LLMO Case Studies: Brands Winning in AI Search

Why LLMO Case Studies Matter — and Why Most Online Are Fake

In short

Most public LLMO case studies are invented. They cite anonymous 'Fortune 500 SaaS' wins with round-number outcomes and no verifiable link. Real LLMO case studies name the client (or programme type), publish dated baselines, and link to the underlying engagement page.

Search "LLMO case study" in May 2026 and you will find dozens of vendor blog posts citing anonymous wins. "A Fortune 500 SaaS grew citations 5x." "A B2B brand 10x-ed its AI traffic."

Almost none of those numbers can be verified. The clients are unnamed. The baselines are undefined. The methodology is absent. The numbers themselves are usually round — a marketing tell, not a measurement one.

A credible LLMO case study has four things. Each is checkable before you trust a number.

• Named programme scope. The client (or programme type), the work performed, and the time window.
• Dated baselines. Where the metric started and when. "+2,092%" without a baseline is meaningless.
• Verifiable outcomes. Specific numbers, not round ones. "+2,092%" is auditable; "+10x" is marketing.
• A link to the engagement page. A live URL that documents the case beyond the marketing claim.

The six Alice Labs cases below each meet that bar. They are the same engagements we reference in sales conversations and board reviews. The metrics are the audited numbers, not the best-case version.

Case 1: Media Company AI SEO Rewrite — +2,092% Clicks in 12 Months

In short

An Alice Labs media client rewrote 178 articles using AI-driven SEO and LLMO tactics. Over 12 months, clicks grew from 141 to 3,091 — a +2,092% increase on 8.77M impressions.

The engagement was a content-rewriting programme. The client's existing article inventory was ranking but not earning clicks — classic mid-position decay against AI Overviews and shifting SERP layouts.

Alice Labs ran 178 articles through an AI-augmented rewrite pipeline. Each rewrite added the patterns Aggarwal et al. (2024) flagged as effective in generative engines — specific statistics, named-source citations, structured answers near the top of each piece.

Programme scope. 178 articles, 12 months, single content domain, AI-augmented rewriting with editorial review on every piece before publication.

Baseline. 141 monthly clicks at programme start. The same article inventory was already indexed and ranking — the rewrite addressed CTR and citation quality, not ranking from scratch.

Outcome. 3,091 monthly clicks by month 12 — a +2,092% increase on 8.77 million total impressions across the rewritten article set. The rewrite economics are documented in full on the engagement page.

Read the full Media Company AI SEO Rewrite case →

Case 2: Accounting Firm AI SEO — 0 to 41 Top-3 Rankings in 3 Months

In short

An Alice Labs accounting firm engagement rebuilt search authority from a near-zero base. Over 3 months, top-3 Google rankings grew from 0 to 41, organic traffic rose +260%, and monthly visitors grew from 2,500 to 9,000+.

This was a brand-authority rebuild, not a tactical sprint. The client was a regional accounting firm with limited search presence outside its core branded queries.

The engagement combined LLMO foundations — entity clarity, Schema.org markup, citation-rich content — with topical authority work on the question patterns AI Overviews now dominate (advisory, regulatory, deadline-driven queries).

Programme scope. 3 months, single market, full content authority rebuild on advisory and regulatory topics, with structured-data and entity-graph cleanup.

Baseline. 0 top-3 Google rankings outside branded queries. ~2,500 monthly organic visitors. Almost no citation share inside ChatGPT, Perplexity, or AI Overviews answers on category-defining prompts.

Outcome. 41 top-3 Google rankings inside 3 months. Organic traffic +260% (2,500 → 9,000+ monthly visitors). The shift was visible inside ChatGPT and AI Overviews answers as well — the same content patterns that move Google also moved citation share.

Read the full Accounting Firm AI SEO case →

Case 3: Media Company Content Production — 1M SEK First-Year Revenue

In short

An Alice Labs media client built a content-production engine that turned into a self-sustaining revenue stream. First-year revenue reached 1M SEK and scaled to 400k SEK/month within 12 months.

This case is the production-discipline version of the LLMO story. The first two cases prove LLMO tactics work. This one proves that the bottleneck for most programmes is execution capacity, not strategy.

The engagement set up a repeatable content-production pipeline with AI in the loop — editorial planning, drafting, citation review, and publication cadence. Over 12 months, the pipeline output became commercially monetizable in its own right.

Programme scope. 12 months, content-production pipeline build, with AI-augmented drafting and editorial review on every piece. The deliverable was throughput, not a single content artefact.

Baseline. Limited monetizable content production. Editorial capacity was the binding constraint — classic media-company economics where strategy is clear but throughput is not.

Outcome. 1M SEK first-year revenue from the content stream. Scaled to 400k SEK per month within 12 months. Self-sustaining — the pipeline pays for itself and funds its own continued production.

Read the full Media Company Content Production case →

The Pattern: Aggarwal's Citation Tactics, Applied at Scale

In short

Every Alice Labs LLMO win shares the same underlying pattern: the citation/statistic/quotation tactics validated by Aggarwal et al. (2024, arXiv:2311.09735) applied consistently across a large content set. The lift is from compounding, not from any single tactic.

The six cases span industries — media, accounting, food, public sector, security, content. The mechanics differ. The underlying pattern does not.

Aggarwal et al. (2024) tested nine content modifications across multiple generative engines (arXiv:2311.09735). The winners were specific — adding inline citations, named statistics, and quotation from authoritative sources. Those patterns delivered up to 40% visibility lift inside generative engines.

Every Alice Labs LLMO engagement applies the same three patterns. The lift comes from scale and consistency, not from any single piece of content.

• Inline citations. Every claim sourced to a named authority, dated where possible. This is the single biggest LLMO content-quality move.
• Specific statistics. Concrete numbers in context — "+2,092%", "83% cost reduction", "6,400-8,000 hours/year" — rather than vague "significant" or "major".
• Authoritative quotation. Named sources, dated work, peer-reviewed studies. Aggarwal's own paper is the clearest example we cite repeatedly.

What makes the pattern repeatable is the production discipline behind it. The media rewrite case rewrote 178 articles. The content-production case shipped consistently for 12 months. The accounting case sustained authority work across 3 months.

Tactics without throughput do not move metrics. That is the single biggest takeaway from looking across all six engagements — the strategy is published in Aggarwal's paper, but the production discipline is what separates programmes that compound from programmes that stall.

How Alice Labs Structures LLMO Engagements

In short

Alice Labs LLMO engagements follow a four-phase structure: citation benchmark, content audit, AI-augmented production sprint, and ongoing measurement. The structure is the same across industries; the scope adjusts to client capacity.

The six cases differ in industry, scope, and outcome metric. They share the same engagement structure. Every LLMO programme Alice Labs ships goes through the same four phases.

Phase 1 — Citation benchmark. Fixed prompt set across ChatGPT, Perplexity, Claude, Gemini, and Google AI Overviews on category-defining queries. Baselines the share of voice today. Repeats quarterly to track progress.

Phase 2 — Content audit. Existing inventory scored on the Aggarwal pattern fit. Identifies which pieces are rewrite candidates, which need fresh content, and which need entity/schema cleanup before any content work begins.

Phase 3 — AI-augmented production sprint. The longest phase. Rewrites or new content shipped at the cadence the client capacity allows — 178 articles in the media case, sustained throughput in the content-production case, focused authority work in the accounting case.

Phase 4 — Ongoing measurement. Quarterly citation benchmark repeat plus GSC and Google Trends tracking on branded search lift. Outcomes are reported with the same metric framing used in the original baseline.

What varies between engagements is depth, not structure. A three-month accounting engagement and a twelve-month media engagement run the same phases — the production phase scales to client capacity and category complexity.

One implication for any internal team. The single biggest controllable variable is production throughput in Phase 3. Programmes that under-resource production consistently underperform, regardless of strategy quality.

Source: Alice Labs proprietary client engagements (verified — see /en/case/* links)