How do you collect carbon data from households or members over years, not once?
Ask the same small set of questions every period, make submitting take under ten minutes, show each participant their own trend back, and never overwrite a submission. Drop-out, not accuracy, is what kills a longitudinal series, so design against it from the first round.
Decide what the series is for before designing it
A one-off survey answers a question. A series answers whether something is changing. Those need different designs, and the second is only worth the effort if a decision depends on the trend: whether an intervention worked, whether a grant achieved what it promised, whether a group is moving in the direction it says it is.
The questions
Fewer than people expect. Domestic footprints are dominated by a handful of activities, so ask about those and estimate the rest:
- Home energy: gas and electricity, in kWh, from the bill or meter.
- Travel: distance by car, public transport and air since the last submission.
- Diet: a simple category rather than a food diary, which nobody sustains.
- Significant purchases: the occasional large item, asked as a yes-or-no with a value.
Each answer is multiplied by a published conversion factor, exactly as in an organisational footprint. The arithmetic is the easy half.
Consistency beats precision
A more accurate question introduced in round four breaks comparison with rounds one to three. If a question must change, keep the old one running alongside for at least one period, and record which version produced each figure. The same applies to conversion factors: fix a factor set for the series and recalculate the whole series when you move to a new one.
Drop-out is the real problem
| Cause | What helps |
|---|---|
| Submission takes too long | Pre-fill last period's answers and let people correct them |
| Nothing comes back | Return each participant's own trend, and the group's, within days |
| Forgetting | One reminder on a fixed day, the same day each period |
| A gap becomes permanent | Accept late and partial submissions rather than treating a missed round as an exit |
Reporting an average across a shrinking group produces a flattering trend, since the least engaged leave first. Report the number of participants next to every figure.
How to store it
- One row per participant per period. Never edit a past row; add a correction row.
- A stable participant identifier that is not a name or an email address.
- The question-set version and factor-set year on every row.
- A separate table mapping identifiers to people, held apart from the measurements.
Household energy use is personal data, so collect the minimum, publish only aggregates, and be able to delete one participant's rows on request (UK GDPR guidance). The separate mapping table is what makes deletion straightforward.
Start in a spreadsheet
A spreadsheet with those four rules holds a series for years, and it is what I built for EcoCounts, the community carbon footprint charity I founded: around 25 participating households, tracked month after month, with the spreadsheet version still in real use. Move to something with a database behind it when several people need to enter data at once, or when participants should see their own history without being sent it.
If you are about to start collecting from a group, the question set and the storage rules are worth an hour before the first round rather than after it. That hour is free.
More answers
- How does a small organisation work out its carbon footprint without buying software?
- What does scope 3 actually require a small organisation to measure?
- What does a net zero roadmap cost when nobody in-house does sustainability?
- Can AI build a custom tool cheaply enough for a small organisation to afford?
- What should a website accessibility audit cover, and what does it cost?
DON'T PANIC
cyberspaceroad.com