Turning household electricity totals into trustworthy explanations for student renters.
We read Designing for Contemporary Challenges through the everyday scene where a person meets environmental sustainability most directly: the electricity, gas and water a household uses. Bills and usage data already exist, but they rarely explain where consumption came from or which behaviours actually mattered.
Resource use is framed as a global problem, but it is experienced as a monthly or quarterly cost.
Most residents hold a vague idea of sustainability. What they track is how much this quarter cost them.
They must manage daily use and bills while holding only partial control over the housing they live in.
How can student renters better understand their household resource use, so they can make better-founded and more sustainable choices without giving up too much everyday comfort?
Our questionnaire suggests awareness is not the gap. Respondents described cutting air-conditioning, switching off lights and unplugging devices. What they could not describe was cause.
Student renters must control a rising cost of living while holding limited understanding of household utilities and limited control over their housing. Adding more data and more reminders introduces a new pressure rather than resolving the first one.
So the design opportunity is not to make utility use more visible. It is to make it understandable.
Exploratory, student-weighted convenience sample. These figures describe reported experience, not prevalence or measured load.
A single total can carry personal use, housemate routines, a hot-water system, fixed appliances, insulation and shared building services. Those inputs are compressed into one number, and the renter has to decide which part was theirs.
The renter pays the whole amount, on time, regardless of what produced it.
One aggregated figure, arriving weeks or months after the use that created it.
Portable appliances are theirs. Fixed systems, meters and tariffs are not.
Student renters cannot reliably attribute a change in utility use to a specific cause, so even when they are willing to cut back they cannot tell which behaviour is worth changing.
Uncertainty pushes people to cut the most visible use first, usually cooling and lighting. The larger load may sit in hot water, fixed equipment or dwelling efficiency. Reducing use is not the same as reducing the right use.
How a resident finds out that usage or cost has changed in a way worth attention.
How they judge whether it came from their own behaviour, shared use or the housing itself.
How they decide what to do next, and whether anyone else needs to be involved.
Current design judgements rest on electricity-related evidence.
The person paying, the people using, the party controlling the equipment and the party holding the data are not the same actor. That separation is why a bill alone cannot settle the question.
Every possible cause must show who can check it, without assigning blame or implying that responsibility is legally settled.
Each method has a defined contribution and a defined limit. We label every claim in this document by the evidence class it came from.
Exploratory patterns, needs and concerns among student renters.
Prevalence, actual loads or behaviour change.
Structural renter and apartment context: hardship, efficiency, tenancy.
That this sample shares every reported issue.
Plausible breakdowns, renter language and real action routes.
Posters are not participants; replies are not diagnoses.
Traceable themes and research-derived design requirements.
Clusters are not universal, and not user validation.
Frozen working snapshot checked 28 August 2026. All respondents confirmed 18+.
Useful exploratory evidence about student and apartment electricity experiences.
A prevalence claim for all Sydney renters, or a claim that all respondents live in apartments.
Because 34 of 38 valid residence answers describe apartments or student accommodation, the findings that follow are read as apartment-weighted.
Self-rated; not an objective knowledge test. Difficulty items were selected, not ranked.
Open-text answers, verbatim. Reasoning rests on recall rather than evidence.
Renters commonly discover an anomaly only when a quarterly bill arrives, then work backwards through weeks or months they can no longer reconstruct.
Begin from an unexplained change. Show plausible causes, name the missing evidence, and offer one next check.
Two ambiguous responses excluded. Intention, awareness and information-seeking were not counted as actions.
AI-assisted, rule-based second-pass audit. Not independent human coding.
Observed in public online threads: several people faced with one bill, offering mutually contradictory causes.
Without attribution, saving becomes trial and error. Identical effort produces very different results, and the resident has no way to tell which happened.
Replace generic saving reminders with one reversible, hypothesis-specific check, and state what result would support or weaken it.
A single total mixes user-controlled and system-controlled consumption, so reading a high bill as overuse is unreliable. Where the resident cannot act, extra data produces blame, anxiety or unnecessary sacrifice instead of understanding.
The renter needs to distinguish a behaviour question from a housing-system question.
Public threads describe residents cutting cooling on hot nights even when heat was already affecting sleep, study or a pet. If a system only repeats that usage is high, data becomes pressure without understanding.
Residents infer a cause from memory or from whichever appliances are visible.
Usually cooling and lighting, often without ever confirming the cause.
The household loses comfort, time or trust, and the issue remains unsolved.
Attribution confidence is low, but willingness to reduce use already exists.
A high bill is followed by guessing, elimination, housemate disputes and suspicion of the housing.
Rental consumption is shaped by structural factors: dwelling performance, infrastructure and billing arrangement.
help student renters turn an unexplained electricity total into a trustworthy account of what changed, what they can control, and what may require action from another stakeholder — without surveillance, false certainty or unnecessary loss of comfort?
These are constraints derived from evidence, not features. Every direction that follows is tested against them.
Near-real-time data may serve as evidence, but the resident decides when to look, how much to see, and whether to investigate. No compulsory attention, no frequent reminders, no scores or comparisons.
The design must help the resident understand what changed in this period, not simply restate a total.
Where data cannot confirm a cause, the system must not present inference as diagnosis.
Each possible cause states whether it sits with the resident, the household, the property, the building or the billing system.
Prefer small, reversible checks over instructions to cut consumption substantially.
Do not judge who wasted more, and never treat a sacrifice of comfort as sustainability success.
When a cause cannot be determined, say what is currently unknown and what further information would settle it. Do not cover uncertainty with an answer that only looks precise.
We began with the familiar options — smart plugs, whole-home monitors, circuit-level sensing, automatic appliance detection, dashboards and alerts. Each solves part of the problem. None closes the gap between seeing a number and knowing what to do.
Each was developed against the same fields: user need, core interaction, benefit, risk, research basis and prototype feasibility.
Explain an unexpected bill or usage change using bill facts plus minimal household context, marking what is fact and what is still inference.
Direct fit with the attribution evidence.
False precision; hypotheses read as diagnoses.
Address information arriving too late: flag a meaningful change quietly and let the resident decide whether to look closer.
Timely support with low attention demand.
Alerts may reintroduce the monitoring and anxiety our research warned about.
Move an unresolved case: collect what is known, what was checked and what remains open, so the resident can speak clearly to a landlord, agent, strata or retailer.
Supports distributed control directly.
Incorrect responsibility guidance; premature escalation.
The strongest evidence we hold points at one thing: residents cannot explain why a bill changed. Cause Lens is also the only direction that can be prototyped honestly with transparent rules and fictional data, without pretending to hold appliance-level sensors.
Quiet Signal needs long-run usage data before “unusual” means anything, and risks the monitoring pressure we found.
Renter Evidence Pathway stays valuable for structural cases, but questionnaire demand for reporting is clearly weaker than for attribution. It becomes the follow-on route for unresolved cases.
On-demand, evidence-labelled, and designed to stop where the available data stops. It converts real or near-real-time usage information into an explanation the resident can act on.
Confirm what actually moved: billing period, days, kWh, cost, tariff, actual or estimated reading, and the closest comparable period.
A small set of possible causes, each carrying its supporting evidence and its missing evidence. No percentages without measurement.
Each cause maps to a control boundary: resident, household, property, building or billing system. No automatic assignment of responsibility.
One low-risk, reversible check, with a statement of what result would support or weaken the hypothesis.
Directly supplied by the bill or confirmed by the resident.
e.g. the meter reading is marked actual, not estimated.
Consistent with some available evidence, but not confirmed by measurement.
e.g. cooling use changed during the hot weeks of the period.
Cannot be resolved with current information; needs more data or another stakeholder.
e.g. there is no comparable earlier reading to compare against.
Fictional scenario. A quarterly bill of $487.60 arrives higher than expected. One housemate blames the air-conditioner, but the period is seven days longer than the last one and the cause is unresolved. The scenario exists to make the interaction testable; it is not a participant story.
Entry is a question, not a dashboard. Nothing is monitored until the resident asks.
Facts and gaps are shown together, so the resident sees what the bill cannot tell them.
Four short questions. Every one is skippable and explains why it is being asked.
Each card carries its evidence state, its supporting and missing evidence, and who can check it.
The case ends in a state, not a verdict: resolved, revised, or carried forward as a private summary the resident chooses whether to share.
Boundaries are described as capability to check, never as fault.
One check at a time, chosen for low effort and no comfort cost.
A negative result is information. Recording it is what makes the next guess better.
Export exists for unresolved cases only, and is never the default path.
The interaction hypothesis to falsify: that a resident can understand an explanation carrying evidence and uncertainty, and make a better-founded decision about what to do next.
A low- to mid-fidelity Cause Lens flow built on a fictional but plausible high-bill scenario: bill facts, a few contextual questions, cause cards, control boundaries and one reversible next action.
No participant walkthrough, expert walkthrough, tutor feedback or user-acceptance evidence has been collected. Human evaluation is deliberately left for the team to conduct with the required permissions.
Approved participant boundary · task script · consented anonymised observations · prototype version · a feedback-to-change record · evidence storage location.
Australian Energy Regulator. (n.d.). Embedded networks customers.
Chalal, M. L., Medjdoub, B., Bezai, N., Bull, R., & Zune, M. (2022). Visualisation in energy eco-feedback systems: A systematic review of good practice. Renewable and Sustainable Energy Reviews, 162, 112447.
Chatzigeorgiou, I. M., & Andreou, G. T. (2021). A systematic review on feedback research for residential energy behavior change through mobile and web interfaces. Renewable and Sustainable Energy Reviews, 135, 110187.
Daniel, L., Moore, T., Baker, E., Beer, A., Willand, N., Horne, R., & Hamilton, C. (2020). Warm, cool and energy-affordable housing policy solutions for low-income renters (AHURI Final Report No. 338). Australian Housing and Urban Research Institute.
Department of Climate Change, Energy, the Environment and Water. (2024). Aligning home energy ratings: Research report.
Dritsa, D., & Houben, S. (2024). How design researchers make sense of data visualizations in data-driven design: An uncertainty-aware sensemaking model. ACM Transactions on Computer-Human Interaction, 31(6), Article 72.
Easthope, H., Palmer, J., Sharam, A., Nethercote, M., Pignatta, G., & Crommelin, L. (2023). Delivering sustainable apartment housing: New build and retrofit (AHURI Final Report No. 400). Australian Housing and Urban Research Institute.
Energy Consumers Australia. (2025). Understanding and measuring energy hardship in Australia.
Franzke, A. S., Bechmann, A., Ess, C. M., Zimmer, M., & Association of Internet Researchers. (2020). Internet research: Ethical guidelines 3.0.
NSW Government. (n.d.). Connection and supply of electricity and gas in rental properties.
Project team. (2026). Anonymous household electricity questionnaire [Unpublished raw data]. Live data checked 28 August 2026, n = 39.
Project team. (2026). Beyond the Bill data and coding audit workbook [Unpublished internal document].
Extended verification notes, the public online observation record and the generative-AI provenance ledger appear in the appendix.
Sources rechecked 28 Aug 2026
Supporting evidence, method boundaries, the full ideation record, concept sheets, prototype specification and the generative-AI provenance ledger.
The questionnaire explored living context, bill access, attribution confidence, saving actions, perceived control, information needs, and privacy, anxiety and comfort concerns.
Explore how student renters access, read and respond to household electricity information, and what constraints shape their response.
Adults aged 18 and over. Convenience sample, strongly student-weighted. No bill upload, address or account information was required.
Self-report supports exploratory needs and concerns. It does not measure appliance loads, diagnose building faults, prove prevalence or validate a concept.
Count only a concrete action reasonably expected to reduce electricity use: reducing heating or cooling time, switching off or unplugging devices, using a fan, or setting a timer.
Do not count intention, awareness, information seeking, paying the bill, contacting a stakeholder, using an app, buying a monitor without a reduction action, “nothing”, or unclear language.
AI-assisted, rule-based second-pass audit. No discrepancy was found against the strict 16 / 39 result. This is not an independent human coding check and does not replace one, should the team choose to perform it.
Attach every cause hypothesis to a control boundary and an appropriate next check. Do not assign automatic legal responsibility, and do not guarantee a stakeholder response.
The self-investigation count is not yet confirmed in the frozen snapshot and must be filled from the source workbook before submission. All other counts are as recorded.
First actions are distributed across supplier, household, landlord or agent, strata or property, self-investigation and “do not know”. There is no single action route.
Clusters are traceable to source evidence, but they are not universal and they are not user validation.
Information arrives delayed or aggregated.
The total cannot be connected to a cause or a moment.
The resident may not control the relevant system.
The next check or contact route is unclear.
Saving can conflict with health, study and sleep.
Structural constraints can be mistaken for carelessness.
The strongest opportunity is not “make people use less”. It is helping renters turn incomplete information into an explanation and a proportionate action, while preserving comfort.
Verification date: 28 August 2026 · primary or recognised research sources only for the claims used.
Franzke et al. (2020) informs the public-observation ethics boundary recorded in Appendix F.
Threads were read only as publicly posted. No contact was made, no comments or private messages were sent, and no bill images or addresses were copied.
Public posters are not project participants. Their replies are hypotheses circulating in a forum, not verified diagnoses, and are used here only to identify plausible breakdowns, renter language and real-world action routes. Approach follows Franzke et al. (2020).
Requires real participant evidence before any claim of comprehension, usefulness or acceptance.
Generated through brainwriting against the five How Might We prompts. Cluster colour indicates the primary job each idea addresses.
Sketch slots below are reserved for the team's hand thumbnails. Idea IDs and captions are final.
Sketch slots reserved for the team's hand thumbnails.
Sketch slots reserved for the team's hand thumbnails.
User question + bill/context + cause cards + self-check + explicit uncertainty + on-demand case
Unusual change + interval data + timeline + investigate or dismiss + uncertainty + opt-in alert
Unresolved problem + bill/context evidence + control map + escalation + missing-data statement + export
“Tell me what may have changed, how strong the evidence is, and what I can check next.”
Select period → confirm known facts → answer minimal context → review evidence-labelled cause hypotheses → see the control map → try one reversible check.
Closest fit with the attribution evidence and with the stated need for source and timing information.
False precision, household sensitivity and interface complexity.
Three uncertainty states, minimal data collection, no appliance percentages without measurement, and an on-demand case rather than continuous monitoring.
QD-01–05 · OE-01 / 02 / 04 · secondary research
High, for a fictional-data low- to mid-fidelity prototype.
“Tell me when something genuinely changes, without making me monitor electricity every day.”
Opt in → set sensitivity and quiet periods → receive a low-pressure flag → see why it appeared → investigate, dismiss or mute.
Responds to the timing need with a low attention demand.
Anxiety, false alerts, a sense of surveillance, and unavailable longitudinal data.
Optional and infrequent, compared only against the household's own history, with transparent data requirements and mute controls.
QD-04 / 06 / 10 · secondary research
Medium. The interaction can be prototyped; the detection cannot yet be validated.
“Help me explain what happened, what I checked, and who may be able to respond.”
Build a timeline → check evidence → map control and responsibility uncertainty → see official contact routes → export a resident-controlled summary.
Reduces the effort of carrying an unresolved case across several actors.
Incorrect legal or technical guidance, premature escalation, and handling sensitive account data.
No legal conclusions, official links only, wording that depends on the actual arrangement, and export controlled entirely by the resident.
QD-07–09 · OE-01 / 03 / 05 · secondary research
High for the case-building flow. Stakeholder response remains external and untestable here.
Cause Lens — closest evidence fit, and honest to prototype with fictional data.
Quiet Signal needs longitudinal data. Renter Evidence Pathway has weaker demand as the primary job. Also deferred: longitudinal anomaly detection, automatic appliance percentages, default escalation.
The scenario exists to make the interaction testable. It is not a participant story.
Fictional data throughout. No diagnosis is claimed at any stage.
Frame slots reserved for the team's illustrations. Captions are final.
In any stop state the case does not proceed to a conclusion. It states what is unknown and what would be needed to continue.
Internal evidence-led critique only. No human feedback has been collected.
The team remains responsible for accuracy, course compliance, final concept decisions and submission.
Generative AI assisted ideation, evidence organisation, draft interaction content, concept comparison and visual production of this document.
None. All figures in this document are drawn from questionnaire data or authored as diagrams and wireframes. The ideation sheets (I1–I3) and the storyboard (M2) carry reserved slots labelled for the team's own sketches.
No participant feedback, expert walkthrough, tutor feedback, workshop photograph, team consensus, user acceptance, measured energy saving, legal advice, or proof that Cause Lens works.
Questionnaire exported and re-audited 28 August 2026. Official and academic sources rechecked 28 August 2026. Public online observation recorded 21 August 2026.
Four prompts (P01 sensemaking ideation sheet, P02 agency and handoff ideation sheet, P03 timing and wellbeing ideation sheet, P04 eight-frame storyboard) are held verbatim in AI_Prompt_Ledger.txt. If the team inserts generated images into I1–I3 or M2, paste the exact prompt text and the file's SHA-256 hash onto a page per asset here before submission.