I answer from the snapshot currently loaded: Jan, Feb, Mar, Apr, May, Jun, Jul 2026 against 2025, covering 149 brands and 68 corporations across $1.38B of measured spend. Ask about a brand, a corporation, a category, a channel, monthly pacing or conversion efficiency. Two further sources sit alongside it: conversation share and verbatims for 26 brands (Jul 1 – 31, 2026), and the TikTok organic analysis — drivers, whitespace, rivals, audience and creative (January – July 2026). Ask about one brand and you get whichever of the three hold it, kept separate, because they measure different universes and do not add together. Right now L'Oréal USA holds 25.5% of category spend, +753 basis points year over year.
Try one of these
Built from what is actually in the loaded snapshot.
What this can and cannot do
It answers from the loaded snapshot only. Ask about a brand that is not in the export and it says so rather than filling the gap — which is the behaviour that matters, because a confident wrong figure here ends up in a deck. There is no model in this path: answers are assembled from computed values by a deterministic router, so the same question always returns the same numbers, and those numbers always match the other tabs. The limitation is phrasing rather than accuracy — it understands a fairly narrow set of question shapes, and an unusual wording may fall through to the “no match” reply even when the data would support an answer.