Tramm
An AEO programme for a GPU-powered transportation management system: answer-engine readiness up from 3/10 to 6/10
The problem
My February 2026 audit scored tramm.global 3/10 for answer-engine readiness. None of its structured data worked: 15 JSON-LD scripts were broken and threw errors on every page load, and the FAQ, Organization and SoftwareApplication schema AI engines rely on was missing. Pages had more than one H1, product copy led with marketing lines instead of answers, four dev and test pages were indexed by Google, and the llms.txt page sat behind a redirect loop.
If AI engines can parse Tramm's pages, with working schema, answer-first copy and search-intent metadata, they will start citing and referring buyers to a product they could not read before.
What I did
- AEO audit (February 2026): readiness scored across schema, content structure, FAQ, content depth and E-E-A-T, with a prioritised roadmap.
- Schema repair: removed the 15 broken JSON-LD scripts (about 14 KB off every page), then validated a FAQPage on the FAQ hub and SoftwareApplication schema on the homepage and the product and industry pages.
- Answer-first content: rewrote the homepage opening as a direct definition of Tramm, added FAQ sections with schema to the homepage and Why Tramm, and fixed the heading structure so every page has a single H1.
- Crawl hygiene: rescued llms.txt from its redirect loop so it loads at the standard address, took the four dev pages out of Google's index and fixed the homepage industry links.
- Metadata: rewrote titles and descriptions on 14 commercial pages through the Webflow MCP, alongside the client team's rewrite of the top 7 blog posts. All 21 pages sit inside Google's display limits, and Open Graph now mirrors the SEO copy.
- Chora white paper page, the GPU vs CPU route-optimisation benchmarks that the site's navigation had promised.
- Phase 3 plan: named customer case studies, a Chora vs NVIDIA cuOpt comparison, a "What is a TMS" pillar page and a buyer's guide for each industry.
How it came together
Where it stands
Re-scored in October 2026 against the same five categories as the February audit, Tramm's answer-engine readiness rose from 3/10 to 6/10. Schema went from 0/10 to 8/10: every page checked now carries valid JSON-LD, with FAQPage markup on the FAQ hub, homepage and Why Tramm, and SoftwareApplication on the homepage and seven product and industry pages. Every page has a single H1, the homepage opens with an answer-first definition, and llms.txt now loads at its standard address.
In Google Search Console, click-through rate rose from 0.39% in May 2026, when the new titles went live, to 1.3% for 1–21 October. Monthly clicks went from 47 in May to between 60 and 70 every month since. In GA4 for 25 July to 22 October 2026, ChatGPT shows up as a referral source, and Google organic search brought 248 sessions.
What's next
The remaining gaps are content, not code: named blog authors, FAQs on the product pages, customer case studies, comparison pages and a "What is a TMS" pillar. Branded searches for "Tramm" still bring most of the clicks, so phase 3 is about winning the category questions buyers ask before they know the name.
The mark
The brand at work
On paper
The website
Planning something similar?
Book a 30-minute discovery call, or run the free AI test to see how answer engines read your site today.