A Ward Is Too Big to Target. Work at Polling Station Level.
An MCA is elected by a ward, so the ward feels like the natural unit of a campaign. It is the natural unit of the ballot. It is far too big to be the unit of the work.
Every instruction a campaign gives at ward level is vague. Cover the ward. Canvass the ward. Get the ward out. None of those can be checked on a Tuesday, which means none of them can be managed.
What changes when you drop one level
At station level the same instructions become specific.
Cover the ward becomes: a named, confirmed agent in each of these eleven streams, and these three are still empty. Canvass the ward becomes: this station has 900 registered voters and we have recorded 140 conversations there. Get the ward out becomes: turnout at this station was low last time and it is where the transport should go.
Nothing about the work changed. What changed is that somebody can now look at a list and see what has not been done.
A station is not the same thing as a stream
This distinction costs campaigns their election night.
A polling station with a large register is divided into streams. Each stream has its own queue, its own count and its own set of result forms, including the Form 36A for the MCA contest. An agent watches a stream, not a building.
A campaign that recruits one agent per station in a dense ward has covered perhaps half of its counts and does not know it, because the recruitment number looks complete. Count streams.
Related: prove your agent coverage stream by stream.
Building the station file
One row per polling station. Six columns, none of which require software to start.
Station name and code. From the IEBC list.
Streams. How many.
Registered voters.
Votes cast last time and your share. From published results.
Canvass contacts recorded there.
Agents named and confirmed.
That file is the campaign. Everything else is commentary on it. When a coordinator says the ward is going well, the file either agrees or it does not, and the disagreement is the useful part of the conversation.
Four things station targeting lets you do
Send people where the votes are. Stations differ enormously in size. Two hours at the largest station in the ward is worth more than two hours at the smallest, and the ward-level plan cannot see the difference.
Spot the ward that is only half canvassed. A ward at sixty per cent coverage sounds respectable. Broken into stations it often means four stations near the road are finished and three further out have never been visited, which is a completely different problem with a completely different fix.
Match transport to distance. The stations that need vehicles on 10 August 2027 are visible in the file a year ahead. They are the far ones with a large register and a history of low turnout.
Reconcile your own result. If you have an agent in every stream, you have your own tally. If you do not, you have somebody else's.
The objection, and the answer
The objection is that this is too much detail for a small campaign.
A single ward has a number of polling stations you can count on your hands and write on one page. A constituency has more, and a county more again, which is exactly why the county campaign needs it in a system rather than in a notebook. The detail does not grow because you wrote it down. It was always there. Writing it down is the only thing that makes it manageable.
Related: what a nerve centre does with all this.
Start with the ten biggest
Do not try to build the whole file this week.
Sort your stations by registered voters and take the top ten. Fill in every column for those ten. Visit any of them you have not visited. Then extend the file downward as you go.
A campaign that knows its ten largest stations properly is ahead of one that knows its whole ward vaguely, and it can be done in an afternoon in 2026 rather than a panic in 2027.
Frequently Asked Questions
What is the difference between a polling station and a polling stream?
A polling station is the location. Where the register for that station is large, it is divided into streams, and each stream has its own queue, its own count and its own set of result forms. Agent coverage has to be counted in streams, because a single agent cannot watch two counts happening at once in the same building.
How many polling stations should a ward campaign track?
All of them, but start with the largest. Sorting stations by registered voters and completing the file for the top ten gives you most of the electorate for a fraction of the effort. Extend downward from there as the campaign grows, rather than trying to build the whole file at once.
Why does ward-level planning fail?
Because every instruction it produces is unverifiable. Cover the ward and canvass the ward cannot be checked, so nobody can tell a coordinator who is working from one who is not. Station-level instructions name a place and a number, which means progress and the absence of progress both become visible on the same page.