Tracking Turnout Ward by Ward While the Polls Are Still Open
On polling day you get one attempt to measure the thing you have spent a year trying to influence, and the measurement has to be built long before you need it.
A campaign that only learns its turnout from the declared result learns it about eight hours too late. A campaign that collects a number every two hours can still act in the afternoon. The difference between those two campaigns is not effort on the day. It is a sheet built in 2026.
The one number, and when it is sent
Each agent sends one figure at agreed times: the number of voters processed at their stream since the station opened.
One number. Not a paragraph, not a photograph, not a comment on the mood. A running total that a person can read off and a coordinator can write down.
Fix the reporting times in advance and keep them few. A handful of scheduled reports through the day is enough to see a trend and few enough that agents actually send them. Reporting every thirty minutes sounds thorough and produces silence by lunchtime, because the agent is inside a room doing the job you sent them to do.
Anything unusual is reported immediately and separately, through the escalation chain rather than mixed into the turnout number.
Related: who receives what, and who calls whom.
The denominator you need before the day
A raw count means nothing on its own. Four hundred voters is excellent at one stream and poor at another.
What turns it into information is the number of registered voters at that stream, which comes from the register published by the IEBC for each polling station. Get it, put it on your sheet next to every stream, and check it well before August rather than on the morning.
With the denominator in place, every report converts to a percentage of the stream's own roll, and percentages can be compared. Without it you have a list of unrelated numbers and a room full of people arguing about what they mean.
Why the reports go to one person and not to a group
A number posted into a group of two hundred people has not been recorded. It has been broadcast, and broadcasting is not recording.
Have coordinators collect from their own agents and pass consolidated ward figures to one named person who writes them onto one sheet. That person is the only source of truth for the day, and everybody else reads the sheet rather than scrolling for it.
This is also what protects you from the worst polling day failure mode, which is a campaign reacting to the loudest report rather than to the weakest ward.
Related: why the group chat is not a field operation.
Reading the numbers without fooling yourself
Three habits keep the reading honest.
Compare a ward to itself. The useful comparison is against what that ward was expected to do, based on its roll and on what you know from your own canvassing, not against a ward with different geography.
Watch the rate of change, not the total. A ward that started slowly and is accelerating needs nothing from you. A ward that started well and has flattened at midday is the one to look at.
Separate a reporting problem from a turnout problem. If a ward has gone quiet, the first question is whether its coordinator has a dead phone. It usually is, and campaigns have spent expensive afternoons solving the wrong problem.
Make it visible in the room
Whatever software you use, put the current picture where everyone in the command room can see it at a glance: ward, registered voters, reported so far, percentage, time of last report.
A wall, a whiteboard or a screen. The point is that the weakest ward is obvious to anybody who walks in, without asking, and that a ward which has not reported for hours is equally obvious.
Related: what to do when one of those wards is behind.
Build the sheet now, and test it in a by-election
One row per polling stream: county, constituency, ward, station, stream, registered voters, agent name, agent phone. The reported figures are columns you add on the day.
Then test the instrument before you depend on it. Any by-election, any party primary, any exercise where your people are deployed is an opportunity to run the reporting chain end to end and find out where it silently stops. The weakest link is almost never the agents. It is usually a coordinator with forty streams and one phone, and you would rather know that in 2026 than at midday on 10 August 2027.
Frequently Asked Questions
What should polling agents report during the day?
One number at fixed times: the total voters processed at their stream since opening. Keeping it to a single figure at a handful of scheduled moments is what makes the reporting survive until the afternoon, because the agent's real job is inside the room. Anything unusual goes up the escalation chain separately rather than being mixed into the turnout report.
How do you know whether a ward's turnout is good or bad?
You need the registered voters for each polling station, which the IEBC publishes, sitting on your sheet before polling day. That turns every report into a percentage of the stream's own roll, which can be compared. Without the denominator you have a list of unrelated numbers and no way to tell a strong ward from a weak one.