Two sections, three approaches each. Pick one per section, or mix. The
numbers are the same in every option so only the presentation differs.
1 · Member adoption
Replaces “How far members actually get”. The question is
whether the people who have access are actually using it.
AThree states, one bar
No funnel. Every member is in exactly one of three states, so the
parts sum to the whole and nothing implies a sequence people did
not take. Fastest to read.
Are members using it?
every member, by what they have done
18 of 99
members have delivered at least one rollout
59
22
18
59 never opened
it22 opened, nothing
delivered18 delivered a
rollout
BFunnel, with real steps only
Keeps the funnel shape but every band is now something a person
did, and the last step measures a habit rather than an arbitrary
count: delivered in two or more separate weeks.
From access to habit
where members stop
99
Given access100%
40
Opened it40%
59 did not get this far
18
Delivered a rollout18%
22 did not get this far
12
Came back another week12%
6 delivered once and stopped
CBy organisation, as a call list
Same data, but per customer, so it answers who to ring. An
aggregate cannot tell you that Acme has eight people with access
and none of them has opened it.
Dormant members by organisation
members with access who have never delivered
OrganisationOpened · delivered · dormantDormant
Acme Retail8 of 8 dormant
BWS6 of 6 dormant
Unicorn12 of 27 dormant
Northbridge Retail6 of 12 dormant
Meridian Group3 of 24 dormant
Harbour Media4 of 9 dormant
2 · What it’s worth
Five replacements, each answering a different commercial question.
Shapes and proportions follow the live production data; the figures
are stand-ins at the scale used elsewhere on this page.
AWhere the growth is coming from
Organisations that delivered work each week, split by whether it
was their first week. Growth and retention in one read.
Organisations delivering work
by week, first-timers separated out
4 last week3 of them for the first time
Arrivals are rising. Returns are not: one organisation has
come back every week, and so far it is the same one.
First weekDelivered before
21 Jun28 Jun5 Jul12 Jul19 Jul26 Jul
A first delivery is the moment an organisation stops being a
signature and starts being a customer. Six have reached it.
BHow much rests on one customer
Share of all delivered work by organisation. The concentration
question a board asks before it asks anything else.
Where the work comes from
share of everything delivered since launch
92%of all delivered work comes from one organisation
Meridian Group5713 people
Northbridge Retail21 person
Harbour Media11 person
Acme Retail11 person
BWS11 person
Every organisation except the first is one person who has
delivered once or twice. The portfolio total moves when a single
account moves.
CHow quickly people get to value
Time from first opening Smart Rollout to a first delivered
rollout. Answers whether the product is hard to start.
Time to a first rollout
measured from the first time a member opened it
Same dayfor two thirds of everyone who gets there
It is not a slow climb. People either deliver something almost
immediately or they do not deliver at all.
Same day12
Within a week1
Longer than a week4
Still nothing23
The slowest first rollout took 31 days. Everyone else who was
going to get there had done it inside a week.
DWork made, work taken away
Members who have delivered a rollout against members who have
exported one. A finish-line gap, and a reason to ring someone.
Does the work leave the building?
delivered against exported, by member
18
members have delivered at least one rollout
5
members have exported one out of the tool
13 made work they never took away
Either the export is not being found, or the output is being used
some other way. Worth one phone call to find out which, because
the answer changes what we build next.
EWho to call this week
The action layer from the metrics brainstorm: named organisations,
what happened, and why it is worth a call. Nothing here is a
number for its own sake.
Worth a call this week
most time-sensitive first
Harbour Media
Delivered their first rollout eight days ago and has not been
back since.
Rescue
Northbridge Retail
Two rollouts delivered, neither exported. The work is being
made and left behind.
Unblock
Acme Retail
First rollout this week, from one person on an eight-seat
plan.
Expand
Meridian Group
Passed fifty rollouts, and thirteen people there have now
delivered one.
Reference
Every row is drawn from a signal already on this dashboard. The
value is in putting them in one place and ordering them.
3 · What it’s worth, second round
Five more, each a different shape rather than another figure in a box.
Every number below is measured on production Smart Rollout data since
launch, at true scale, not a stand-in.
FThe verdict, in a sentence
The graphic is the sentence. Written from the data each time it
loads, so it can be read out as it stands.
Where this stands
the week, in one line
Four organisations delivered work last week,
three of them for the first time. Only
one has ever come back.
6organisations have delivered at all
128rollouts since launch
18members have delivered one
90%of attempts reach an output
GDoes it stick?
A return curve rather than a total. The shape says it before the
numbers do.
Do they come back?
members still delivering in a second week, and a third
12 of 18
delivered a rollout in one week and have not delivered in
another since.
The drop from the first week to the second is the steepest thing
on this dashboard. Everything after it is a rounding error.
HWhat the machine did instead of a person
Work displaced rather than money charged. Reads as a ledger, and
the one assumption is printed on it rather than hidden.
What was done, and what it took
production work since launch
Rollouts deliveredreached a finished output128
Machine steps runtool calls the agent executed1,933
Model calls madereasoning and generation3,529
Model spendinference only, at cost$84.19
Per rollout15 steps and 28 model calls apiece$0.66
At an hour and a half to build one of these by hand, 128 rollouts
is about 192 hours of work. The hourly rate is yours to set.
IDoes it work when they try?
Two measures that disagree: attempts look strong, people less so.
The tension is the point.
Reliability
by attempt, and by person
By attempt
90% reach an output
128 delivered9 failed6 cancelled
By person
1 in 3 has seen it fail
6 of the 18 who
delivered also hit a failure
Nine failures spread thinly would be noise. Six people out of
eighteen meeting one is a first impression. The weakest step is
plan building, which lands 92 times in 100.
JThe account, up close
One account profiled instead of twelve averaged, because almost
everything delivered comes from it. The case study and the risk in
the same card.
Meridian Group
the account behind most of the delivered work
Rollouts delivered by each of its thirteen active members
13members have delivered work
5of them account for 103 of 118
Every weekactive since its first delivery
Depth like this is what a reference customer looks like. It is
also the reason the portfolio total moves when one account moves.