last 15 minutes Β· wall-clock presence Β· refreshes with the dash
core: neurons β visits β views Β· derivatives: depth = views/visit Β· dwell = attention/visit Β· att/neuron = total looking per unique key Β· scroll = max % reached per neuronΓpage (from attention sensor) Β· NEW = first-seen in window Β· RETURNING = seen before window
shape of engagement Β· flat left = healthy long tail Β· spike on 32+ = whales / bots Β· top list is heaviest keys (truncated)
cohort = week of first-seen Β· W0 = that week Β· Wn = still active n weeks later Β· % of cohort size Β· returns = first seen in window, came back within 7d
pick any two metrics Β· left axis green Β· right axis magenta Β· each scaled independently Β· day buckets (hour in 24h)
neurons = unique visitors who entered / left here Β· visits = passage count Β· grouped by first path segment (click to expand)
consecutive pageviews inside a visit Β· from β to Β· thickness = neurons who took the edge Β· answers "after X, where?"
ordered path steps Β· neuron must hit each step later than the previous (any visit, full window) Β· v1 = page paths only Β· convert/tx goals come next
grouped by first path segment Β· click a folder to expand children Β· segment neurons/visits are sums of child paths (may overlap)