Nobody could say how this month's field programme was going

Every month your field staff have a programme to run. There is a planning window, in which they line up their farmer meetings, and an execution window, in which they carry them out. Partway through, you would like to know one thing: is this month’s programme on track, and if not, in which states?

Nobody could tell you. The portal could report on one employee at a time, so a leader could open a single person and see how he was doing. There was no screen that answered how the cycle as a whole was going. By the time anyone added it up by hand, the window to carry out the meetings was nearly shut and the answer was no use.

When we built the screen, the first thing it showed was the number nobody had seen. The current cycle stood at 9.2 per cent complete, 116 of 1,262 tasks, with its execution window closing. Every cycle since May had stood at between 54 and 66 per cent.

What was actually going on

The system is the field-activity module of the ordering app that Aries Agro, a crop-nutrition manufacturer, runs for its sales staff. It held all the facts. It just had nothing that added them up across the reporting hierarchy, which is the question a leader opens the portal with.

Before building anything we checked the data against production, and three things were not what the design documents said.

The table meant to record which task types each cycle runs was empty for all 30 cycles. The only way to know what a cycle contained was to look at the tasks people had actually created, of which there are three types company-wide. Farmer records were attached to their task through a different link than the schema suggested. And the stored farmer count on a task was unreliable: it read 0 on 580 completed tasks in one cycle that did have farmer records behind them.

What we changed

A cycle dashboard, the first item in the field-activity menu. It shows completion with the phase and days left, planned against done for each task type, a trend over the last eight cycles, completion by State Head, how many completed tasks carry photos, video or remarks, how many farmers were reached, and coverage by state and district.

Farmers are counted as records, never from the stored count. A manager sees only the people under him, by the same reporting rules as the rest of the module, and a plain field rep is refused the screen.

Every headline figure was checked against a direct query on production and matched exactly. As of the log entry the screen was on the staging system, not yet released to production.

What it did not fix

It shows that a cycle is behind. It does not say why. The work log records no cause for the 9.2 per cent and we will not guess one here.

The empty table is still empty. The dashboard works around it by reading what people created, so a task type nobody has created yet will not appear.

The pattern, for anyone running a monthly field programme

Ask who can see the whole programme’s completion today, not last month’s. If the answer is that someone compiles it from individual reports, you find out about a bad month when it is over.

Then ask what the trend normally looks like. A figure of 9.2 per cent means little alone; it alarms because the last several months were between 54 and 66.

Where this ends up

Sazinga Field gives a leader the state of the programme across the whole team, so the question is answered while there is still time to act on it.