EQU: how we measure adoption
We use EQU to understand whether our team is making SparkAI a regular part of their work. We look at two things: how many days someone uses it and how much qualifying activity they have.
The change we are adding is a 10-day minimum. We need at least 10 days someone was expected to work before we assign a usage level. That gives us more to go on than a day or two of activity.
What counts as an eligible day?
An eligible day is a Monday-Friday day in the reporting period, minus full days of leave recorded in Harvest. If a period has 22 weekdays and someone took four full days off, we assess them against 18 eligible days.
The dashboard defaults to the last 30 days. We can also look at month to date, last month, a selected month or year to date. Dates use Eastern time.
What happens below 10 days?
The person shows Not enough eligible days. We can still see their activity, but they are left out of the EQU percentage and the needs-support list. We keep their actual day count; we do not change it to 10.
How someone reaches each level
An active day is a weekday with at least one qualifying work activity. Activities include work such as drafting, research, planning, analysis, editing and building. We count credited work sessions or tool activities, rather than each individual chat message.
Both requirements below have to be met. We round required active days up to the next whole day.
| Level | Active days needed | Activities needed |
|---|---|---|
| Progressing | 25% of eligible days | Required active days x 1 |
| Qualified | 50% of eligible days | Required active days x 2 |
| Power user | 75% of eligible days | Required active days x 3 |
| Flying Pig | 90% of eligible days | Required active days x 5 |
Activities do not have to be spread evenly across those days. Someone can also qualify using one kind of work; category variety is not a requirement.
A few examples
At the 10-day minimum, Qualified means using SparkAI on five distinct weekdays and having at least 10 qualifying activities. As eligible days increase, the requirements increase with them.
| Eligible days | Active days for Qualified | Activities for Qualified |
|---|---|---|
| Fewer than 10 | Not assessed | Not assessed |
| 10 | 5 | 10 |
| 18 | 9 | 18 |
| 20 | 10 | 20 |
| 22 | 11 | 22 |
How we handle time off
Harvest leave includes PTO, holidays, sick time, Summer Fridays, unpaid time off and bereavement. A full day of leave covers at least 90% of normal daily capacity: 7.2 hours for someone who normally works an eight-hour day. Half days remain eligible.
We link people through an admin-selected Harvest person or an unambiguous name match. If someone is not linked, or leave data is unavailable, we use all weekdays. Missing time entries by themselves do not remove a day.
Weekend work counts toward activity volume, but does not add active weekdays. Work on a weekday taken off still counts, even though the leave reduces expected days.
What goes into the leadership percentage
We divide the number of Qualified, Power user and Flying Pig users by the number of people eligible to be assessed. If 20 people meet the minimum and 12 are Qualified or higher, our EQU rate is 60%.
People below the minimum stay visible without affecting that rate. Tester/QA and developer/administrator roles are tracked separately from the leadership adoption cohort.
What the number tells us
EQU gives us a consistent way to track adoption: enough eligible days to assess someone, regular use during that period and a minimum amount of work activity. It does not prove time savings or work quality. Chat credit reflects the work requested; tool credit uses recorded completed actions.
Release note - October 2, 2026. The 10-day rule is implemented in PR #135, pending merge and deployment when this document was prepared. All 38 focused tests passed, including the nine-day/10-day boundary and percentage calculations. TypeScript, lint and syntax checks passed. Live browser verification is still outstanding.