Fictional study: business software adoption and continuation
This is a newly written example for script testing, author training and a potential demonstration. It has no commissioning client, fieldwork, real responses or research findings. Its wording is not copied from a client questionnaire. Publishing or claiming a qualified hosted study requires a separate review.
The imagined sponsor wants to understand how New Zealand organisations evaluate business software, which difficulties discourage adoption, and what could affect continuation of an existing tool. The study asks about work tools, not personal subscriptions. It includes adults who decide, approve or formally recommend software; people with no purchasing influence screen out. Organisations using none of the listed categories remain eligible for the adoption questions.
Recruit one knowledgeable person per organisation from a fictional opted-in business panel. Target approximately 300 completed interviews, with recruitment spread across organisation-size bands and broad sectors. This is a planning assumption, not an achieved sample or power calculation. Neither the questionnaire nor this example enforces panel eligibility, deduplication, contact permissions or cross-case quota authority. Those belong to the host and fieldwork plan. No payment or incentive workflow is included.
The unit of analysis is the decision-maker’s account of one organisation. Current usage means tools used at the time of interview; continuation and adoption refer to the next twelve months. For detailed experience, the respondent chooses the tool whose selection or renewal they most influence and gives it a short non-confidential label. A portfolio module collects category satisfaction, the availability and size of a tool-count estimate, coordination between multiple tools, category purchasing role, and conditional replacement/consolidation reasons. It retains selected mention order and distinct category records. The design does not assume that the person answering directly uses every tool; Cannot judge is a substantive missing category, not the scale midpoint.
The target interview duration is 20–25 minutes for a current user. It has not been timed. There are 53 numbered item definitions; the portfolio module repeats for selected categories, so the number of screens varies. People using no listed software have a shorter path. A small mobile and desktop cognitive pilot should test the category distinctions, the chosen-tool instruction, renewal timing and perceived burden before fieldwork. Budget approvers and sole owners may interpret access counts differently; interviewers or cognitive probes should check this rather than silently equating seats with active users. Q35 asks whether an access-count estimate is available before Q12. Don’t know and Prefer not to answer leave the count absent, preserve the distinct gate response, and allow the remaining tool questions to continue. Report access counts only among respondents supplying an estimate; unavailable counts must not become zero.
The purchasing module separates the most recent personally influenced decision from future plans. Recent decisions collect suppliers evaluated, pilot use and approval gates. Future projects collect up to three category identities, business-case maturity, overseas-hosting approval and mandatory requirements. Budget availability gates NZD annual spend bands; unavailable estimates remain absent. The 10% price vignette applies only to a known paid licence and is an intention measure, not a pricing model.
Report separate denominators: all eligible completes for adoption criteria and barriers; current users for tool experience and recommendation; renewable paid-tool users for contract timing; AI-permitted/trial respondents for AI use cases. Each Q09 category rating uses respondents selecting that category, rather than all current users. Multiselect percentages use respondents as the denominator and may sum above 100%. Rating codes 1–5 are ordered categories, with code 6 excluded from substantive summaries and reported separately. Recommendation 0–10 is a descriptive intention indicator; do not present it as a validated revenue forecast or a comparable supplier league table. Different respondents select different tools, so unweighted overall scores are not vendor estimates. Unanswered, skipped and Cannot judge must remain distinct.
Q24 randomizes the seven substantive purchase criteria while keeping Other last. Each engine uses its own deterministic seeded ordering; identical ODIN/QSS permutations are not claimed. Topic order and rating direction are fixed to reduce accidental context changes. The repeated-category contract specifies selected mention order, preserving the chosen sequence.
Participation is voluntary and a respondent may leave. Free text requests generic labels and excludes client names and confidential figures, but this does not guarantee anonymous responses. Storage, access, retention and deletion policies must be supplied by the actual host before any real use. Fictional test fixtures use invented labels and statements only.
The paired scripts and independent scenarios qualify only the executed paths, validations and answer records documented by the focused harness. They are not vendor-DLL equivalence, ODIN production approval, a survey-method validation, browser accessibility proof, hosted fieldwork rehearsal or a general NIPO capability claim.