Measurement method
Before-and-After Visibility Measurement
Source trust work is only worth doing if the effect can be observed. This page states the method, and publishes BackTier's own dated runs as they are captured.
The method
Frozen prompt set
A fixed list of buyer-language prompts is written before any change is made. The same wording is used in every later run, so a difference in the result is not a difference in the question.
Named platforms
Each prompt is run against the assistants that are actually being measured, one at a time, and the response is recorded verbatim with the date.
Three recorded outcomes
For every prompt and platform: was the brand mentioned, was it cited with a link, and which URL was cited. Nothing is scored on impression.
Baseline before, same run after
The baseline is captured before implementation. The post-implementation run repeats the identical prompt set so the two are directly comparable.
Stated limits
Assistant outputs are non-deterministic and change without notice. A single run is a sample, not a ranking. Differences are reported as observed counts, never as guaranteed outcomes.
BackTier's own runs
No runs have been published yet. Results appear here once a baseline run is captured and verified — no projected or illustrative numbers are shown in the meantime.