11 March 2026
DAU still hides the people who keep the workspace alive
Daily active users is a comfortable number. It moves every morning. It can be compared to last Tuesday. It also treats a tourist who opened settings and a finance lead who closed payroll as the same pulse. For SaaS usage pattern analysis, that is not a rounding error. It is the wrong object.
Two clicks, one chart
In the Khlong San archive we keep a synthetic invoicing product. On any given Wednesday, DAU is dominated by people who export a CSV and leave. The workspaces that actually keep paying are quieter: one admin, a handful of approvers, a month-end spike. If you only watch the daily total, the exporters look like your “power users.” They are not. They are a printer.
App analytics teams inherit DAU because mobile consumer products made it famous. B2B seating does not behave like a social feed. Shared logins, Line invites, and prepaid cards in Thailand make “a user” even less stable. Grain has to be named before the chart is drawn.
What we ask seats to do instead
In Sequence Atlas we ask for a weekly tenant return by feature family, not a daily headcount. Feature Gravity Lab then asks which screens still appear after week two. The quiet week (days 8–14) is where DAU is most misleading: the chart can look healthy while the approval queue is never opened again.
None of this produces a single score. That is the point. A score would hide the same people DAU already hides, only with more decimal places.
When DAU is still allowed
Consumer wrappers, viral loops, and anything billed per seat with true individual identity can still use a daily count — alongside a path library. We do not ban the metric. We refuse to let it stand alone in a QBR about expansion.
If your Monday board still opens with DAU, bring the export to Sequence Atlas or start with the shorter Quiet-Week Reading desk.