A research hub for GEO and LLMs.
Domino is built on deep academic expertise in how large language models work — how they retrieve information, reason over it and decide which sources to cite. The Domino Lab turns that understanding into research the industry can use, and into every service we deliver.
A customer asks
Sources are gathered
- Your website
- Review sites
- News & directories
- Wikipedia
The model weighs them
Relevance, consistency and trust across sources decide who makes the cut.
One answer, few citations
"…for small companies in Dublin, your business is a strong option…"
✓ cited: your websiteGEO is too new for guesswork.
Much of today's GEO advice is recycled SEO. But generative engines don't rank pages — they compose answers, and the models behind them change constantly.
So we study the mechanics directly: we form a question, test it across ChatGPT, Claude, Gemini and Perplexity, and only turn what holds up into method. That's the difference between following trends and understanding the system.
What the Lab studies.
Six questions sit at the centre of our work. Each one shapes how we audit, fix and measure AI visibility.
Citation & source selection
How models decide which businesses and sources make it into an answer — and which are left out.
Read: the anatomy of an AI answer → AREA 02Retrieval & knowledge sources
Which sources each model learns from and retrieves at answer time, and how a brand's footprint across them shapes visibility.
See the Footprint Scanner → AREA 03Measurement without clicks
How to measure visibility when answers replace click-throughs: mention rate, accuracy and share of answer.
Read: measurement without clicks → AREA 04Accuracy & hallucination
When and why models misdescribe a business, and which signals reduce the risk of invented claims.
See the Hallucination Risk Audit → AREA 05Structured content & schema
How answer blocks, FAQ schema and llms.txt change what models can extract and reuse.
Read: why FAQ schema matters → AREA 06Behaviour across platforms
Why the same question gets different answers from ChatGPT, Claude, Gemini and Perplexity — and what that means for strategy.
Read: why brands get left out →Every finding sets off the next tile.
Research at Domino isn't a side project. It's the first domino in everything we deliver.
Question
Start from what businesses actually need to know about AI answers.
Experiment
Test it with structured prompts across the major models.
Method
Turn what holds up into repeatable audit and optimisation methods.
Results
Apply it for clients — and publish what we learn.
Latest research notes.
Why Brands Get left out of AI search even when SEO looks healthy
There is rarely one big reason.
Read →The Anatomy of an AI Answer: How Models Decide Who to Cite
Search is no longer a list of blue links. It is a direct answer.
Read →Artificial intelligence search engine optimization starts with a visibility audit and topic graph
Are you doing SEO and still not showing up in AI answers. Are your competitors getting named in Google AI Overviews while you get ignored. Are you see
Read →Research with us.
We work with academics, partners and businesses who want to understand how LLMs really behave. If you have a question worth testing — or want to take part in a study — talk to us.