Ship on Tuesday, measure on Tuesday.
Adoption, activation and retention questions arrive the moment a release does. Ask them yourself instead of filing a ticket and waiting a sprint.
Net revenue by region
Quarter to date
The questions a release raises.
Every ship date starts the same list, and every item on it is urgent for about a week.
Is anyone using it
How many accounts have used the new export since Tuesday, and how many came back a second time? Answered while the release is still warm.
Did activation move
What share of accounts that signed up in June reached their first report inside a week? Then ask it again by plan, or by where they came from.
The ticket queue
The question is small and the queue is not. By the time the analysis lands, the decision it was meant to inform has already been made.
Two teams, two funnels
Your activation rate and the growth review's activation rate were built in different places, from different filters, and neither one is obviously wrong.
Which channels drove the most new customers in July?
Organic search brought 1,284 new customers, a third of the 3,891 total. Paid social added 962 at a blended cost of $41.20 each.
New customers by channel
July 2026 · first touch attribution
Why the funnel means one thing
One definition, one person counted once.
Usage data is the easiest data to over-count, and what counts as active is the easiest thing for two teams to quietly disagree on.
Activated means one thing
Activated is defined once, in one shared place, not in whoever wrote the report that day. Your number and the growth review's number agree.
One person, one count
One person clicking nine times still counts as one person. iDash reduces to one line per user before it counts, so the cohort size is real.
Two questions at once
Ask for sessions and support tickets against the same accounts in one go and iDash stops, because it cannot prove both counts together. Ask them separately.
It shows its working
Every answer names what it counted, the filters it applied and the window it covered. A question that was subtly the wrong question is visible, not just plausible.
Semantic model
4 models · 3 relationships · derived from foreign keys
- PKcustomer_iduuid
- FKregion_idint
- first_order_attimestamptz
- channeltext
- PKorder_iduuid
- FKcustomer_iduuid
- placed_attimestamptz
- net_revenuenumeric
- PKproduct_idint
- categorytext
- unit_costnumeric
- PKitem_iduuid
- FKorder_iduuid
- FKproduct_idint
- quantityint
- customers1:Norders
- orders1:Norder_items
- products1:Norder_items
What changes after a release.
Measure the day you ship
Open chat, ask what the new flow did overnight, and read the answer yourself instead of adding a line to somebody's backlog.
A release board, pinned
Pin the four charts that decide whether a feature stays. They ask their questions again against live data, so the board is still true a month later.
Fewer arguments about definitions
When activated is defined in one place, a review stops relitigating the number and spends its time on what to do about it.
Revenue analytics
Saved results. Every tile shows what it got when it last ran.
Net revenue
Monthly, trailing 12 months
Orders
12,847
+11.2%vs the prior 90 days
Average order value
$333.30
+4.0%vs the prior 90 days
Revenue by channel
Quarter to date
- Organic search37.0%
- Paid social24.0%
- Email19.0%
- Referral12.1%
- Direct7.9%
Revenue by region
Quarter to date
Measure the release, not the ticket.
Connect the product database read-only and ask what last week's launch actually changed.