—Case study · Find and be found
How one of the world's largest freight forwarders went from #54 to #16 in AI search citations
AI engines already recommended the brand. This work gave them something to quote.
#54 → #16
AI citation rank in two months
+45%
AI-assistant sessions, now 1 in 9 site visits
17
pages cited by ChatGPT, from zero
3 of 4
AI platforms' citation pools entered within one month
One of the world's largest freight forwarders had the highest AI brand visibility among five global logistics brands, yet AI engines quoted its website in just 1.3% of tracked answers. EvolvWorks rebuilt its US content for retrieval, not rankings. Within two months, its AI citation rank rose from #54 to #16 and AI-assistant sessions grew 45%.
The challenge: recommended by AI, quoted by almost no one
The brand was not invisible to AI. Across ChatGPT, Gemini, Google AI Overviews and Perplexity, it had the highest brand visibility in its benchmark set, at 5.6%. But when those engines needed a source to quote, they chose someone else.
- Mentioned, not cited. AI engines named the brand, then cited other websites. Its domain appeared as a source in only 1.3% of tracked AI answers.
- Cited by the wrong people. YouTube, LinkedIn and specialist logistics publishers were quoted ahead of every forwarder. The brand's domain ranked #54 in the citation pool, with 12 pages cited. None of the 148 videos AI engines cited was its own.
- No entry points. Keywords worth 90,500 monthly searches had a rival on page one and the brand nowhere. The site had no glossary pages and no route pages.
The shift was already under way. AI-assistant sessions had risen 43% while search sessions fell 21%. A standard SEO report would only have shown the organic decline, not the leak to AI answers.
What we did
Step 1: Build the measurement layer the category lacked
- Track: 475 keywords (280,000 monthly searches) and 98 AI prompts across four platforms, benchmarked monthly against four rivals.
- Separate: Brand and non-brand performance at query level, so falling brand demand could be read apart from ranking loss.
- Audit the citation pool: For each platform, which sources were quoted, on which page, for which prompt. This moved the focus from competitors to the sources AI actually retrieves.
- Isolate the cause: 89% of the non-brand loss sat in just five URLs. The decline in the sustainability cluster was a technical fault, not a competitive defeat.
The diagnosis changed the brief. The problem was retrieval, not ranking.
Step 2: Four workstreams, sequenced by what AI engines reward
- 01
On-site quotability
Key pages rewritten as standalone, numbers-led answers a large language model can lift directly, supported by FAQ schema, markdown mirrors and an llms.txt file.
- 02
New citable inventory
A 39-page glossary hub covering terms with 739,000 combined monthly searches, approved for production, plus a blueprint for a 53-route shipping hub.
- 03
Earned citations where AI looks
Placements in the publications AI engines already quote, including Inbound Logistics, Supply Chain Management Review, Centric Software, Buske and sector trade titles.
- 04
Owned media built for AI retrieval
YouTube explainers titled as the exact buyer question, seven to eight a month, with transcripts on the matching web page. Executive LinkedIn articles, three to five a month.
Results: AI visibility moved first, and moved fast
| Metric | May 2026 | July 2026 | Change |
|---|---|---|---|
| AI citation rank (domain pool) | #54 | #16 | Up 38 places |
| Share of AI answers citing the domain | 1.3% | 2.0% | +54% |
| Share of voice | 1.3% | 1.8% | +38% |
| AI brand visibility | 5.6% | 6.3% | +13%, the only brand in the set to rise |
| Pages cited by ChatGPT | 0 | 17 | Entered ChatGPT's citation pool at #10 |
| AI-assistant sessions | Baseline | +45% | Now 1 in 9 site visits |
Starting from no presence, the brand entered the citation pools of three of the four tracked AI platforms within a single month.
