Data engineer, AI engineer
AI Search Visibility Harness
The instrument behind the search work. Public code, and a sample report you can run with no API key.
The measurement code is public. Read it, clone it, and run a full sample report offline without an API key or a cent of spend.
A Python measurement harness that audits how often a brand and its competitors get named when buyers ask ChatGPT, Perplexity, or Gemini a question in their category. Config-driven from YAML, so swapping the query set and the competitor list points it at any market. It is the instrument behind the search-visibility work: rather than asserting that AI answers matter, it measures what the answers actually say, repeatedly, and shows the runs.
The problem
Answer engine optimization is filling up with people selling confidence off screenshots. The honest version of the work needs an instrument: something that asks the questions a buyer would actually ask, records what each engine answered, and does it enough times to survive the fact that AI answers vary run to run. Without that, every claim about AI visibility is a story, and there is no way to show a client whether anything you did moved the number.
What we built
A Python harness that runs a configured query set across multiple answer engines with repeated sampling, then scores four things that matter separately. Citation inclusion: how often the brand is named at all. Answer absorption: whether the brand's own framing shows up inside the answer text rather than merely as a link. Gatekeeper domain share: which third-party sources the engines lean on when they compose an answer about the category. And segment gaps: where the brand holds ground and where competitors own the framing. Everything is config-driven in YAML, and a sample mode runs the whole pipeline on synthetic responses so anyone can see the output shape before spending anything.
What was delivered
- Python harness with per-engine adapters, repeated sampling, and structured storage
- Scoring across citation inclusion, answer absorption, gatekeeper domains, and segment gaps
- YAML config for the query set and competitor list, so it retargets to any category
- Offline sample mode that produces every output file with no API keys and no spend
- Report generator: baseline report, competitive gap analysis, and a ranked action shortlist
Outcomes
- Visibility claims come with a dated run behind them rather than a screenshot
- Repeated sampling accounts for AI answers varying between runs
- Separating absorption from citation shows whether a brand is being read or merely linked
- Gatekeeper analysis names the third-party sources actually authoring category answers
- Re-running the same query set after the work is what shows a client whether it moved
Services: Python, AI answer engines, Measurement design, YAML config, Reporting