Reading your results
Data you collect and never read is just doors you tired your volunteers out on. This is how to read a week of field data honestly and turn it into next week's plan.
Dispositions in your language, one scale underneath
You define your own dispositions – your own outcome codes, in the words your volunteers actually use – and every one of them maps to a single five-point support scale, from STRONG_SUPPORT through to STRONG_OPPOSE.
That combination is the point. "Rusted on", "leaning our way" and "polite but no" all sort onto the same scale automatically, so you get local language on the doorstep and comparable data underneath. Nobody has to argue about what a "3" means.
Design your dispositions once, before the first shift. Changing them mid-campaign makes this week's data incomparable with last week's. If you must change them, note the date and treat it as a new series.
Read three numbers together, always
Any one of these on its own will mislead you. Together they keep you honest.
Disposition breakdown – what actually happened at the doors, counted raw. The sanity check: if half your outcomes are a single code, either your codes are wrong or your volunteers are defaulting.
Five-point support distribution – where your contacts sit on the scale. This is your real support picture, not the one you were hoping for.
Contact funnel – how many doors became conversations, and how many conversations became scored contacts. This is where problems show up first. A beautiful support number means nothing if your contact rate is quietly collapsing.
The temptation in field work is to read only the number that flatters you. Looking at all three at once removes the option.
Contact rate is your early warning
Support numbers move slowly. Contact rate moves immediately, and it tells you about your operation rather than the electorate:
- Falling contact rate, same turf type – you are knocking at the wrong times. Weekday evenings and weekend mornings are not interchangeable.
- Low contact rate on one turf only – it is the turf, not the team. High-rise, holiday area, or an address list that has drifted.
- High contact rate, low completion – people are answering and then the survey is losing them. It is too long. See Building a branching survey.
Pace, not a wish
A target with no pace attached is a wish. Goals track pace against target, so you can see whether today's rate reaches the number by the date that matters.
Use it to make decisions while there is still time to act on them. Behind pace with three weekends left is not a moment for optimism – it is a signal to add a shift, recut turf toward higher-density areas, or move volunteers off low-yield lists.
Check pace weekly, on the same day, with the same person doing it. Irregular checking produces panic rather than adjustment.
Beware the numbers that flatter
- Support percentage among contacts is not support in the electorate. The people who open the door are not a random sample.
- A rising support percentage on a falling contact rate usually means you are only reaching your own supporters.
- Small turf, big claims. Sixty conversations in one street tells you about that street.
Close the loop
Reading the results is only worth it if the reading changes something. Every week, the output should be three decisions:
- Where volunteers go next weekend – recut the turf. See Turf and walk lists.
- Who gets contacted on another channel – roll supporters and persuadables into audiences. See Audiences and segments.
- What the script changes to – based on what came back in the text field and the debrief.
Getting the data out
When you need it elsewhere – a coalition partner, a mailing house, a scrutineering plan – results export straight to CSV. No screenshots, no retyping, no version that lives only in one organiser's head.