LBOX University
Research on how AI engines find, judge and recommend companies.
Buyers now ask ChatGPT, Gemini, Perplexity and Claude which company to choose before they visit a single website. LBOX University studies how those answers are made, what the engines read, and what decides who gets named. Every figure carries its source and its date.
Read the research
The whole argument with one point per slide: clicks fell, almost nobody clicks inside AI answers, the visit moved to a second search, named brands get 2.3 times the visits, dashboards cannot see it, those buyers convert better, everyone sells the mention, and what LBOX.com does about it. Opens full screen for presenting.
What LBOX.com believes: the buying decision moved into the AI answer, and the visit moved to a second step that goes to the brand the answer named. Six claims with their evidence and status, the before-and-after journey map per 100 people with drop-off, redirect and gain, the named-versus-left-out proof, and the tests that would prove it wrong.
The buyer now decides inside the AI answer and acts in a second step: a search for the chosen brand on Google, Amazon or the brand site. Before and after charts of the decoupling, the five-step journey with evidence at each step, the new stated value of AI visibility, why "get found by AI" stops short, and the correct-or-embrace map LBOX.com works from.
Who documented the decoupling and with what data, every credible click-loss study with sample, method and limits, the Search Console reporting events that distort the picture (num=100, the 2025 impression logging error, core updates), and how third-party tools estimate traffic.
Two procedures a practitioner can follow: the outside-in estimate from public data (formulas, confidence by step, worked example) and the Search Console confirmation with clean date windows and a control group. Ends with how to measure the gain side: branded search, AI referrals and answer share.
The research is free. The full picture is on LBOX.com.
Everything behind these studies, from the companies and products to the people and the AI answers, is available for your own industry, your competitors and the accounts you want to win.
More than 50 million companies and over 1 billion products across 250+ industries, with competitors side by side.
Explore industriesThe decision makers at the companies you sell to, found by industry, role and location.
Find professionalsHow ChatGPT, Gemini, Perplexity and Claude describe your company and your competitors, and what to change.
Start freeWhat we are testing now
About 20 impact buying questions across clean energy, sustainable goods, recycling and impact services, each asked as a general question and as a sustainability question on ChatGPT, Gemini and Perplexity. Every impact claim in the answers is traced to its source: certification registries, independent coverage, forums, retailers, or the company marketing itself.
Method fixed before the first run. Results will be published here in full.
Explore renewable energy companiesOur research standards
Studies are read in full at the source. A figure we cannot trace to its original stays out.
Each number carries its source, sample size, engines and date, because AI answers change from week to week.
Measured findings, single tests and reasoned hypotheses are labelled differently and never mixed.
For our own studies the questions and the method are fixed before the first run, then published with the results.
Companies, contacts, products and AI answers for your industry, in one place.