Log 000 – July 2026: our baseline is zero

Nothing yet. This is entry 000 – the reading taken before any of the work, so every entry after it has something honest to be measured against.

At launch, Southlume’s own AI visibility is zero: asked the questions we ask about everyone else, no assistant named us.

The reading: mention rate 0, citation rate 0 – not one of the 48 checks across ChatGPT, Perplexity, Gemini and Claude. Measured 15 July 2026, question set v1 (5 July 2026), market Australia.

The point of this log is to watch it move, or fail to.

For context, the category we measure was itself quiet at our start: on 6 July 2026, no assistant could point to an independent, cross-industry index of who AI recommends in Australia. And in the same pre-launch scan, ChatGPT answered 11 of 12 category questions without citing a single source – being named and being cited really are different things.

We ran the same measurement on ourselves that the methodology describes for everyone else. The full question set, the four assistants, our own site as the subject – run before this site went live, so the baseline predates any effect the site could have.

Run 15 July 2026 · question set v1 (5 July 2026) · market Australia · ChatGPT, Perplexity, Gemini and Claude · 48 checks, all complete.

No special treatment, no favourable phrasing. If the method is good enough to read others, it’s good enough to read us from zero.

Nothing mysterious. A brand that had published nothing was not yet recommended by anything – assistants can’t name what they’ve never seen. The interesting part isn’t this entry; it’s whether the next ones move.

What we’ll watch from here: when a first mention appears, which assistant it appears in, and whether anything we publish can be traced to it. Movement only counts when it clears the stability band – wobble gets reported as wobble.

And the standing question this log exists to answer in public: how long does it take a brand publishing measured, dated work to enter the answers at all?