A mid-sized B2B online shop for industrial shelving and warehouse equipment — over 10,000 products, established in the German market — discovered it was virtually invisible in AI-generated answers. Neither ChatGPT nor Perplexity mentioned the company for typical industry prompts. Competitors — including a large industrial supplier and a price-aggressive online retailer — were recommended regularly.
Starting Position
The initial analysis painted a clear picture: across ten typical prompts such as "Which heavy-duty shelving supplier do you recommend?" or "Where can I buy pallet racking online?", the shop was mentioned in zero cases. The larger competitor appeared in seven of ten prompts, the price-aggressive retailer in four.
Technically, the shop was missing nearly everything relevant for GEO: robots.txt blocked GPTBot and ClaudeBot. No Schema.org markup beyond the standard minimum. No LLM.txt. No FAQ sections. No author box on the existing but thinly maintained advice blog.
The 8-Week Plan
Weeks 1–2: Technical Foundations
The robots.txt was updated to explicitly allow GPTBot, ClaudeBot, and PerplexityBot. An LLM.txt was created describing the company, its product range, differentiators, and core categories. Schema.org was expanded: Organization with detailed company description, Person for the managing director with industry experience, and FAQPage on the ten most important category pages.
In parallel, GEO monitoring was established using Rankscale.ai: 30 prompts across ten clusters, three competitors as benchmarks.
Weeks 3–4: Content Optimization
The existing advice blog was overhauled. Five articles covering the highest-revenue categories received: citable statements in the introduction, expanded FAQ sections with five to eight questions per article, author box with industry experience and qualifications, and clearer structure with question-answer headings.
Additionally, three new guide articles were created addressing typical comparison and advisory questions — exactly the type of content LLMs prefer as sources.
Weeks 5–6: Off-Page Measures
The Google Business Profile was fully updated with all services, current photos, and an active review strategy. Existing customers were specifically asked for Google reviews. The company was listed in two industry directories and placed an expert article in a trade publication.
Weeks 7–8: Monitoring and Adjustment
Initial tracking data from Rankscale showed movement: the shop appeared as a source for three of 30 prompts — exclusively for prompts pointing to the newly created guide articles. Sentiment was consistently positive. Share of Voice rose from zero to eight percent.
Results After 8 Weeks
Progress was measurable, though still early-stage: AI visibility moved from zero to a trackable score in Rankscale. Three of 30 prompts showed mentions — all in positive context. GA4 reports showed AI referrer traffic for the first time, albeit at low levels. The technical GEO score (robots.txt, Schema.org, LLM.txt) rose from 15 to 85 percent.
Competitors maintained their positions, but the gap had narrowed significantly.
Lessons Learned
Technical quick wins deliver fastest: robots.txt adjustment and LLM.txt creation showed initial effects within three weeks. Citable content is the lever: new mentions came exclusively from revised and new guide articles, not from product pages. Off-page takes time: directory listings and trade article had no measurable AI visibility impact after eight weeks. Regular monitoring is essential: without Rankscale, progress would have been invisible and adjustments impossible.
FAQ
Is this timeline realistic for any business? Technical measures (Weeks 1–2) are universally implementable. Content velocity depends on available resources. Smaller companies can stretch the plan; the sequence remains the same.
What did the project cost? Tool costs were €20 per month (Rankscale base plan). The bulk of effort went into content creation and technical adjustments — internal time budget plus external consulting.
Can I achieve these results without an agency? Technical measures and monitoring are manageable independently. For content strategy and creation, external support can significantly improve quality and speed.
