Through the referral source in the client’s own GA4 property. Assistant traffic arrives with its own source and medium, so it can be isolated as a channel and read against the purchase events already on the property. It counts clicked-through sales only: a buyer who reads the answer and returns later through a branded search is not in the number.
Do you have a case study in my industry?
Quite possibly not: eight published engagements do not cover sixteen verticals. The retrieval mechanics are the same everywhere, which is why the diagnostic is identical for all of them. What changes by industry is the prompt set, the entity map and the sources an engine already trusts in your category.
How long before something shows?
The first movement we can measure is usually crawler access and citation presence, inside the first cycle. Position and referral volume follow each engine’s own refresh rate; in the law firm case, most of the movement came in the second half of the 90 days.
3 business days · No sales call

See where you stand before you commit.

A written diagnostic on your own domain, measured the same way every figure on this page was. Complimentary, and yours to keep whatever you decide next.

Prefer to talk first? Book a strategy call →
Serviced Fire EquipmentCase NineMary Vance NCOITOOSingenuityBenarrivati Travel ServicesThe Avenue Cookery SchoolJulia Nicole Wellness Coach

Every figure on this page comes from client analytics we can walk you through