# Household connectivity experience and switching

A fictional service provider wants to decide whether reliability, support and value should receive priority before its next renewal campaign. This study separates experienced problems, support contact, perceived service quality and future switching intent. An intention to switch is not a prediction of actual churn.

Interview adults who share responsibility for choosing or paying for household fixed broadband. Recruit an online opt-in sample of 500 eligible households, one respondent per household. The host may set marginal targets for age band and broad settlement type; keep provider-user bases visible in reporting and avoid implying that a convenience sample represents all households. Mobile-only households are outside this particular fixed-broadband universe and should be reported as screening dispositions. A separate mobile-only instrument would require a revised sampling plan.

The intended median interview length is 10–12 minutes; a timed pilot must verify it. Use a four-week reliability period and a three-month support period. Ask support evaluations only of respondents who contacted support. Present Don't know separately from low satisfaction and do not impute an answer. A separate availability question permits unknown, inseparable bundled costs and refusal; ask a whole-pound broadband-only amount only when the respondent can estimate it. Zero means free. Exclude TV/mobile/device add-ons and retain unavailable amounts as absent.

Report reliability incidents, service ratings, support ratings and switching intent as separate measures. Do not label the questionnaire a validated satisfaction index or NPS instrument. Report hypothetical switching triggers among all eligible respondents and, separately, among respondents with stated switching intention, showing each denominator. External longitudinal work is needed to relate intention to actual provider changes.

The fictional market uses pounds (£). All providers, places and fixture respondents are invented. No account number, telephone number, address, IP address or contact information is requested. The hosting organization supplies its real privacy notice, retention and withdrawal procedure before collection.

Files: `SPEC.md`, independent `study.qube`/`study.odin`, and synthetic `fixtures.json`. Run `node tools/test_fictional_consumer_studies.mjs` from the repository root. Local checks do not qualify real invitations, quotas, hosting, browser/mobile layout, accessibility or timed length. Host-controlled quotas, permissions and services are not implemented by these standalone sources. Legacy DLL oracle approval and human acceptance remain separate.