Delivery sees a program drift before the quarter ends
Programs slip gradually, and quarterly reporting is built to notice late. A look at the early signals and how to watch them without adding paperwork.

A program rarely changes course on a single day. Attendance at one site thins out over a month. A line in the budget runs a little hot. Each week looks close enough to the last that nobody raises a hand.
Quarterly reporting is poorly suited to catching this. It compares the end of the quarter to the plan, after the weeks in which someone could have acted.
The signals that move first
Two measures tend to move before outcomes do. Attendance is one, especially repeat attendance: a site can hold its headcount while losing its regulars. Spending pace is the other. A program that is underspending in month two is often understaffed or started late, and that shows up in results a quarter later.
Staff usually sense these things. A site lead knows the Tuesday group feels smaller. What they lack is the time to total sign-in sheets every week, so the sense never becomes a number anyone can act on.
Watching without more paperwork
The practical obstacle has always been data entry. This is where recent AI is useful in an unglamorous way. A language model can read a photographed sign-in sheet or a site lead’s text the way a person would, and software can then keep the running count. The reading is the new part. The counting is ordinary arithmetic on what was read, which is why the totals hold up.
Delivery does this from what the team already produces. It keeps each program’s attendance and spending current against the plan, and it tells the program director when a number moves away from it, with the source attached. If Tuesday attendance at one site has fallen by a third over four weeks, the director hears in week four and can open the sign-in sheets behind the figure.
Nobody fills out anything new. The sheet on the table and the text to the office are the same ones the team was already using.
What the number is for
A count raises a question, and answering it takes someone who knows the neighborhood. Software can see that Tuesday attendance fell. A site lead learns that the bus route changed. Delivery brings the first to the people who can find out the second, while most of the quarter is still ahead of them.
The same data carries forward. When the mid-year report comes due, Delivery drafts it from the program data it has been keeping and has it waiting for review a week early. The director adds what the numbers can’t say, and the funder reads about a change the team already acted on.
There is a quieter benefit for the people running sites. A site lead who mentions a problem in a text has now reported it. The concern is on file with the date it was raised, and it reaches the director without a form or a meeting.
If you run programs, try this whichever tools you use. For each one, name the number that would tell you in week three that something has shifted, and who would see it. If the answer is the quarterly report, that is the gap.


