Recruit Polling Agents for Attendance, Not for Enthusiasm

Every campaign has the same recruitment failure and describes it the same way afterwards: we had the names.

Names are not the product. The product is a person standing in a specific room before dawn on 10 August 2027, still standing there when the count finishes, holding a result form. Enthusiasm at the recruitment meeting predicts almost none of that.

So screen for attendance.

The four questions that actually predict turning up

Ask these in the recruitment conversation, and write down the answers.

How will you get to the station, and how long does it take? An agent who lives a fifteen-minute walk away turns up in rain, in a transport strike, at half past five. An agent who needs two matatus and a boda in the dark is a coin toss, and the coin is weighted by weather.

Who else depends on you that day? A person with a sick relative, a market stall that cannot close, or small children and no cover has a competing obligation that will win. Better to know now than at seven in the morning.

Can you sit in one room for fifteen hours and stay alert at the end of it? This is a real question and people answer it honestly when it is asked plainly. The count comes at the end of the longest day of the campaign, and the agent who has already left is the agent who cost you the stream.

Who vouches for you? The single strongest signal is that a ward coordinator will put their own name behind this person, because that coordinator is the one who will be phoning them at five in the morning.

Where reliable agents actually come from

Not from an open call, and not from a form on a website.

They come from people who have already done work for you. Somebody who has knocked doors through a wet Saturday has demonstrated something that no interview can. Somebody who registered voters for you and reported their numbers on time has demonstrated more.

They come from people with a standing reason to be at that building. The person whose children attend the school, who worships nearby, who knows the head teacher and the chief, will not get lost, will not be turned away at the gate over a misunderstanding, and will not be intimidated as easily as a stranger.

And they come from your own volunteer base, which is why the base has to be built long before it is needed. This is the failure described in the volunteer nobody called: a campaign collects willing people, never gives them anything to do, and finds them unreachable when it finally needs them.

Say the terms out loud at recruitment

Tell people what the day is before they agree to it. The hour they must arrive. The hour they can realistically expect to leave, which is after the count and not after the voting. Whether food is provided. What they will be paid and when.

An agent who learns the terms on polling day is an agent negotiating at the worst possible moment, and a campaign that has not said the terms out loud has agreed to whatever the agent assumed. Set the money out properly, along the lines in agent pay.

Confirm twice, and treat an unconfirmed name as empty

A name on a list that nobody has phoned is not an agent.

Confirm once at recruitment, then again close to the day, by voice. Not by broadcast message, which people read and do not answer. A short call that ends with the person saying the station name back to you is the whole test.

Anything unconfirmed goes back into the uncovered column, where it belongs, so the gap stays visible. Recruiting early, as set out in the eleven-month plan, exists precisely so there is time to lose people and replace them.

What to do with this week

Give every ward coordinator a target drawn from your stream sheet, and a simple rule: no name is submitted until they have spoken to the person and can answer the four questions above.

Then hold them to the gap rather than the total. A coordinator measured on how many names they submit will submit names. A coordinator measured on how many streams are covered and confirmed will bring you agents.

Recruitment done this way in 2026 is unhurried and cheap. The same work in the last fortnight before the poll is a bidding war you will not win.